diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,74454 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 12404,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.680965147453083e-06,
+      "loss": 4.5056,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.361930294906166e-06,
+      "loss": 4.7007,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.04289544235925e-06,
+      "loss": 4.6743,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0723860589812333e-05,
+      "loss": 4.3821,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3404825737265417e-05,
+      "loss": 4.4353,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.60857908847185e-05,
+      "loss": 4.7395,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.876675603217158e-05,
+      "loss": 4.5067,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1447721179624665e-05,
+      "loss": 4.4043,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4128686327077747e-05,
+      "loss": 4.4801,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6809651474530833e-05,
+      "loss": 4.6597,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9490616621983912e-05,
+      "loss": 4.4885,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2171581769437e-05,
+      "loss": 4.411,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.485254691689008e-05,
+      "loss": 4.4398,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.753351206434316e-05,
+      "loss": 4.4488,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.021447721179625e-05,
+      "loss": 4.4335,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.289544235924933e-05,
+      "loss": 4.2612,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.557640750670242e-05,
+      "loss": 4.4847,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8257372654155495e-05,
+      "loss": 4.304,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.093833780160858e-05,
+      "loss": 4.4179,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.3619302949061666e-05,
+      "loss": 4.2927,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.630026809651475e-05,
+      "loss": 4.2074,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.8981233243967824e-05,
+      "loss": 4.3569,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.16621983914209e-05,
+      "loss": 4.1638,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.4343163538874e-05,
+      "loss": 4.3648,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.702412868632708e-05,
+      "loss": 4.4101,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.970509383378017e-05,
+      "loss": 4.0898,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.238605898123325e-05,
+      "loss": 4.2903,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.506702412868632e-05,
+      "loss": 4.3643,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.774798927613941e-05,
+      "loss": 3.9349,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.04289544235925e-05,
+      "loss": 4.1064,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.310991957104558e-05,
+      "loss": 4.2157,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.579088471849866e-05,
+      "loss": 4.2978,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.847184986595174e-05,
+      "loss": 4.2138,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.115281501340484e-05,
+      "loss": 4.2219,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.383378016085791e-05,
+      "loss": 3.9725,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.651474530831099e-05,
+      "loss": 4.0764,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.919571045576407e-05,
+      "loss": 4.2098,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010187667560321715,
+      "loss": 4.2297,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010455764075067025,
+      "loss": 4.1287,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010723860589812333,
+      "loss": 4.0883,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010991957104557641,
+      "loss": 3.9625,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001126005361930295,
+      "loss": 4.1021,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011528150134048257,
+      "loss": 3.9127,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011796246648793565,
+      "loss": 3.9697,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012064343163538874,
+      "loss": 4.0852,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001233243967828418,
+      "loss": 3.9649,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012600536193029492,
+      "loss": 3.8829,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000128686327077748,
+      "loss": 3.9912,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001313672922252011,
+      "loss": 3.8896,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013404825737265417,
+      "loss": 3.9951,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013672922252010725,
+      "loss": 3.9065,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013941018766756033,
+      "loss": 3.8521,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014209115281501342,
+      "loss": 4.0513,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001447721179624665,
+      "loss": 3.9905,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014745308310991955,
+      "loss": 3.8583,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015013404825737264,
+      "loss": 3.9013,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015281501340482572,
+      "loss": 3.9285,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015549597855227883,
+      "loss": 3.8923,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001581769436997319,
+      "loss": 3.8573,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000160857908847185,
+      "loss": 3.7849,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016353887399463807,
+      "loss": 3.9592,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016621983914209116,
+      "loss": 4.0169,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00016890080428954424,
+      "loss": 3.8744,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017158176943699732,
+      "loss": 3.5715,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001742627345844504,
+      "loss": 3.7266,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017694369973190349,
+      "loss": 3.768,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00017962466487935657,
+      "loss": 3.6529,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018230563002680968,
+      "loss": 3.9392,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018498659517426276,
+      "loss": 3.7641,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00018766756032171581,
+      "loss": 3.9693,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0001903485254691689,
+      "loss": 3.7977,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019302949061662198,
+      "loss": 3.7701,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019571045576407506,
+      "loss": 3.6338,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00019839142091152814,
+      "loss": 3.8021,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00020107238605898123,
+      "loss": 3.9095,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002037533512064343,
+      "loss": 3.6367,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002064343163538874,
+      "loss": 3.7558,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002091152815013405,
+      "loss": 3.6858,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021179624664879358,
+      "loss": 3.8657,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021447721179624666,
+      "loss": 3.7741,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021715817694369975,
+      "loss": 3.5222,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00021983914209115283,
+      "loss": 3.8026,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002225201072386059,
+      "loss": 3.6976,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000225201072386059,
+      "loss": 3.6276,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00022788203753351208,
+      "loss": 3.7541,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023056300268096513,
+      "loss": 3.9139,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00023324396782841821,
+      "loss": 4.082,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002359249329758713,
+      "loss": 3.7524,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002386058981233244,
+      "loss": 3.7427,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002412868632707775,
+      "loss": 3.7775,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00024396782841823057,
+      "loss": 3.8743,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002466487935656836,
+      "loss": 3.7983,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002493297587131367,
+      "loss": 3.6228,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00025201072386058984,
+      "loss": 3.7209,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002546916890080429,
+      "loss": 3.7958,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000257372654155496,
+      "loss": 3.6996,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002600536193029491,
+      "loss": 3.7013,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002627345844504022,
+      "loss": 3.4004,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00026541554959785526,
+      "loss": 3.5795,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00026809651474530834,
+      "loss": 3.7328,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002707774798927614,
+      "loss": 3.8102,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002734584450402145,
+      "loss": 3.7942,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002761394101876676,
+      "loss": 3.7152,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00027882037533512067,
+      "loss": 3.634,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028150134048257375,
+      "loss": 3.622,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00028418230563002683,
+      "loss": 3.7229,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002868632707774799,
+      "loss": 3.6406,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000289544235924933,
+      "loss": 3.7836,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000292225201072386,
+      "loss": 3.5748,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002949061662198391,
+      "loss": 3.7754,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002975871313672922,
+      "loss": 3.6404,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030026809651474527,
+      "loss": 3.6145,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030294906166219835,
+      "loss": 3.8322,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030563002680965144,
+      "loss": 3.6069,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00030831099195710457,
+      "loss": 3.7303,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031099195710455765,
+      "loss": 3.6266,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00031367292225201074,
+      "loss": 3.6942,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003163538873994638,
+      "loss": 3.63,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003190348525469169,
+      "loss": 3.6495,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032171581769437,
+      "loss": 3.5662,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032439678284182307,
+      "loss": 3.3465,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032707774798927615,
+      "loss": 3.6504,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00032975871313672923,
+      "loss": 3.6322,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003324396782841823,
+      "loss": 3.3552,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003351206434316354,
+      "loss": 3.6682,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003378016085790885,
+      "loss": 3.6302,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034048257372654156,
+      "loss": 3.6298,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00034316353887399464,
+      "loss": 3.7576,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003458445040214477,
+      "loss": 3.4768,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003485254691689008,
+      "loss": 3.507,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003512064343163539,
+      "loss": 3.6654,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035388739946380697,
+      "loss": 3.3679,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035656836461126005,
+      "loss": 3.6443,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00035924932975871314,
+      "loss": 3.5562,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036193029490616627,
+      "loss": 3.4396,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036461126005361935,
+      "loss": 3.6713,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00036729222520107244,
+      "loss": 3.6486,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003699731903485255,
+      "loss": 3.5865,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003726541554959786,
+      "loss": 3.5484,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00037533512064343163,
+      "loss": 3.7849,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003780160857908847,
+      "loss": 3.6027,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003806970509383378,
+      "loss": 3.821,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003833780160857909,
+      "loss": 3.6333,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038605898123324396,
+      "loss": 3.4564,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00038873994638069704,
+      "loss": 3.6285,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003914209115281501,
+      "loss": 3.7933,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003941018766756032,
+      "loss": 3.7927,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0003967828418230563,
+      "loss": 3.6335,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00039946380697050937,
+      "loss": 3.535,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040214477211796245,
+      "loss": 3.7445,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00040482573726541553,
+      "loss": 3.4131,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004075067024128686,
+      "loss": 3.5931,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004101876675603217,
+      "loss": 3.5004,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004128686327077748,
+      "loss": 3.3845,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00041554959785522786,
+      "loss": 3.7084,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000418230563002681,
+      "loss": 3.5225,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004209115281501341,
+      "loss": 3.8217,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042359249329758717,
+      "loss": 3.5852,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042627345844504025,
+      "loss": 3.4497,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00042895442359249333,
+      "loss": 3.5197,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004316353887399464,
+      "loss": 3.7623,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004343163538873995,
+      "loss": 3.5317,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004369973190348526,
+      "loss": 3.5683,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00043967828418230566,
+      "loss": 3.3049,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00044235924932975874,
+      "loss": 3.5602,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004450402144772118,
+      "loss": 3.6711,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004477211796246649,
+      "loss": 3.3921,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000450402144772118,
+      "loss": 3.6147,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045308310991957107,
+      "loss": 3.7874,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00045576407506702415,
+      "loss": 3.6169,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004584450402144772,
+      "loss": 3.5472,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046112600536193026,
+      "loss": 3.5859,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046380697050938335,
+      "loss": 3.5365,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00046648793565683643,
+      "loss": 3.5386,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004691689008042895,
+      "loss": 3.5617,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004718498659517426,
+      "loss": 3.3853,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00047453083109919573,
+      "loss": 3.4019,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004772117962466488,
+      "loss": 3.3252,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004798927613941019,
+      "loss": 3.3393,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.000482573726541555,
+      "loss": 3.6015,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00048525469168900806,
+      "loss": 3.606,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.00048793565683646114,
+      "loss": 3.4438,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004906166219839142,
+      "loss": 3.4984,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004932975871313673,
+      "loss": 3.6215,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004959785522788203,
+      "loss": 3.4542,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0004986595174262734,
+      "loss": 3.6443,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005013404825737266,
+      "loss": 3.4622,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005040214477211797,
+      "loss": 3.4298,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005067024128686328,
+      "loss": 3.483,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005093833780160859,
+      "loss": 3.4811,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005120643431635389,
+      "loss": 3.4373,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000514745308310992,
+      "loss": 3.4904,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005174262734584451,
+      "loss": 3.5644,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005201072386058982,
+      "loss": 3.4527,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005227882037533513,
+      "loss": 3.4208,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005254691689008043,
+      "loss": 3.3163,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005281501340482574,
+      "loss": 3.5268,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005308310991957105,
+      "loss": 3.316,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005335120643431636,
+      "loss": 3.4611,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005361930294906167,
+      "loss": 3.4167,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005388739946380698,
+      "loss": 3.6485,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005415549597855228,
+      "loss": 3.4581,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005442359249329759,
+      "loss": 3.6436,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000546916890080429,
+      "loss": 3.7011,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005495978552278821,
+      "loss": 3.65,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005522788203753352,
+      "loss": 3.5182,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005549597855227883,
+      "loss": 3.4927,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005576407506702413,
+      "loss": 3.34,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005603217158176944,
+      "loss": 3.5522,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005630026809651475,
+      "loss": 3.5606,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005656836461126006,
+      "loss": 3.4391,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005683646112600537,
+      "loss": 3.6014,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005710455764075067,
+      "loss": 3.6005,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005737265415549598,
+      "loss": 3.7645,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005764075067024129,
+      "loss": 3.6632,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000579088471849866,
+      "loss": 3.4965,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005817694369973191,
+      "loss": 3.5575,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000584450402144772,
+      "loss": 3.6135,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005871313672922251,
+      "loss": 3.541,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005898123324396782,
+      "loss": 3.6028,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005924932975871313,
+      "loss": 3.402,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005951742627345844,
+      "loss": 3.4476,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0005978552278820375,
+      "loss": 3.4834,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006005361930294905,
+      "loss": 3.5106,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006032171581769436,
+      "loss": 3.5723,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006058981233243967,
+      "loss": 3.547,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006085790884718498,
+      "loss": 3.5552,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006112600536193029,
+      "loss": 3.5637,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006139410187667561,
+      "loss": 3.4603,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006166219839142091,
+      "loss": 3.4384,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006193029490616622,
+      "loss": 3.4509,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006219839142091153,
+      "loss": 3.4964,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006246648793565684,
+      "loss": 3.4809,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006273458445040215,
+      "loss": 3.6007,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006300268096514746,
+      "loss": 3.5872,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006327077747989276,
+      "loss": 3.6099,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006353887399463807,
+      "loss": 3.4948,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006380697050938338,
+      "loss": 3.6397,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006407506702412869,
+      "loss": 3.4219,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00064343163538874,
+      "loss": 3.5236,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000646112600536193,
+      "loss": 3.5653,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006487935656836461,
+      "loss": 3.6204,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006514745308310992,
+      "loss": 3.5153,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006541554959785523,
+      "loss": 3.4594,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006568364611260054,
+      "loss": 3.1602,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006595174262734585,
+      "loss": 3.5707,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006621983914209115,
+      "loss": 3.5101,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006648793565683646,
+      "loss": 3.5676,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006675603217158177,
+      "loss": 3.5808,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006702412868632708,
+      "loss": 3.3707,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006729222520107239,
+      "loss": 3.4087,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000675603217158177,
+      "loss": 3.327,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00067828418230563,
+      "loss": 3.4561,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006809651474530831,
+      "loss": 3.419,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006836461126005362,
+      "loss": 3.3771,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006863270777479893,
+      "loss": 3.3801,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006890080428954424,
+      "loss": 3.3452,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006916890080428954,
+      "loss": 3.337,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006943699731903485,
+      "loss": 3.5394,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006970509383378016,
+      "loss": 3.4903,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0006997319034852547,
+      "loss": 3.2122,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007024128686327078,
+      "loss": 3.4186,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007050938337801609,
+      "loss": 3.3163,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007077747989276139,
+      "loss": 3.4557,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000710455764075067,
+      "loss": 3.3332,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007131367292225201,
+      "loss": 3.4628,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007158176943699732,
+      "loss": 3.4516,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007184986595174263,
+      "loss": 3.5307,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007211796246648794,
+      "loss": 3.1228,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007238605898123325,
+      "loss": 3.2845,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007265415549597856,
+      "loss": 3.3893,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007292225201072387,
+      "loss": 3.4385,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007319034852546918,
+      "loss": 3.4979,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007345844504021449,
+      "loss": 3.5945,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000737265415549598,
+      "loss": 3.3905,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000739946380697051,
+      "loss": 3.5767,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007426273458445041,
+      "loss": 3.5386,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007453083109919572,
+      "loss": 3.5799,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007479892761394103,
+      "loss": 3.3117,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007506702412868633,
+      "loss": 3.6162,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007533512064343163,
+      "loss": 3.4789,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007560321715817694,
+      "loss": 3.5055,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007587131367292225,
+      "loss": 3.4608,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007613941018766756,
+      "loss": 3.3967,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007640750670241287,
+      "loss": 3.5569,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007667560321715818,
+      "loss": 3.3741,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007694369973190348,
+      "loss": 3.5724,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007721179624664879,
+      "loss": 3.6305,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000774798927613941,
+      "loss": 3.5216,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007774798927613941,
+      "loss": 3.3702,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007801608579088472,
+      "loss": 3.3584,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007828418230563002,
+      "loss": 3.3977,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007855227882037533,
+      "loss": 3.5203,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007882037533512064,
+      "loss": 3.3817,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007908847184986595,
+      "loss": 3.2694,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007935656836461126,
+      "loss": 3.5191,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007962466487935657,
+      "loss": 3.3964,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0007989276139410187,
+      "loss": 3.4319,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008016085790884718,
+      "loss": 3.1677,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008042895442359249,
+      "loss": 3.6935,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.000806970509383378,
+      "loss": 3.4752,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008096514745308311,
+      "loss": 3.279,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008123324396782842,
+      "loss": 3.2315,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008150134048257372,
+      "loss": 3.6522,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008176943699731903,
+      "loss": 3.371,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008203753351206434,
+      "loss": 3.5735,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008230563002680965,
+      "loss": 3.6451,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008257372654155496,
+      "loss": 3.3793,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008284182305630026,
+      "loss": 3.3573,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0008310991957104557,
+      "loss": 3.3081,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008337801608579089,
+      "loss": 3.5121,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000836461126005362,
+      "loss": 3.3999,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008391420911528151,
+      "loss": 3.7084,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008418230563002682,
+      "loss": 3.1661,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008445040214477212,
+      "loss": 3.3997,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008471849865951743,
+      "loss": 3.5647,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008498659517426274,
+      "loss": 3.7293,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008525469168900805,
+      "loss": 3.5796,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008552278820375336,
+      "loss": 3.4183,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008579088471849867,
+      "loss": 3.4069,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008605898123324397,
+      "loss": 3.42,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008632707774798928,
+      "loss": 3.4943,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008659517426273459,
+      "loss": 3.4958,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000868632707774799,
+      "loss": 3.5379,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008713136729222521,
+      "loss": 3.4513,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008739946380697052,
+      "loss": 3.7136,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008766756032171582,
+      "loss": 3.3402,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008793565683646113,
+      "loss": 3.4035,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008820375335120644,
+      "loss": 3.263,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008847184986595175,
+      "loss": 3.3754,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008873994638069706,
+      "loss": 3.5344,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008900804289544236,
+      "loss": 3.2584,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008927613941018767,
+      "loss": 3.5329,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008954423592493298,
+      "loss": 3.4527,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0008981233243967829,
+      "loss": 3.6162,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000900804289544236,
+      "loss": 3.4061,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009034852546916891,
+      "loss": 3.4858,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009061662198391421,
+      "loss": 3.2216,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009088471849865952,
+      "loss": 3.4746,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009115281501340483,
+      "loss": 3.5282,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009142091152815014,
+      "loss": 3.47,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009168900804289544,
+      "loss": 3.4327,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009195710455764074,
+      "loss": 3.5238,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009222520107238605,
+      "loss": 3.5225,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009249329758713136,
+      "loss": 3.4252,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009276139410187667,
+      "loss": 3.5481,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009302949061662198,
+      "loss": 3.4861,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009329758713136729,
+      "loss": 3.3141,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009356568364611259,
+      "loss": 3.4315,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000938337801608579,
+      "loss": 3.4716,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009410187667560321,
+      "loss": 3.2669,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009436997319034852,
+      "loss": 3.3611,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009463806970509384,
+      "loss": 3.562,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009490616621983915,
+      "loss": 3.6294,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009517426273458445,
+      "loss": 3.3456,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009544235924932976,
+      "loss": 3.5747,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009571045576407507,
+      "loss": 3.3573,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009597855227882038,
+      "loss": 3.3589,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009624664879356569,
+      "loss": 3.4414,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00096514745308311,
+      "loss": 3.5348,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000967828418230563,
+      "loss": 3.4129,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009705093833780161,
+      "loss": 3.4775,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009731903485254692,
+      "loss": 3.3529,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009758713136729223,
+      "loss": 3.4056,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009785522788203754,
+      "loss": 3.3092,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009812332439678283,
+      "loss": 3.3507,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009839142091152815,
+      "loss": 3.3695,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009865951742627345,
+      "loss": 3.6229,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009892761394101877,
+      "loss": 3.4639,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009919571045576407,
+      "loss": 3.6431,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009946380697050939,
+      "loss": 3.4117,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009973190348525468,
+      "loss": 3.2746,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.001,
+      "loss": 3.6911,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999999829534579,
+      "loss": 3.5913,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999999318138324,
+      "loss": 3.3611,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999998465811271,
+      "loss": 3.42,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999999727255348,
+      "loss": 3.4089,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999995738365032,
+      "loss": 3.4272,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999993863246028,
+      "loss": 3.1569,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00099999916471966,
+      "loss": 3.4177,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999989090216897,
+      "loss": 3.3649,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999986192307092,
+      "loss": 3.4385,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999982953467386,
+      "loss": 3.551,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999979373698001,
+      "loss": 3.4344,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999975452999175,
+      "loss": 3.4804,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999997119137118,
+      "loss": 3.2594,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999966588814305,
+      "loss": 3.5102,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999961645328864,
+      "loss": 3.4857,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999956360915194,
+      "loss": 3.252,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999950735573655,
+      "loss": 3.381,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999944769304631,
+      "loss": 3.5984,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999938462108531,
+      "loss": 3.4279,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999931813985783,
+      "loss": 3.4316,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999924824936838,
+      "loss": 3.3633,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999917494962177,
+      "loss": 3.7086,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999909824062296,
+      "loss": 3.3578,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999990181223772,
+      "loss": 3.3453,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999893459488995,
+      "loss": 3.6693,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999884765816691,
+      "loss": 3.5419,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999875731221402,
+      "loss": 3.3315,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999986635570374,
+      "loss": 3.4508,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999856639264346,
+      "loss": 3.5241,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999846581903885,
+      "loss": 3.5136,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999983618362304,
+      "loss": 3.3944,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999982544442252,
+      "loss": 3.4823,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999814364303062,
+      "loss": 3.3143,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999802943265416,
+      "loss": 3.406,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999979118131036,
+      "loss": 3.3259,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00099997790784387,
+      "loss": 3.3173,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999976663465126,
+      "loss": 3.5675,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999975384994889,
+      "loss": 3.4501,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999740724332459,
+      "loss": 3.3541,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999972725780286,
+      "loss": 3.5157,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999713450361016,
+      "loss": 3.5022,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999699302007867,
+      "loss": 3.3483,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999684812744378,
+      "loss": 3.2538,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999669982571537,
+      "loss": 3.4579,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999654811490353,
+      "loss": 3.2424,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999639299501863,
+      "loss": 3.4564,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999623446607122,
+      "loss": 3.3272,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999607252807214,
+      "loss": 3.3997,
+      "step": 421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999959071810324,
+      "loss": 3.5204,
+      "step": 422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999573842496333,
+      "loss": 3.4343,
+      "step": 423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999955662598764,
+      "loss": 3.4295,
+      "step": 424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999539068578332,
+      "loss": 3.4521,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999952117026961,
+      "loss": 3.6927,
+      "step": 426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999502931062694,
+      "loss": 3.5073,
+      "step": 427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999484350958826,
+      "loss": 3.53,
+      "step": 428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999465429959276,
+      "loss": 3.4015,
+      "step": 429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999944616806533,
+      "loss": 3.4883,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999426565278306,
+      "loss": 3.3811,
+      "step": 431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999406621599535,
+      "loss": 3.4137,
+      "step": 432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0009999386337030383,
+      "loss": 3.3525,
+      "step": 433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.000999936571157223,
+      "loss": 3.5153,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999344745226482,
+      "loss": 3.4236,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999932343799457,
+      "loss": 3.3508,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999301789877946,
+      "loss": 3.4205,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999279800878085,
+      "loss": 3.359,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999257470996489,
+      "loss": 3.2616,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999234800234678,
+      "loss": 3.6935,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00099992117885942,
+      "loss": 3.3756,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999188436076623,
+      "loss": 3.5184,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999164742683538,
+      "loss": 3.5026,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999140708416563,
+      "loss": 3.4803,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999116333277336,
+      "loss": 3.4668,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999091617267518,
+      "loss": 3.286,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999066560388797,
+      "loss": 3.448,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999041162642877,
+      "loss": 3.4755,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009999015424031494,
+      "loss": 3.1095,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998989344556402,
+      "loss": 3.3439,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998962924219377,
+      "loss": 3.3642,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998936163022223,
+      "loss": 3.4001,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998909060966762,
+      "loss": 3.6748,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998881618054848,
+      "loss": 3.5855,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998853834288345,
+      "loss": 3.3505,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998825709669151,
+      "loss": 3.4729,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998797244199185,
+      "loss": 3.3987,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998768437880384,
+      "loss": 3.4487,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998739290714715,
+      "loss": 3.3256,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998709802704166,
+      "loss": 3.3681,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998679973850747,
+      "loss": 3.376,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998649804156489,
+      "loss": 3.4768,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998619293623454,
+      "loss": 3.3241,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998588442253719,
+      "loss": 3.4088,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999855725004939,
+      "loss": 3.3225,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999852571701259,
+      "loss": 3.4442,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998493843145473,
+      "loss": 3.3072,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999846162845021,
+      "loss": 3.5965,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998429072929,
+      "loss": 3.4591,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998396176584062,
+      "loss": 3.4283,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998362939417638,
+      "loss": 3.4306,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998329361431994,
+      "loss": 3.508,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999829544262942,
+      "loss": 3.3934,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999826118301223,
+      "loss": 3.4651,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999822658258276,
+      "loss": 3.4041,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998191641343369,
+      "loss": 3.3188,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998156359296436,
+      "loss": 3.2855,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998120736444372,
+      "loss": 3.3568,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998084772789603,
+      "loss": 3.3127,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998048468334581,
+      "loss": 3.4545,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009998011823081786,
+      "loss": 3.4263,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999797483703371,
+      "loss": 3.4163,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999793751019288,
+      "loss": 3.3252,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997899842561838,
+      "loss": 3.4273,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999786183414315,
+      "loss": 3.4478,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997823484939417,
+      "loss": 3.3217,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997784794953247,
+      "loss": 3.2223,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997745764187278,
+      "loss": 3.1939,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997706392644173,
+      "loss": 3.3764,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997666680326615,
+      "loss": 3.4984,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997626627237315,
+      "loss": 3.5173,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997586233379003,
+      "loss": 3.1547,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999754549875443,
+      "loss": 3.294,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997504423366378,
+      "loss": 3.247,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997463007217645,
+      "loss": 3.444,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997421250311056,
+      "loss": 3.3308,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997379152649458,
+      "loss": 3.2643,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999733671423572,
+      "loss": 3.3512,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997293935072738,
+      "loss": 3.1478,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997250815163429,
+      "loss": 3.4098,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999720735451073,
+      "loss": 3.2401,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999716355311761,
+      "loss": 3.1969,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999711941098705,
+      "loss": 3.652,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009997074928122065,
+      "loss": 3.3869,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999703010452568,
+      "loss": 3.4123,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996984940200964,
+      "loss": 3.3021,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996939435150984,
+      "loss": 3.3385,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999689358937885,
+      "loss": 3.2853,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996847402887687,
+      "loss": 3.4674,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996800875680643,
+      "loss": 3.6196,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999675400776089,
+      "loss": 3.4151,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996706799131625,
+      "loss": 3.3922,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996659249796066,
+      "loss": 3.4546,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996611359757456,
+      "loss": 3.1734,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996563129019062,
+      "loss": 3.4481,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996514557584172,
+      "loss": 3.3703,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996465645456093,
+      "loss": 3.3851,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996416392638167,
+      "loss": 3.5865,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996366799133749,
+      "loss": 3.2499,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999631686494622,
+      "loss": 3.4819,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996266590078986,
+      "loss": 3.5128,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996215974535476,
+      "loss": 3.2911,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999616501831914,
+      "loss": 3.3759,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996113721433453,
+      "loss": 3.4357,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996062083881911,
+      "loss": 3.4758,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009996010105668036,
+      "loss": 3.4036,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995957786795373,
+      "loss": 3.4995,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995905127267491,
+      "loss": 3.3549,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995852127087977,
+      "loss": 3.339,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995798786260445,
+      "loss": 3.2404,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995745104788537,
+      "loss": 3.4025,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995691082675907,
+      "loss": 3.2095,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995636719926243,
+      "loss": 3.338,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999558201654325,
+      "loss": 3.2491,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995526972530657,
+      "loss": 3.4088,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995471587892222,
+      "loss": 3.3172,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995415862631715,
+      "loss": 3.3998,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999535979675294,
+      "loss": 3.3732,
+      "step": 538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995303390259718,
+      "loss": 3.2961,
+      "step": 539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995246643155896,
+      "loss": 3.336,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995189555445344,
+      "loss": 3.1351,
+      "step": 541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995132127131952,
+      "loss": 3.3059,
+      "step": 542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009995074358219638,
+      "loss": 3.7163,
+      "step": 543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999501624871234,
+      "loss": 3.2936,
+      "step": 544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994957798614024,
+      "loss": 3.1359,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999489900792867,
+      "loss": 3.2418,
+      "step": 546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994839876660288,
+      "loss": 3.1253,
+      "step": 547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994780404812913,
+      "loss": 3.4503,
+      "step": 548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994720592390598,
+      "loss": 3.3443,
+      "step": 549
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999466043939742,
+      "loss": 3.4015,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994599945837483,
+      "loss": 3.4427,
+      "step": 551
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994539111714913,
+      "loss": 3.4033,
+      "step": 552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994477937033854,
+      "loss": 3.4402,
+      "step": 553
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999441642179848,
+      "loss": 3.4508,
+      "step": 554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994354566012983,
+      "loss": 3.3105,
+      "step": 555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994292369681584,
+      "loss": 3.3445,
+      "step": 556
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0009994229832808522,
+      "loss": 3.2719,
+      "step": 557
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.000999416695539806,
+      "loss": 3.4676,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999410373745449,
+      "loss": 3.5065,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009994040178982117,
+      "loss": 3.3213,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999397627998528,
+      "loss": 3.2626,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993912040468331,
+      "loss": 3.4737,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993847460435653,
+      "loss": 3.4689,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999378253989165,
+      "loss": 3.2702,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993717278840746,
+      "loss": 3.2668,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993651677287394,
+      "loss": 3.3033,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993585735236065,
+      "loss": 3.4698,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993519452691257,
+      "loss": 3.1562,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993452829657487,
+      "loss": 3.2824,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993385866139302,
+      "loss": 3.544,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993318562141265,
+      "loss": 3.2954,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993250917667963,
+      "loss": 3.4697,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009993182932724013,
+      "loss": 3.3927,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999311460731405,
+      "loss": 3.415,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999304594144273,
+      "loss": 3.337,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992976935114734,
+      "loss": 3.5119,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992907588334773,
+      "loss": 3.5456,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999283790110757,
+      "loss": 3.3833,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992767873437882,
+      "loss": 3.306,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999269750533048,
+      "loss": 3.4638,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992626796790163,
+      "loss": 3.4891,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992555747821753,
+      "loss": 3.3656,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992484358430091,
+      "loss": 3.1225,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999241262862005,
+      "loss": 3.3914,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992340558396518,
+      "loss": 3.2088,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992268147764412,
+      "loss": 3.2564,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992195396728665,
+      "loss": 3.3883,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999212230529424,
+      "loss": 3.3769,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009992048873466123,
+      "loss": 3.2832,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991975101249317,
+      "loss": 3.4222,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991900988648855,
+      "loss": 3.3652,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999182653566979,
+      "loss": 3.4577,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991751742317195,
+      "loss": 3.5464,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991676608596175,
+      "loss": 3.5043,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999160113451185,
+      "loss": 3.2752,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999152532006937,
+      "loss": 3.3061,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099914491652739,
+      "loss": 3.3134,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991372670130636,
+      "loss": 3.4025,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999129583464479,
+      "loss": 3.3688,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991218658821608,
+      "loss": 3.3664,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009991141142666345,
+      "loss": 3.4155,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999106328618429,
+      "loss": 3.4838,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999098508938075,
+      "loss": 3.377,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999090655226106,
+      "loss": 3.3603,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990827674830572,
+      "loss": 3.2046,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990748457094666,
+      "loss": 3.3727,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990668899058743,
+      "loss": 3.3533,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999058900072823,
+      "loss": 3.4137,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990508762108572,
+      "loss": 3.3099,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990428183205239,
+      "loss": 3.3983,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999034726402373,
+      "loss": 3.4254,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990266004569557,
+      "loss": 3.5047,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990184404848267,
+      "loss": 3.5458,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000999010246486542,
+      "loss": 3.5098,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009990020184626604,
+      "loss": 3.3291,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989937564137427,
+      "loss": 3.3409,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989854603403528,
+      "loss": 3.3414,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998977130243056,
+      "loss": 3.4428,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989687661224203,
+      "loss": 3.2908,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998960367979016,
+      "loss": 3.3096,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998951935813416,
+      "loss": 3.2323,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989434696261949,
+      "loss": 3.5354,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989349694179302,
+      "loss": 3.4213,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989264351892015,
+      "loss": 3.3584,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009989178669405907,
+      "loss": 3.3435,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998909264672682,
+      "loss": 3.3714,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998900628386062,
+      "loss": 3.5086,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988919580813194,
+      "loss": 3.4126,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988832537590456,
+      "loss": 3.3376,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998874515419834,
+      "loss": 3.3397,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988657430642805,
+      "loss": 3.3221,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988569366929832,
+      "loss": 3.3186,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988480963065427,
+      "loss": 3.3353,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988392219055615,
+      "loss": 3.406,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988303134906451,
+      "loss": 3.2147,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988213710624009,
+      "loss": 3.5313,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998812394621438,
+      "loss": 3.3304,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009988033841683691,
+      "loss": 3.3923,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987943397038088,
+      "loss": 3.3192,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987852612283732,
+      "loss": 3.2905,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987761487426816,
+      "loss": 3.2152,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987670022473553,
+      "loss": 3.5944,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987578217430179,
+      "loss": 3.4223,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987486072302956,
+      "loss": 3.2871,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987393587098166,
+      "loss": 3.1335,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987300761822116,
+      "loss": 3.3807,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987207596481131,
+      "loss": 3.4807,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987114091081569,
+      "loss": 3.1697,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009987020245629805,
+      "loss": 3.391,
+      "step": 649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986926060132234,
+      "loss": 3.4914,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986831534595283,
+      "loss": 3.3702,
+      "step": 651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986736669025393,
+      "loss": 3.3688,
+      "step": 652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986641463429035,
+      "loss": 3.391,
+      "step": 653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986545917812701,
+      "loss": 3.7516,
+      "step": 654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986450032182907,
+      "loss": 3.5396,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986353806546188,
+      "loss": 3.2298,
+      "step": 656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986257240909107,
+      "loss": 3.291,
+      "step": 657
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986160335278247,
+      "loss": 3.3103,
+      "step": 658
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009986063089660216,
+      "loss": 3.3454,
+      "step": 659
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985965504061647,
+      "loss": 3.3392,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985867578489193,
+      "loss": 3.3068,
+      "step": 661
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985769312949528,
+      "loss": 3.2301,
+      "step": 662
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985670707449359,
+      "loss": 3.1872,
+      "step": 663
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985571761995402,
+      "loss": 3.322,
+      "step": 664
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998547247659441,
+      "loss": 3.2689,
+      "step": 665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985372851253147,
+      "loss": 3.3687,
+      "step": 666
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998527288597841,
+      "loss": 3.3079,
+      "step": 667
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985172580777018,
+      "loss": 3.3789,
+      "step": 668
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009985071935655803,
+      "loss": 3.3395,
+      "step": 669
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984970950621633,
+      "loss": 3.3044,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984869625681392,
+      "loss": 3.2048,
+      "step": 671
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984767960841987,
+      "loss": 3.3032,
+      "step": 672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984665956110355,
+      "loss": 3.1582,
+      "step": 673
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984563611493448,
+      "loss": 3.3726,
+      "step": 674
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984460926998247,
+      "loss": 3.4356,
+      "step": 675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000998435790263175,
+      "loss": 3.4163,
+      "step": 676
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984254538400981,
+      "loss": 3.1272,
+      "step": 677
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984150834312995,
+      "loss": 3.4953,
+      "step": 678
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009984046790374856,
+      "loss": 3.3458,
+      "step": 679
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009983942406593662,
+      "loss": 3.2337,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0009983837682976532,
+      "loss": 3.444,
+      "step": 681
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00099837326195306,
+      "loss": 3.4116,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983627216263039,
+      "loss": 3.3617,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983521473181027,
+      "loss": 3.4515,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998341539029178,
+      "loss": 3.4494,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983308967602533,
+      "loss": 3.1546,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983202205120535,
+      "loss": 3.259,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009983095102853075,
+      "loss": 3.2744,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982987660807447,
+      "loss": 3.2561,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982879878990983,
+      "loss": 3.4546,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982771757411032,
+      "loss": 3.3575,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998266329607496,
+      "loss": 3.3696,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982554494990172,
+      "loss": 3.3431,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998244535416408,
+      "loss": 3.1348,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998233587360413,
+      "loss": 3.4828,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982226053317782,
+      "loss": 3.3066,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982115893312528,
+      "loss": 3.2476,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009982005393595881,
+      "loss": 3.3148,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998189455417537,
+      "loss": 3.3208,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998178337505856,
+      "loss": 3.1563,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981671856253023,
+      "loss": 3.2503,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981559997766373,
+      "loss": 3.4143,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981447799606226,
+      "loss": 3.1632,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981335261780243,
+      "loss": 3.3515,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009981222384296091,
+      "loss": 3.4414,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998110916716147,
+      "loss": 3.3919,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980995610384099,
+      "loss": 3.2062,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000998088171397172,
+      "loss": 3.3334,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980767477932098,
+      "loss": 3.1464,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980652902273025,
+      "loss": 3.2536,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980537987002312,
+      "loss": 3.5775,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980422732127794,
+      "loss": 3.5207,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980307137657334,
+      "loss": 3.2446,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980191203598806,
+      "loss": 3.3348,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009980074929960125,
+      "loss": 3.4116,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979958316749212,
+      "loss": 3.1309,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997984136397402,
+      "loss": 3.523,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979724071642526,
+      "loss": 3.4203,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979606439762723,
+      "loss": 3.4133,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979488468342638,
+      "loss": 3.3029,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997937015739031,
+      "loss": 3.2216,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979251506913806,
+      "loss": 3.3944,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997913251692122,
+      "loss": 3.2372,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009979013187420664,
+      "loss": 3.3772,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978893518420274,
+      "loss": 3.3513,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997877350992821,
+      "loss": 3.4224,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978653161952656,
+      "loss": 3.6568,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978532474501816,
+      "loss": 3.3648,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978411447583919,
+      "loss": 3.2869,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978290081207222,
+      "loss": 3.3302,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978168375379994,
+      "loss": 3.2188,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009978046330110537,
+      "loss": 3.2823,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977923945407175,
+      "loss": 3.371,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997780122127825,
+      "loss": 3.3615,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977678157732128,
+      "loss": 3.227,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977554754777205,
+      "loss": 3.2793,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977431012421893,
+      "loss": 3.3436,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977306930674629,
+      "loss": 3.3431,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977182509543876,
+      "loss": 3.1953,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009977057749038115,
+      "loss": 3.3713,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976932649165852,
+      "loss": 3.2067,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976807209935623,
+      "loss": 3.4077,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976681431355977,
+      "loss": 3.1681,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976555313435489,
+      "loss": 3.3894,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976428856182762,
+      "loss": 3.5122,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976302059606416,
+      "loss": 3.4382,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976174923715098,
+      "loss": 3.3988,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009976047448517476,
+      "loss": 3.1913,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975919634022242,
+      "loss": 3.1708,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975791480238113,
+      "loss": 3.1197,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975662987173827,
+      "loss": 3.2682,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975534154838144,
+      "loss": 3.2055,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975404983239848,
+      "loss": 3.2128,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997527547238775,
+      "loss": 3.4254,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975145622290677,
+      "loss": 3.3995,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009975015432957487,
+      "loss": 3.2589,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997488490439705,
+      "loss": 3.5871,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974754036618278,
+      "loss": 3.4633,
+      "step": 758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974622829630082,
+      "loss": 3.4118,
+      "step": 759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974491283441416,
+      "loss": 3.3674,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974359398061247,
+      "loss": 3.2783,
+      "step": 761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974227173498568,
+      "loss": 3.1736,
+      "step": 762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009974094609762397,
+      "loss": 3.3233,
+      "step": 763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997396170686177,
+      "loss": 3.361,
+      "step": 764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997382846480575,
+      "loss": 3.2328,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973694883603423,
+      "loss": 3.3107,
+      "step": 766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973560963263898,
+      "loss": 3.2461,
+      "step": 767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973426703796305,
+      "loss": 3.2147,
+      "step": 768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099732921052098,
+      "loss": 3.4843,
+      "step": 769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973157167513559,
+      "loss": 3.3314,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009973021890716785,
+      "loss": 3.1888,
+      "step": 771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00099728862748287,
+      "loss": 3.1761,
+      "step": 772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997275031985855,
+      "loss": 3.4401,
+      "step": 773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997261402581561,
+      "loss": 3.3106,
+      "step": 774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997247739270917,
+      "loss": 3.3128,
+      "step": 775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972340420548547,
+      "loss": 3.2983,
+      "step": 776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997220310934308,
+      "loss": 3.2617,
+      "step": 777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009972065459102133,
+      "loss": 3.1561,
+      "step": 778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971927469835092,
+      "loss": 3.3255,
+      "step": 779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971789141551364,
+      "loss": 3.2847,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971650474260383,
+      "loss": 3.5162,
+      "step": 781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971511467971603,
+      "loss": 3.3326,
+      "step": 782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971372122694504,
+      "loss": 3.5011,
+      "step": 783
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971232438438584,
+      "loss": 3.0777,
+      "step": 784
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009971092415213371,
+      "loss": 3.3389,
+      "step": 785
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970952053028413,
+      "loss": 3.3596,
+      "step": 786
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970811351893278,
+      "loss": 3.5063,
+      "step": 787
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997067031181756,
+      "loss": 3.2103,
+      "step": 788
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970528932810878,
+      "loss": 3.3622,
+      "step": 789
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997038721488287,
+      "loss": 3.3303,
+      "step": 790
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009970245158043203,
+      "loss": 3.2297,
+      "step": 791
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000997010276230156,
+      "loss": 3.2554,
+      "step": 792
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.000996996002766765,
+      "loss": 3.2221,
+      "step": 793
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969816954151205,
+      "loss": 3.4725,
+      "step": 794
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969673541761984,
+      "loss": 3.2278,
+      "step": 795
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969529790509763,
+      "loss": 3.1754,
+      "step": 796
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969385700404345,
+      "loss": 3.3298,
+      "step": 797
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969241271455555,
+      "loss": 3.2653,
+      "step": 798
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009969096503673243,
+      "loss": 3.4171,
+      "step": 799
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968951397067273,
+      "loss": 3.2347,
+      "step": 800
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968805951647548,
+      "loss": 3.361,
+      "step": 801
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968660167423982,
+      "loss": 3.3027,
+      "step": 802
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968514044406513,
+      "loss": 3.3653,
+      "step": 803
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968367582605109,
+      "loss": 3.3501,
+      "step": 804
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968220782029752,
+      "loss": 3.1632,
+      "step": 805
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0009968073642690454,
+      "loss": 3.4604,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967926164597249,
+      "loss": 3.194,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967778347760193,
+      "loss": 3.4233,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967630192189362,
+      "loss": 3.343,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967481697894863,
+      "loss": 3.41,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967332864886816,
+      "loss": 3.4878,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967183693175372,
+      "loss": 3.3135,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009967034182770701,
+      "loss": 3.1233,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966884333683,
+      "loss": 3.3501,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966734145922486,
+      "loss": 3.25,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00099665836194994,
+      "loss": 3.4078,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966432754424002,
+      "loss": 3.4668,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009966281550706582,
+      "loss": 3.5786,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996613000835745,
+      "loss": 3.3501,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965978127386942,
+      "loss": 3.3603,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965825907805408,
+      "loss": 3.5399,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965673349623231,
+      "loss": 3.3968,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965520452850814,
+      "loss": 3.4268,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996536721749858,
+      "loss": 3.2737,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965213643576979,
+      "loss": 3.1914,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009965059731096482,
+      "loss": 3.3965,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964905480067585,
+      "loss": 3.234,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964750890500804,
+      "loss": 3.1734,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964595962406683,
+      "loss": 3.3603,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996444069579578,
+      "loss": 3.3834,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964285090678687,
+      "loss": 3.4272,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009964129147066012,
+      "loss": 3.2351,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996397286496839,
+      "loss": 3.4244,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963816244396477,
+      "loss": 3.3898,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963659285360952,
+      "loss": 3.29,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963501987872514,
+      "loss": 3.3456,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963344351941895,
+      "loss": 3.3746,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963186377579837,
+      "loss": 3.4154,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009963028064797117,
+      "loss": 3.3538,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962869413604528,
+      "loss": 3.0583,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962710424012886,
+      "loss": 3.2169,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962551096033034,
+      "loss": 3.4192,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962391429675834,
+      "loss": 3.2176,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962231424952175,
+      "loss": 3.3232,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009962071081872968,
+      "loss": 3.3875,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961910400449142,
+      "loss": 3.4432,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961749380691658,
+      "loss": 3.0354,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961588022611492,
+      "loss": 3.3419,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961426326219648,
+      "loss": 3.3342,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996126429152715,
+      "loss": 3.2251,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009961101918545049,
+      "loss": 3.3113,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960939207284414,
+      "loss": 3.487,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000996077615775634,
+      "loss": 3.3382,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960612769971946,
+      "loss": 3.54,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960449043942374,
+      "loss": 3.2412,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960284979678785,
+      "loss": 3.0542,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009960120577192366,
+      "loss": 3.1917,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995995583649433,
+      "loss": 3.385,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959790757595906,
+      "loss": 3.346,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959625340508354,
+      "loss": 3.1679,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959459585242951,
+      "loss": 3.164,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959293491810998,
+      "loss": 3.2492,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009959127060223823,
+      "loss": 3.0746,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958960290492773,
+      "loss": 3.2709,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995879318262922,
+      "loss": 3.2653,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958625736644557,
+      "loss": 3.3322,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958457952550202,
+      "loss": 3.1895,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958289830357598,
+      "loss": 3.3289,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009958121370078205,
+      "loss": 3.2731,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957952571723512,
+      "loss": 3.1952,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957783435305028,
+      "loss": 3.2226,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957613960834284,
+      "loss": 3.3767,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995744414832284,
+      "loss": 3.2272,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995727399778227,
+      "loss": 3.4644,
+      "step": 874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009957103509224182,
+      "loss": 3.1841,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956932682660194,
+      "loss": 3.0566,
+      "step": 876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995676151810196,
+      "loss": 3.259,
+      "step": 877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956590015561145,
+      "loss": 3.4079,
+      "step": 878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956418175049448,
+      "loss": 3.3082,
+      "step": 879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956245996578586,
+      "loss": 3.3026,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009956073480160298,
+      "loss": 3.427,
+      "step": 881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955900625806344,
+      "loss": 3.1807,
+      "step": 882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955727433528516,
+      "loss": 3.3116,
+      "step": 883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995555390333862,
+      "loss": 3.2328,
+      "step": 884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955380035248488,
+      "loss": 3.2252,
+      "step": 885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955205829269979,
+      "loss": 3.3491,
+      "step": 886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009955031285414967,
+      "loss": 3.2847,
+      "step": 887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954856403695353,
+      "loss": 3.4084,
+      "step": 888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954681184123069,
+      "loss": 3.4461,
+      "step": 889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954505626710052,
+      "loss": 3.3449,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954329731468282,
+      "loss": 3.2009,
+      "step": 891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009954153498409747,
+      "loss": 3.2065,
+      "step": 892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953976927546467,
+      "loss": 3.5274,
+      "step": 893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953800018890479,
+      "loss": 3.2793,
+      "step": 894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953622772453845,
+      "loss": 3.453,
+      "step": 895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953445188248652,
+      "loss": 3.522,
+      "step": 896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953267266287012,
+      "loss": 3.2826,
+      "step": 897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009953089006581053,
+      "loss": 3.4001,
+      "step": 898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995291040914293,
+      "loss": 3.353,
+      "step": 899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952731473984822,
+      "loss": 3.4168,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995255220111893,
+      "loss": 3.442,
+      "step": 901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952372590557477,
+      "loss": 3.2462,
+      "step": 902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952192642312713,
+      "loss": 3.4453,
+      "step": 903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009952012356396903,
+      "loss": 3.5079,
+      "step": 904
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951831732822342,
+      "loss": 3.1491,
+      "step": 905
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951650771601347,
+      "loss": 3.3999,
+      "step": 906
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951469472746258,
+      "loss": 3.1524,
+      "step": 907
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951287836269436,
+      "loss": 3.3426,
+      "step": 908
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009951105862183265,
+      "loss": 3.3888,
+      "step": 909
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950923550500154,
+      "loss": 3.228,
+      "step": 910
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950740901232531,
+      "loss": 3.436,
+      "step": 911
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950557914392858,
+      "loss": 3.2105,
+      "step": 912
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950374589993605,
+      "loss": 3.3398,
+      "step": 913
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009950190928047274,
+      "loss": 3.4984,
+      "step": 914
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000995000692856639,
+      "loss": 3.2069,
+      "step": 915
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949822591563497,
+      "loss": 3.1534,
+      "step": 916
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949637917051166,
+      "loss": 3.3625,
+      "step": 917
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949452905041987,
+      "loss": 3.417,
+      "step": 918
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949267555548578,
+      "loss": 3.3448,
+      "step": 919
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009949081868583576,
+      "loss": 3.298,
+      "step": 920
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948895844159643,
+      "loss": 3.2519,
+      "step": 921
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994870948228946,
+      "loss": 3.2324,
+      "step": 922
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994852278298574,
+      "loss": 3.399,
+      "step": 923
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009948335746261209,
+      "loss": 3.2256,
+      "step": 924
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994814837212862,
+      "loss": 3.3446,
+      "step": 925
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947960660600752,
+      "loss": 3.2941,
+      "step": 926
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947772611690402,
+      "loss": 3.3827,
+      "step": 927
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947584225410396,
+      "loss": 3.3274,
+      "step": 928
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0009947395501773576,
+      "loss": 3.1669,
+      "step": 929
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.000994720644079281,
+      "loss": 3.3624,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994701704248099,
+      "loss": 3.344,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009946827306851031,
+      "loss": 3.2472,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994663723391587,
+      "loss": 3.2132,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994644682368847,
+      "loss": 3.1746,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994625607618181,
+      "loss": 3.1889,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099460649914089,
+      "loss": 3.3148,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945873569382766,
+      "loss": 3.3248,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945681810116461,
+      "loss": 3.0177,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945489713623062,
+      "loss": 3.3974,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945297279915666,
+      "loss": 3.2613,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009945104509007396,
+      "loss": 3.3894,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944911400911396,
+      "loss": 3.0474,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944717955640832,
+      "loss": 3.1822,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944524173208893,
+      "loss": 3.4363,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944330053628795,
+      "loss": 3.3585,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009944135596913772,
+      "loss": 3.365,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943940803077086,
+      "loss": 3.2219,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943745672132019,
+      "loss": 3.0479,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943550204091873,
+      "loss": 3.5194,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994335439896998,
+      "loss": 3.1262,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009943158256779689,
+      "loss": 3.2203,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942961777534374,
+      "loss": 3.2913,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942764961247432,
+      "loss": 3.3778,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942567807932285,
+      "loss": 3.0438,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942370317602377,
+      "loss": 3.329,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009942172490271168,
+      "loss": 3.3556,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941974325952153,
+      "loss": 3.1885,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941775824658845,
+      "loss": 3.1931,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941576986404774,
+      "loss": 3.3645,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099413778112035,
+      "loss": 3.1232,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009941178299068606,
+      "loss": 3.4145,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940978450013694,
+      "loss": 3.3234,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000994077826405239,
+      "loss": 3.4186,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940577741198347,
+      "loss": 3.3533,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940376881465234,
+      "loss": 3.1962,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009940175684866753,
+      "loss": 3.3004,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939974151416614,
+      "loss": 3.2611,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939772281128566,
+      "loss": 3.318,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993957007401637,
+      "loss": 3.3719,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939367530093818,
+      "loss": 3.3358,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009939164649374716,
+      "loss": 3.4179,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00099389614318729,
+      "loss": 3.2199,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938757877602225,
+      "loss": 3.4683,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993855398657657,
+      "loss": 3.537,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938349758809842,
+      "loss": 3.1637,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009938145194315965,
+      "loss": 3.471,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937940293108882,
+      "loss": 3.5021,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993773505520257,
+      "loss": 3.3944,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937529480611026,
+      "loss": 3.4085,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937323569348259,
+      "loss": 3.5192,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009937117321428314,
+      "loss": 3.2562,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936910736865255,
+      "loss": 3.1477,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936703815673167,
+      "loss": 3.2715,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936496557866159,
+      "loss": 3.3015,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936288963458362,
+      "loss": 3.202,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009936081032463934,
+      "loss": 3.2726,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935872764897052,
+      "loss": 3.2487,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935664160771916,
+      "loss": 3.214,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935455220102748,
+      "loss": 3.4499,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009935245942903801,
+      "loss": 3.4099,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993503632918934,
+      "loss": 3.3038,
+      "step": 991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934826378973657,
+      "loss": 3.2506,
+      "step": 992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993461609227107,
+      "loss": 3.3343,
+      "step": 993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934405469095918,
+      "loss": 3.357,
+      "step": 994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009934194509462562,
+      "loss": 3.1593,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933983213385384,
+      "loss": 3.1574,
+      "step": 996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933771580878797,
+      "loss": 3.2079,
+      "step": 997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933559611957226,
+      "loss": 3.2919,
+      "step": 998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933347306635128,
+      "loss": 3.4227,
+      "step": 999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009933134664926976,
+      "loss": 3.2807,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932921686847272,
+      "loss": 3.1485,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932708372410537,
+      "loss": 3.4832,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932494721631316,
+      "loss": 3.222,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932280734524176,
+      "loss": 3.3207,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009932066411103711,
+      "loss": 3.0343,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931851751384534,
+      "loss": 3.294,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931636755381279,
+      "loss": 3.397,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931421423108608,
+      "loss": 3.3944,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009931205754581203,
+      "loss": 3.1909,
+      "step": 1009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993098974981377,
+      "loss": 3.2331,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993077340882104,
+      "loss": 3.1963,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000993055673161776,
+      "loss": 3.3735,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930339718218705,
+      "loss": 3.3504,
+      "step": 1013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009930122368638676,
+      "loss": 3.3538,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992990468289249,
+      "loss": 3.3409,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929686660994992,
+      "loss": 3.395,
+      "step": 1016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929468302961046,
+      "loss": 3.1611,
+      "step": 1017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929249608805542,
+      "loss": 3.3364,
+      "step": 1018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009929030578543392,
+      "loss": 3.4771,
+      "step": 1019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928811212189532,
+      "loss": 3.4048,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992859150975892,
+      "loss": 3.3522,
+      "step": 1021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928371471266532,
+      "loss": 3.315,
+      "step": 1022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009928151096727377,
+      "loss": 3.2574,
+      "step": 1023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927930386156478,
+      "loss": 3.1686,
+      "step": 1024
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927709339568886,
+      "loss": 3.3531,
+      "step": 1025
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927487956979675,
+      "loss": 2.9856,
+      "step": 1026
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009927266238403936,
+      "loss": 3.2103,
+      "step": 1027
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992704418385679,
+      "loss": 3.1881,
+      "step": 1028
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992682179335338,
+      "loss": 3.2806,
+      "step": 1029
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926599066908866,
+      "loss": 3.3028,
+      "step": 1030
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926376004538435,
+      "loss": 3.3605,
+      "step": 1031
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009926152606257298,
+      "loss": 3.3944,
+      "step": 1032
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992592887208069,
+      "loss": 3.2655,
+      "step": 1033
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925704802023862,
+      "loss": 3.3816,
+      "step": 1034
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925480396102099,
+      "loss": 3.2354,
+      "step": 1035
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925255654330695,
+      "loss": 3.0872,
+      "step": 1036
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009925030576724977,
+      "loss": 3.1928,
+      "step": 1037
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924805163300293,
+      "loss": 3.1939,
+      "step": 1038
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924579414072015,
+      "loss": 3.2929,
+      "step": 1039
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992435332905553,
+      "loss": 3.1351,
+      "step": 1040
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009924126908266262,
+      "loss": 3.3865,
+      "step": 1041
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923900151719643,
+      "loss": 3.2781,
+      "step": 1042
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923673059431136,
+      "loss": 3.4139,
+      "step": 1043
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923445631416228,
+      "loss": 3.2967,
+      "step": 1044
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009923217867690427,
+      "loss": 3.1606,
+      "step": 1045
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922989768269257,
+      "loss": 3.2362,
+      "step": 1046
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992276133316828,
+      "loss": 3.2175,
+      "step": 1047
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922532562403063,
+      "loss": 3.2829,
+      "step": 1048
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922303455989214,
+      "loss": 3.1423,
+      "step": 1049
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009922074013942348,
+      "loss": 3.5081,
+      "step": 1050
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921844236278116,
+      "loss": 3.2433,
+      "step": 1051
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921614123012178,
+      "loss": 3.2573,
+      "step": 1052
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000992138367416023,
+      "loss": 3.2309,
+      "step": 1053
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0009921152889737985,
+      "loss": 3.2346,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920921769761175,
+      "loss": 3.2367,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920690314245567,
+      "loss": 3.2615,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920458523206936,
+      "loss": 3.1614,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009920226396661089,
+      "loss": 3.4263,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919993934623856,
+      "loss": 3.1291,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919761137111086,
+      "loss": 3.3762,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919528004138651,
+      "loss": 3.0191,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919294535722452,
+      "loss": 3.4227,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009919060731878402,
+      "loss": 3.2347,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918826592622448,
+      "loss": 3.2732,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918592117970555,
+      "loss": 3.2173,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009918357307938709,
+      "loss": 3.1767,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991812216254292,
+      "loss": 3.1759,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917886681799222,
+      "loss": 3.0424,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917650865723675,
+      "loss": 3.2954,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917414714332355,
+      "loss": 3.2203,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009917178227641364,
+      "loss": 3.296,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991694140566683,
+      "loss": 3.5314,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00099167042484249,
+      "loss": 3.2378,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916466755931742,
+      "loss": 3.1956,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009916228928203551,
+      "loss": 3.2044,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915990765256546,
+      "loss": 3.2328,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915752267106966,
+      "loss": 3.3659,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915513433771072,
+      "loss": 3.3775,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915274265265147,
+      "loss": 3.408,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009915034761605504,
+      "loss": 3.1012,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914794922808467,
+      "loss": 3.0662,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914554748890398,
+      "loss": 3.2889,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914314239867668,
+      "loss": 3.2461,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009914073395756676,
+      "loss": 3.3221,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913832216573847,
+      "loss": 3.3805,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913590702335625,
+      "loss": 3.3112,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991334885305848,
+      "loss": 3.442,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009913106668758896,
+      "loss": 3.073,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912864149453395,
+      "loss": 3.4024,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912621295158509,
+      "loss": 3.3377,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912378105890796,
+      "loss": 3.1258,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009912134581666842,
+      "loss": 3.1141,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911890722503253,
+      "loss": 3.2873,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000991164652841665,
+      "loss": 3.2227,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911401999423688,
+      "loss": 3.3764,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009911157135541041,
+      "loss": 3.0383,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910911936785406,
+      "loss": 3.0659,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910666403173498,
+      "loss": 3.2204,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910420534722064,
+      "loss": 3.3961,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009910174331447864,
+      "loss": 3.3546,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990992779336769,
+      "loss": 3.3757,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990968092049835,
+      "loss": 3.2662,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990943371285668,
+      "loss": 3.1051,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009909186170459529,
+      "loss": 3.243,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908938293323787,
+      "loss": 3.2204,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908690081466345,
+      "loss": 3.2006,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908441534904136,
+      "loss": 3.3972,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009908192653654102,
+      "loss": 3.2418,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907943437733216,
+      "loss": 3.4031,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990769388715847,
+      "loss": 3.327,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990744400194688,
+      "loss": 3.2582,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009907193782115484,
+      "loss": 3.285,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906943227681346,
+      "loss": 3.0688,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990669233866155,
+      "loss": 3.3462,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00099064411150732,
+      "loss": 3.3593,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009906189556933428,
+      "loss": 3.2412,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905937664259387,
+      "loss": 3.179,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990568543706825,
+      "loss": 3.1134,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990543287537722,
+      "loss": 3.2856,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009905179979203516,
+      "loss": 3.3905,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990492674856438,
+      "loss": 3.0071,
+      "step": 1121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904673183477083,
+      "loss": 3.3348,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904419283958909,
+      "loss": 3.3033,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009904165050027179,
+      "loss": 3.361,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903910481699218,
+      "loss": 3.3468,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903655578992392,
+      "loss": 3.2961,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009903400341924077,
+      "loss": 3.2634,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990314477051168,
+      "loss": 3.1851,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902888864772627,
+      "loss": 3.2031,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902632624724365,
+      "loss": 3.2343,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902376050384368,
+      "loss": 3.2905,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009902119141770127,
+      "loss": 3.153,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901861898899166,
+      "loss": 3.3621,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000990160432178902,
+      "loss": 3.4625,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901346410457256,
+      "loss": 3.3635,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009901088164921458,
+      "loss": 3.2928,
+      "step": 1136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900829585199234,
+      "loss": 3.3355,
+      "step": 1137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900570671308219,
+      "loss": 3.2491,
+      "step": 1138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900311423266062,
+      "loss": 3.2276,
+      "step": 1139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009900051841090444,
+      "loss": 3.2615,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899791924799064,
+      "loss": 3.1889,
+      "step": 1141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899531674409645,
+      "loss": 3.3473,
+      "step": 1142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899271089939931,
+      "loss": 3.2735,
+      "step": 1143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009899010171407693,
+      "loss": 3.3218,
+      "step": 1144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989874891883072,
+      "loss": 3.2166,
+      "step": 1145
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898487332226826,
+      "loss": 3.2936,
+      "step": 1146
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009898225411613848,
+      "loss": 3.4014,
+      "step": 1147
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897963157009645,
+      "loss": 3.2864,
+      "step": 1148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897700568432099,
+      "loss": 3.1055,
+      "step": 1149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897437645899117,
+      "loss": 3.3988,
+      "step": 1150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009897174389428623,
+      "loss": 3.0137,
+      "step": 1151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989691079903857,
+      "loss": 3.2463,
+      "step": 1152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989664687474693,
+      "loss": 3.3567,
+      "step": 1153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009896382616571703,
+      "loss": 3.1607,
+      "step": 1154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00098961180245309,
+      "loss": 3.108,
+      "step": 1155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989585309864257,
+      "loss": 3.3998,
+      "step": 1156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895587838924772,
+      "loss": 3.1365,
+      "step": 1157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009895322245395597,
+      "loss": 3.4168,
+      "step": 1158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989505631807315,
+      "loss": 3.1955,
+      "step": 1159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989479005697557,
+      "loss": 3.2377,
+      "step": 1160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894523462121007,
+      "loss": 3.2624,
+      "step": 1161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009894256533527642,
+      "loss": 3.3662,
+      "step": 1162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893989271213674,
+      "loss": 3.3004,
+      "step": 1163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893721675197328,
+      "loss": 3.231,
+      "step": 1164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893453745496849,
+      "loss": 3.2205,
+      "step": 1165
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009893185482130506,
+      "loss": 3.2235,
+      "step": 1166
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892916885116594,
+      "loss": 3.4184,
+      "step": 1167
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892647954473424,
+      "loss": 3.1102,
+      "step": 1168
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892378690219334,
+      "loss": 3.3397,
+      "step": 1169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009892109092372687,
+      "loss": 3.1174,
+      "step": 1170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989183916095186,
+      "loss": 3.5304,
+      "step": 1171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891568895975267,
+      "loss": 3.2747,
+      "step": 1172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989129829746133,
+      "loss": 3.3855,
+      "step": 1173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009891027365428497,
+      "loss": 3.1807,
+      "step": 1174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.000989075609989525,
+      "loss": 3.2915,
+      "step": 1175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890484500880083,
+      "loss": 3.2248,
+      "step": 1176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009890212568401512,
+      "loss": 3.2089,
+      "step": 1177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0009889940302478084,
+      "loss": 3.2659,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988966770312836,
+      "loss": 3.2506,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009889394770370926,
+      "loss": 3.1416,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00098891215042244,
+      "loss": 3.364,
+      "step": 1181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888847904707405,
+      "loss": 3.5777,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888573971838603,
+      "loss": 3.3889,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888299705636673,
+      "loss": 3.2904,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009888025106120312,
+      "loss": 3.2574,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887750173308246,
+      "loss": 3.2507,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988747490721922,
+      "loss": 3.349,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009887199307872007,
+      "loss": 3.102,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886923375285396,
+      "loss": 3.3116,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886647109478204,
+      "loss": 3.038,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886370510469268,
+      "loss": 3.3859,
+      "step": 1191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009886093578277445,
+      "loss": 3.2622,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988581631292162,
+      "loss": 3.2779,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885538714420703,
+      "loss": 3.2247,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009885260782793615,
+      "loss": 3.4938,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884982518059313,
+      "loss": 3.3125,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884703920236767,
+      "loss": 3.3102,
+      "step": 1197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884424989344976,
+      "loss": 3.2409,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009884145725402957,
+      "loss": 3.1013,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883866128429753,
+      "loss": 3.1908,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988358619844443,
+      "loss": 3.2474,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883305935466072,
+      "loss": 3.2832,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009883025339513794,
+      "loss": 3.5441,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882744410606723,
+      "loss": 3.3213,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882463148764018,
+      "loss": 3.3451,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009882181554004858,
+      "loss": 3.1586,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000988189962634844,
+      "loss": 3.1862,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881617365813994,
+      "loss": 3.3512,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881334772420757,
+      "loss": 3.3662,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009881051846188007,
+      "loss": 3.2516,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880768587135031,
+      "loss": 3.4293,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880484995281145,
+      "loss": 3.286,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009880201070645684,
+      "loss": 3.2752,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987991681324801,
+      "loss": 3.2171,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879632223107502,
+      "loss": 3.2861,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987934730024357,
+      "loss": 3.2249,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009879062044675637,
+      "loss": 3.2016,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878776456423157,
+      "loss": 3.2009,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009878490535505603,
+      "loss": 3.2635,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987820428194247,
+      "loss": 3.346,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877917695753275,
+      "loss": 3.2916,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987763077695756,
+      "loss": 3.2694,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877343525574892,
+      "loss": 3.3758,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009877055941624852,
+      "loss": 3.2416,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876768025127054,
+      "loss": 3.2356,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876479776101128,
+      "loss": 3.1121,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009876191194566728,
+      "loss": 3.2014,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875902280543534,
+      "loss": 3.3066,
+      "step": 1228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875613034051242,
+      "loss": 3.2735,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875323455109577,
+      "loss": 3.3121,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009875033543738285,
+      "loss": 3.2455,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874743299957132,
+      "loss": 3.305,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987445272378591,
+      "loss": 3.3624,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009874161815244431,
+      "loss": 3.0238,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009873870574352531,
+      "loss": 3.3949,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987357900113007,
+      "loss": 3.4786,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987328709559693,
+      "loss": 3.3253,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987299485777301,
+      "loss": 3.3269,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872702287678243,
+      "loss": 3.2456,
+      "step": 1239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872409385332575,
+      "loss": 3.1481,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009872116150755977,
+      "loss": 3.2231,
+      "step": 1241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871822583968446,
+      "loss": 3.3232,
+      "step": 1242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009871528684989998,
+      "loss": 3.2057,
+      "step": 1243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987123445384067,
+      "loss": 3.2364,
+      "step": 1244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987093989054053,
+      "loss": 3.2897,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000987064499510966,
+      "loss": 3.3317,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870349767568166,
+      "loss": 3.2313,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009870054207936183,
+      "loss": 3.3018,
+      "step": 1248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869758316233862,
+      "loss": 3.1413,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869462092481375,
+      "loss": 3.2062,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009869165536698927,
+      "loss": 3.1577,
+      "step": 1251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868868648906734,
+      "loss": 3.313,
+      "step": 1252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868571429125042,
+      "loss": 3.0348,
+      "step": 1253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009868273877374116,
+      "loss": 3.3729,
+      "step": 1254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867975993674248,
+      "loss": 2.9735,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867677778045746,
+      "loss": 3.5309,
+      "step": 1256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009867379230508943,
+      "loss": 3.305,
+      "step": 1257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00098670803510842,
+      "loss": 3.3241,
+      "step": 1258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866781139791895,
+      "loss": 3.2485,
+      "step": 1259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986648159665243,
+      "loss": 3.1288,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009866181721686228,
+      "loss": 3.2308,
+      "step": 1261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009865881514913737,
+      "loss": 3.2341,
+      "step": 1262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986558097635543,
+      "loss": 3.1682,
+      "step": 1263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009865280106031795,
+      "loss": 3.275,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864978903963352,
+      "loss": 3.2617,
+      "step": 1265
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864677370170635,
+      "loss": 3.2904,
+      "step": 1266
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864375504674205,
+      "loss": 3.2683,
+      "step": 1267
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009864073307494646,
+      "loss": 3.1774,
+      "step": 1268
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863770778652564,
+      "loss": 3.2831,
+      "step": 1269
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863467918168585,
+      "loss": 3.1305,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009863164726063363,
+      "loss": 3.3395,
+      "step": 1271
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986286120235757,
+      "loss": 3.2163,
+      "step": 1272
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862557347071902,
+      "loss": 3.0939,
+      "step": 1273
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009862253160227076,
+      "loss": 3.2142,
+      "step": 1274
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861948641843837,
+      "loss": 3.2598,
+      "step": 1275
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861643791942948,
+      "loss": 3.0794,
+      "step": 1276
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861338610545193,
+      "loss": 3.2041,
+      "step": 1277
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009861033097671383,
+      "loss": 3.3327,
+      "step": 1278
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000986072725334235,
+      "loss": 3.0757,
+      "step": 1279
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860421077578947,
+      "loss": 3.2517,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009860114570402056,
+      "loss": 3.3519,
+      "step": 1281
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859807731832568,
+      "loss": 3.2188,
+      "step": 1282
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985950056189141,
+      "loss": 3.2728,
+      "step": 1283
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009859193060599528,
+      "loss": 3.176,
+      "step": 1284
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858885227977887,
+      "loss": 3.4192,
+      "step": 1285
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858577064047478,
+      "loss": 3.268,
+      "step": 1286
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009858268568829313,
+      "loss": 3.1882,
+      "step": 1287
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857959742344427,
+      "loss": 3.3156,
+      "step": 1288
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857650584613878,
+      "loss": 3.1562,
+      "step": 1289
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857341095658746,
+      "loss": 3.3029,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009857031275500134,
+      "loss": 3.4106,
+      "step": 1291
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985672112415917,
+      "loss": 3.1767,
+      "step": 1292
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856410641656997,
+      "loss": 3.183,
+      "step": 1293
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009856099828014788,
+      "loss": 3.1662,
+      "step": 1294
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855788683253736,
+      "loss": 3.3689,
+      "step": 1295
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985547720739506,
+      "loss": 3.2252,
+      "step": 1296
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009855165400459993,
+      "loss": 3.162,
+      "step": 1297
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854853262469799,
+      "loss": 3.2448,
+      "step": 1298
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.000985454079344576,
+      "loss": 3.2917,
+      "step": 1299
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009854227993409187,
+      "loss": 3.4996,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00098539148623814,
+      "loss": 3.38,
+      "step": 1301
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0009853601400383757,
+      "loss": 3.2552,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009853287607437628,
+      "loss": 3.1445,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852973483564413,
+      "loss": 3.3456,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852659028785527,
+      "loss": 3.2391,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852344243122414,
+      "loss": 3.1234,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009852029126596537,
+      "loss": 3.2245,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851713679229384,
+      "loss": 3.2572,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851397901042462,
+      "loss": 3.4106,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009851081792057303,
+      "loss": 3.3502,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850765352295465,
+      "loss": 3.2065,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000985044858177852,
+      "loss": 3.3246,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009850131480528069,
+      "loss": 3.325,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849814048565732,
+      "loss": 3.0659,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984949628591316,
+      "loss": 3.2938,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009849178192592011,
+      "loss": 3.0755,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848859768623983,
+      "loss": 3.3898,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848541014030783,
+      "loss": 3.2682,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009848221928834148,
+      "loss": 3.3562,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847902513055833,
+      "loss": 3.2516,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847582766717617,
+      "loss": 3.3785,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009847262689841308,
+      "loss": 3.2102,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846942282448724,
+      "loss": 3.4843,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846621544561717,
+      "loss": 3.1351,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009846300476202155,
+      "loss": 3.2748,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845979077391929,
+      "loss": 3.2833,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845657348152956,
+      "loss": 3.3042,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845335288507175,
+      "loss": 3.3728,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009845012898476542,
+      "loss": 3.0908,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844690178083043,
+      "loss": 3.3241,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984436712734868,
+      "loss": 3.2521,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009844043746295484,
+      "loss": 3.2668,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009843720034945504,
+      "loss": 3.1987,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984339599332081,
+      "loss": 3.1541,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00098430716214435,
+      "loss": 3.2057,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009842746919335693,
+      "loss": 3.2204,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009842421887019525,
+      "loss": 3.301,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000984209652451716,
+      "loss": 3.1149,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841770831850785,
+      "loss": 3.5398,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841444809042606,
+      "loss": 3.114,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009841118456114853,
+      "loss": 3.3167,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840791773089782,
+      "loss": 3.2167,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840464759989665,
+      "loss": 3.2147,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009840137416836801,
+      "loss": 3.3109,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983980974365351,
+      "loss": 3.1616,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839481740462135,
+      "loss": 3.145,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009839153407285041,
+      "loss": 3.337,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838824744144617,
+      "loss": 3.3787,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983849575106327,
+      "loss": 3.2105,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009838166428063435,
+      "loss": 3.424,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983783677516757,
+      "loss": 3.2087,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009837506792398146,
+      "loss": 3.2534,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983717647977767,
+      "loss": 3.3082,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836845837328662,
+      "loss": 3.1868,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836514865073666,
+      "loss": 3.321,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009836183563035252,
+      "loss": 3.3113,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835851931236008,
+      "loss": 3.2113,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835519969698548,
+      "loss": 3.1353,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009835187678445508,
+      "loss": 3.038,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834855057499544,
+      "loss": 3.1124,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009834522106883336,
+      "loss": 3.4666,
+      "step": 1361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983418882661959,
+      "loss": 3.3244,
+      "step": 1362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009833855216731028,
+      "loss": 3.3022,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009833521277240397,
+      "loss": 3.1026,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000983318700817047,
+      "loss": 3.0675,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832852409544035,
+      "loss": 3.105,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832517481383912,
+      "loss": 3.066,
+      "step": 1367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009832182223712936,
+      "loss": 3.2154,
+      "step": 1368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831846636553967,
+      "loss": 3.1445,
+      "step": 1369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009831510719929886,
+      "loss": 3.3494,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00098311744738636,
+      "loss": 3.1249,
+      "step": 1371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830837898378037,
+      "loss": 3.208,
+      "step": 1372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830500993496145,
+      "loss": 3.3531,
+      "step": 1373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009830163759240898,
+      "loss": 3.1829,
+      "step": 1374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829826195635288,
+      "loss": 3.2353,
+      "step": 1375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829488302702334,
+      "loss": 3.199,
+      "step": 1376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009829150080465077,
+      "loss": 3.2467,
+      "step": 1377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828811528946576,
+      "loss": 3.2059,
+      "step": 1378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828472648169917,
+      "loss": 3.1561,
+      "step": 1379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009828133438158206,
+      "loss": 3.2889,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827793898934574,
+      "loss": 3.3283,
+      "step": 1381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827454030522172,
+      "loss": 3.4043,
+      "step": 1382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009827113832944175,
+      "loss": 3.0856,
+      "step": 1383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982677330622378,
+      "loss": 3.3109,
+      "step": 1384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826432450384204,
+      "loss": 3.3416,
+      "step": 1385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009826091265448692,
+      "loss": 2.963,
+      "step": 1386
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825749751440505,
+      "loss": 3.4554,
+      "step": 1387
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825407908382931,
+      "loss": 3.4537,
+      "step": 1388
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009825065736299278,
+      "loss": 3.0897,
+      "step": 1389
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982472323521288,
+      "loss": 3.2195,
+      "step": 1390
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009824380405147087,
+      "loss": 3.1831,
+      "step": 1391
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982403724612528,
+      "loss": 3.5238,
+      "step": 1392
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009823693758170852,
+      "loss": 3.4282,
+      "step": 1393
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000982334994130723,
+      "loss": 3.2669,
+      "step": 1394
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009823005795557853,
+      "loss": 3.4679,
+      "step": 1395
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822661320946187,
+      "loss": 3.3631,
+      "step": 1396
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009822316517495724,
+      "loss": 3.2738,
+      "step": 1397
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009821971385229973,
+      "loss": 3.4866,
+      "step": 1398
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009821625924172467,
+      "loss": 3.516,
+      "step": 1399
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009821280134346762,
+      "loss": 3.3115,
+      "step": 1400
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820934015776436,
+      "loss": 3.2817,
+      "step": 1401
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820587568485088,
+      "loss": 3.2574,
+      "step": 1402
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009820240792496344,
+      "loss": 3.2349,
+      "step": 1403
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819893687833846,
+      "loss": 3.1778,
+      "step": 1404
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819546254521264,
+      "loss": 3.2589,
+      "step": 1405
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009819198492582288,
+      "loss": 3.1966,
+      "step": 1406
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981885040204063,
+      "loss": 3.3425,
+      "step": 1407
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009818501982920026,
+      "loss": 3.1677,
+      "step": 1408
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009818153235244232,
+      "loss": 3.1501,
+      "step": 1409
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817804159037028,
+      "loss": 3.1657,
+      "step": 1410
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009817454754322214,
+      "loss": 3.1,
+      "step": 1411
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981710502112362,
+      "loss": 3.3481,
+      "step": 1412
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816754959465087,
+      "loss": 3.3774,
+      "step": 1413
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981640456937049,
+      "loss": 3.3256,
+      "step": 1414
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009816053850863716,
+      "loss": 3.1737,
+      "step": 1415
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009815702803968682,
+      "loss": 3.2597,
+      "step": 1416
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009815351428709323,
+      "loss": 3.1632,
+      "step": 1417
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814999725109598,
+      "loss": 3.4122,
+      "step": 1418
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814647693193491,
+      "loss": 3.3762,
+      "step": 1419
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009814295332985002,
+      "loss": 3.3409,
+      "step": 1420
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.000981394264450816,
+      "loss": 3.3307,
+      "step": 1421
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009813589627787013,
+      "loss": 3.4053,
+      "step": 1422
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009813236282845627,
+      "loss": 3.4189,
+      "step": 1423
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009812882609708104,
+      "loss": 3.2643,
+      "step": 1424
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009812528608398553,
+      "loss": 3.2276,
+      "step": 1425
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0009812174278941114,
+      "loss": 3.203,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811819621359946,
+      "loss": 3.0774,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811464635679234,
+      "loss": 3.3406,
+      "step": 1428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009811109321923182,
+      "loss": 3.2609,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810753680116018,
+      "loss": 3.3466,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810397710281991,
+      "loss": 3.1864,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009810041412445375,
+      "loss": 3.5672,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980968478663046,
+      "loss": 3.2425,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980932783286157,
+      "loss": 3.3624,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980897055116304,
+      "loss": 3.1804,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808612941559231,
+      "loss": 3.4064,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009808255004074527,
+      "loss": 3.2519,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807896738733338,
+      "loss": 3.1634,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807538145560088,
+      "loss": 3.1385,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009807179224579233,
+      "loss": 3.174,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806819975815243,
+      "loss": 3.1082,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806460399292615,
+      "loss": 3.1726,
+      "step": 1442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009806100495035866,
+      "loss": 3.35,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805740263069537,
+      "loss": 3.2989,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805379703418191,
+      "loss": 2.8506,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009805018816106414,
+      "loss": 3.2357,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009804657601158815,
+      "loss": 3.2159,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000980429605860002,
+      "loss": 3.0104,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803934188454683,
+      "loss": 3.4125,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803571990747478,
+      "loss": 3.1938,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009803209465503102,
+      "loss": 3.1115,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802846612746274,
+      "loss": 3.0662,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802483432501737,
+      "loss": 3.3014,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009802119924794254,
+      "loss": 3.4082,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009801756089648612,
+      "loss": 3.07,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009801391927089617,
+      "loss": 3.2422,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009801027437142103,
+      "loss": 3.222,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800662619830923,
+      "loss": 3.3565,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009800297475180948,
+      "loss": 3.2076,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979993200321708,
+      "loss": 3.3811,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009799566203964238,
+      "loss": 3.1571,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009799200077447367,
+      "loss": 3.3046,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798833623691427,
+      "loss": 2.9702,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009798466842721408,
+      "loss": 3.0666,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979809973456232,
+      "loss": 3.109,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979773229923919,
+      "loss": 3.2313,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979736453677708,
+      "loss": 3.297,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979699644720106,
+      "loss": 3.1246,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979662803053623,
+      "loss": 3.2373,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009796259286807712,
+      "loss": 3.2482,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009795890216040648,
+      "loss": 3.3224,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009795520818260204,
+      "loss": 3.169,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979515109349157,
+      "loss": 3.0385,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979478104175995,
+      "loss": 3.0906,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794410663090585,
+      "loss": 3.2518,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009794039957508722,
+      "loss": 3.1085,
+      "step": 1476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793668925039644,
+      "loss": 3.1419,
+      "step": 1477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009793297565708645,
+      "loss": 3.1506,
+      "step": 1478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979292587954105,
+      "loss": 3.2288,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009792553866562201,
+      "loss": 3.2227,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009792181526797465,
+      "loss": 3.174,
+      "step": 1481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791808860272232,
+      "loss": 3.1462,
+      "step": 1482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791435867011911,
+      "loss": 3.0793,
+      "step": 1483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009791062547041933,
+      "loss": 3.3043,
+      "step": 1484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009790688900387758,
+      "loss": 3.3196,
+      "step": 1485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000979031492707486,
+      "loss": 3.1362,
+      "step": 1486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978994062712874,
+      "loss": 3.1657,
+      "step": 1487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789566000574918,
+      "loss": 3.2723,
+      "step": 1488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009789191047438943,
+      "loss": 3.3197,
+      "step": 1489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009788815767746378,
+      "loss": 3.2393,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978844016152281,
+      "loss": 3.2343,
+      "step": 1491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009788064228793856,
+      "loss": 3.3475,
+      "step": 1492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009787687969585145,
+      "loss": 3.2853,
+      "step": 1493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009787311383922333,
+      "loss": 3.1276,
+      "step": 1494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009786934471831099,
+      "loss": 3.0238,
+      "step": 1495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009786557233337145,
+      "loss": 3.4672,
+      "step": 1496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978617966846619,
+      "loss": 3.0538,
+      "step": 1497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978580177724398,
+      "loss": 3.1025,
+      "step": 1498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785423559696283,
+      "loss": 3.34,
+      "step": 1499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009785045015848887,
+      "loss": 3.4973,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009784666145727604,
+      "loss": 3.2582,
+      "step": 1501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009784286949358267,
+      "loss": 3.2497,
+      "step": 1502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009783907426766732,
+      "loss": 3.4648,
+      "step": 1503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009783527577978878,
+      "loss": 3.3389,
+      "step": 1504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009783147403020606,
+      "loss": 3.1634,
+      "step": 1505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782766901917836,
+      "loss": 3.0033,
+      "step": 1506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782386074696516,
+      "loss": 3.4366,
+      "step": 1507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009782004921382612,
+      "loss": 3.1573,
+      "step": 1508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009781623442002114,
+      "loss": 3.2322,
+      "step": 1509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000978124163658103,
+      "loss": 3.2143,
+      "step": 1510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009780859505145398,
+      "loss": 3.1486,
+      "step": 1511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009780477047721273,
+      "loss": 3.2709,
+      "step": 1512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009780094264334733,
+      "loss": 3.295,
+      "step": 1513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779711155011878,
+      "loss": 3.2488,
+      "step": 1514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009779327719778832,
+      "loss": 3.1384,
+      "step": 1515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009778943958661738,
+      "loss": 3.106,
+      "step": 1516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009778559871686768,
+      "loss": 3.2392,
+      "step": 1517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009778175458880105,
+      "loss": 3.2088,
+      "step": 1518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777790720267964,
+      "loss": 3.2647,
+      "step": 1519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777405655876578,
+      "loss": 3.2873,
+      "step": 1520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009777020265732204,
+      "loss": 3.2372,
+      "step": 1521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977663454986112,
+      "loss": 3.25,
+      "step": 1522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009776248508289627,
+      "loss": 3.2739,
+      "step": 1523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775862141044045,
+      "loss": 3.2754,
+      "step": 1524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775475448150722,
+      "loss": 3.1747,
+      "step": 1525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009775088429636023,
+      "loss": 3.2301,
+      "step": 1526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977470108552634,
+      "loss": 3.1109,
+      "step": 1527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009774313415848081,
+      "loss": 3.2561,
+      "step": 1528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009773925420627683,
+      "loss": 3.2892,
+      "step": 1529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00097735370998916,
+      "loss": 3.4228,
+      "step": 1530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977314845366631,
+      "loss": 3.2573,
+      "step": 1531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009772759481978313,
+      "loss": 3.2314,
+      "step": 1532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009772370184854134,
+      "loss": 3.2346,
+      "step": 1533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009771980562320315,
+      "loss": 3.2844,
+      "step": 1534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009771590614403424,
+      "loss": 3.0655,
+      "step": 1535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000977120034113005,
+      "loss": 3.2455,
+      "step": 1536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770809742526802,
+      "loss": 3.3245,
+      "step": 1537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770418818620319,
+      "loss": 3.4041,
+      "step": 1538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009770027569437252,
+      "loss": 3.2937,
+      "step": 1539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976963599500428,
+      "loss": 3.342,
+      "step": 1540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009769244095348103,
+      "loss": 3.2331,
+      "step": 1541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976885187049544,
+      "loss": 3.3409,
+      "step": 1542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009768459320473042,
+      "loss": 3.2672,
+      "step": 1543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976806644530767,
+      "loss": 3.362,
+      "step": 1544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009767673245026115,
+      "loss": 3.1519,
+      "step": 1545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009767279719655189,
+      "loss": 3.1874,
+      "step": 1546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976688586922172,
+      "loss": 3.2645,
+      "step": 1547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009766491693752569,
+      "loss": 3.2147,
+      "step": 1548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000976609719327461,
+      "loss": 3.1253,
+      "step": 1549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0009765702367814744,
+      "loss": 3.129,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009765307217399889,
+      "loss": 3.3202,
+      "step": 1551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764911742056995,
+      "loss": 3.139,
+      "step": 1552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764515941813022,
+      "loss": 3.0317,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009764119816694962,
+      "loss": 3.2281,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009763723366729824,
+      "loss": 3.1409,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000976332659194464,
+      "loss": 3.2624,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762929492366464,
+      "loss": 3.1379,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762532068022376,
+      "loss": 3.343,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009762134318939471,
+      "loss": 3.4197,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009761736245144871,
+      "loss": 2.9407,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009761337846665721,
+      "loss": 3.1097,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760939123529183,
+      "loss": 3.2143,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760540075762447,
+      "loss": 3.1495,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009760140703392723,
+      "loss": 3.2204,
+      "step": 1564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975974100644724,
+      "loss": 3.2472,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009759340984953253,
+      "loss": 3.168,
+      "step": 1566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758940638938038,
+      "loss": 3.2716,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758539968428895,
+      "loss": 3.3427,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009758138973453141,
+      "loss": 3.2848,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757737654038121,
+      "loss": 3.2219,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009757336010211197,
+      "loss": 3.4269,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009756934041999758,
+      "loss": 3.3517,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975653174943121,
+      "loss": 3.171,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009756129132532986,
+      "loss": 3.342,
+      "step": 1574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009755726191332537,
+      "loss": 3.0788,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975532292585734,
+      "loss": 3.1267,
+      "step": 1576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754919336134891,
+      "loss": 3.1106,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975451542219271,
+      "loss": 3.0629,
+      "step": 1578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009754111184058339,
+      "loss": 3.1413,
+      "step": 1579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753706621759337,
+      "loss": 3.2087,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009753301735323294,
+      "loss": 3.0786,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009752896524777818,
+      "loss": 3.3562,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009752490990150535,
+      "loss": 3.0987,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009752085131469101,
+      "loss": 3.3743,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009751678948761188,
+      "loss": 3.3576,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000975127244205449,
+      "loss": 3.0414,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750865611376729,
+      "loss": 3.4141,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750458456755644,
+      "loss": 3.1771,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009750050978218996,
+      "loss": 3.3936,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749643175794569,
+      "loss": 3.4375,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009749235049510172,
+      "loss": 3.1577,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009748826599393632,
+      "loss": 3.2411,
+      "step": 1592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097484178254728,
+      "loss": 3.219,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009748008727775548,
+      "loss": 3.2459,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747599306329771,
+      "loss": 3.0856,
+      "step": 1595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009747189561163387,
+      "loss": 3.1684,
+      "step": 1596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746779492304334,
+      "loss": 3.305,
+      "step": 1597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009746369099780574,
+      "loss": 3.3902,
+      "step": 1598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745958383620088,
+      "loss": 3.1331,
+      "step": 1599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745547343850883,
+      "loss": 3.2869,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009745135980500987,
+      "loss": 3.1267,
+      "step": 1601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744724293598445,
+      "loss": 3.2014,
+      "step": 1602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009744312283171335,
+      "loss": 3.2912,
+      "step": 1603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009743899949247744,
+      "loss": 3.3687,
+      "step": 1604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009743487291855792,
+      "loss": 3.3269,
+      "step": 1605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009743074311023614,
+      "loss": 3.2742,
+      "step": 1606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000974266100677937,
+      "loss": 3.5447,
+      "step": 1607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009742247379151242,
+      "loss": 3.2738,
+      "step": 1608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741833428167435,
+      "loss": 3.2622,
+      "step": 1609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741419153856172,
+      "loss": 3.3385,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009741004556245703,
+      "loss": 3.3622,
+      "step": 1611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740589635364297,
+      "loss": 3.0512,
+      "step": 1612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009740174391240245,
+      "loss": 3.1723,
+      "step": 1613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009739758823901862,
+      "loss": 3.3169,
+      "step": 1614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009739342933377484,
+      "loss": 3.062,
+      "step": 1615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738926719695469,
+      "loss": 3.0764,
+      "step": 1616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738510182884196,
+      "loss": 3.2159,
+      "step": 1617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009738093322972069,
+      "loss": 3.2532,
+      "step": 1618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737676139987509,
+      "loss": 3.3178,
+      "step": 1619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009737258633958965,
+      "loss": 3.1407,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009736840804914904,
+      "loss": 3.2772,
+      "step": 1621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009736422652883816,
+      "loss": 3.2902,
+      "step": 1622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009736004177894215,
+      "loss": 3.372,
+      "step": 1623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009735585379974632,
+      "loss": 3.2039,
+      "step": 1624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009735166259153625,
+      "loss": 3.1913,
+      "step": 1625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734746815459772,
+      "loss": 3.5439,
+      "step": 1626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009734327048921674,
+      "loss": 3.3026,
+      "step": 1627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009733906959567953,
+      "loss": 3.2096,
+      "step": 1628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009733486547427252,
+      "loss": 3.3537,
+      "step": 1629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009733065812528239,
+      "loss": 3.2737,
+      "step": 1630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00097326447548996,
+      "loss": 3.2711,
+      "step": 1631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009732223374570049,
+      "loss": 3.0832,
+      "step": 1632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731801671568315,
+      "loss": 3.25,
+      "step": 1633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009731379645923153,
+      "loss": 3.2938,
+      "step": 1634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000973095729766334,
+      "loss": 3.285,
+      "step": 1635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730534626817674,
+      "loss": 3.1736,
+      "step": 1636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009730111633414976,
+      "loss": 3.0948,
+      "step": 1637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009729688317484087,
+      "loss": 3.0757,
+      "step": 1638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009729264679053873,
+      "loss": 3.2471,
+      "step": 1639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009728840718153218,
+      "loss": 3.2111,
+      "step": 1640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009728416434811033,
+      "loss": 3.4106,
+      "step": 1641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727991829056247,
+      "loss": 3.1463,
+      "step": 1642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972756690091781,
+      "loss": 3.3388,
+      "step": 1643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009727141650424698,
+      "loss": 3.1478,
+      "step": 1644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009726716077605909,
+      "loss": 3.0819,
+      "step": 1645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972629018249046,
+      "loss": 3.3983,
+      "step": 1646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725863965107388,
+      "loss": 3.1832,
+      "step": 1647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009725437425485761,
+      "loss": 3.1948,
+      "step": 1648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972501056365466,
+      "loss": 3.3484,
+      "step": 1649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724583379643189,
+      "loss": 3.1165,
+      "step": 1650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009724155873480479,
+      "loss": 3.1872,
+      "step": 1651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009723728045195679,
+      "loss": 3.2294,
+      "step": 1652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009723299894817962,
+      "loss": 3.099,
+      "step": 1653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722871422376521,
+      "loss": 3.1557,
+      "step": 1654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722442627900571,
+      "loss": 3.2598,
+      "step": 1655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009722013511419351,
+      "loss": 3.1848,
+      "step": 1656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000972158407296212,
+      "loss": 3.2691,
+      "step": 1657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009721154312558162,
+      "loss": 3.1182,
+      "step": 1658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720724230236778,
+      "loss": 3.0946,
+      "step": 1659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009720293826027295,
+      "loss": 3.1069,
+      "step": 1660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000971986309995906,
+      "loss": 3.0771,
+      "step": 1661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009719432052061444,
+      "loss": 3.0731,
+      "step": 1662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009719000682363836,
+      "loss": 3.1346,
+      "step": 1663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718568990895651,
+      "loss": 3.3081,
+      "step": 1664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009718136977686325,
+      "loss": 3.0804,
+      "step": 1665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717704642765313,
+      "loss": 3.2342,
+      "step": 1666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009717271986162098,
+      "loss": 3.0488,
+      "step": 1667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716839007906178,
+      "loss": 3.2148,
+      "step": 1668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009716405708027076,
+      "loss": 3.3712,
+      "step": 1669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009715972086554341,
+      "loss": 3.3345,
+      "step": 1670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009715538143517535,
+      "loss": 3.0079,
+      "step": 1671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.000971510387894625,
+      "loss": 3.0143,
+      "step": 1672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009714669292870096,
+      "loss": 3.0275,
+      "step": 1673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0009714234385318706,
+      "loss": 3.4131,
+      "step": 1674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009713799156321734,
+      "loss": 3.4685,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009713363605908855,
+      "loss": 3.1936,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009712927734109772,
+      "loss": 3.0534,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009712491540954202,
+      "loss": 3.221,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009712055026471889,
+      "loss": 3.2435,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009711618190692596,
+      "loss": 3.316,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000971118103364611,
+      "loss": 3.1069,
+      "step": 1681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710743555362238,
+      "loss": 3.3348,
+      "step": 1682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009710305755870811,
+      "loss": 3.2909,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009709867635201682,
+      "loss": 3.3615,
+      "step": 1684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009709429193384722,
+      "loss": 3.2666,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009708990430449828,
+      "loss": 3.4084,
+      "step": 1686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009708551346426918,
+      "loss": 3.2238,
+      "step": 1687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009708111941345931,
+      "loss": 3.0447,
+      "step": 1688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707672215236828,
+      "loss": 3.2318,
+      "step": 1689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009707232168129593,
+      "loss": 3.1314,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706791800054231,
+      "loss": 3.2421,
+      "step": 1691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009706351111040768,
+      "loss": 3.1464,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009705910101119254,
+      "loss": 3.2862,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000970546877031976,
+      "loss": 3.232,
+      "step": 1694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009705027118672378,
+      "loss": 3.3948,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704585146207221,
+      "loss": 3.1659,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009704142852954429,
+      "loss": 3.1898,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009703700238944158,
+      "loss": 3.1326,
+      "step": 1698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009703257304206587,
+      "loss": 3.3239,
+      "step": 1699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009702814048771921,
+      "loss": 3.2105,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009702370472670384,
+      "loss": 3.3665,
+      "step": 1701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701926575932217,
+      "loss": 3.1217,
+      "step": 1702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701482358587693,
+      "loss": 3.273,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009701037820667097,
+      "loss": 3.2329,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700592962200744,
+      "loss": 3.3025,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009700147783218965,
+      "loss": 3.1827,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009699702283752117,
+      "loss": 3.2177,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009699256463830575,
+      "loss": 3.099,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969881032348474,
+      "loss": 3.2358,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009698363862745028,
+      "loss": 3.1674,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009697917081641887,
+      "loss": 3.2334,
+      "step": 1711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009697469980205777,
+      "loss": 3.326,
+      "step": 1712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009697022558467188,
+      "loss": 3.2788,
+      "step": 1713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696574816456624,
+      "loss": 3.3592,
+      "step": 1714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009696126754204618,
+      "loss": 3.3658,
+      "step": 1715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969567837174172,
+      "loss": 3.2139,
+      "step": 1716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009695229669098505,
+      "loss": 3.2615,
+      "step": 1717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009694780646305565,
+      "loss": 3.2588,
+      "step": 1718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969433130339352,
+      "loss": 3.0496,
+      "step": 1719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009693881640393009,
+      "loss": 3.2762,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009693431657334692,
+      "loss": 3.1363,
+      "step": 1721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969298135424925,
+      "loss": 3.3337,
+      "step": 1722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009692530731167391,
+      "loss": 3.2992,
+      "step": 1723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000969207978811984,
+      "loss": 3.3654,
+      "step": 1724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009691628525137344,
+      "loss": 3.34,
+      "step": 1725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009691176942250672,
+      "loss": 2.9573,
+      "step": 1726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009690725039490619,
+      "loss": 3.0562,
+      "step": 1727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009690272816887995,
+      "loss": 3.1927,
+      "step": 1728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689820274473638,
+      "loss": 3.0423,
+      "step": 1729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009689367412278404,
+      "loss": 3.2263,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009688914230333172,
+      "loss": 3.2724,
+      "step": 1731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009688460728668842,
+      "loss": 3.1751,
+      "step": 1732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009688006907316338,
+      "loss": 3.2967,
+      "step": 1733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687552766306603,
+      "loss": 3.1853,
+      "step": 1734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009687098305670605,
+      "loss": 3.2541,
+      "step": 1735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009686643525439329,
+      "loss": 3.2266,
+      "step": 1736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009686188425643788,
+      "loss": 3.1351,
+      "step": 1737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009685733006315012,
+      "loss": 3.373,
+      "step": 1738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009685277267484053,
+      "loss": 3.2234,
+      "step": 1739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684821209181987,
+      "loss": 3.3449,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009684364831439913,
+      "loss": 3.1857,
+      "step": 1741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009683908134288946,
+      "loss": 3.1965,
+      "step": 1742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000968345111776023,
+      "loss": 3.2618,
+      "step": 1743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682993781884925,
+      "loss": 3.3037,
+      "step": 1744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682536126694215,
+      "loss": 3.2471,
+      "step": 1745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009682078152219307,
+      "loss": 3.0348,
+      "step": 1746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009681619858491426,
+      "loss": 3.17,
+      "step": 1747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009681161245541824,
+      "loss": 3.3263,
+      "step": 1748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009680702313401772,
+      "loss": 3.4096,
+      "step": 1749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009680243062102561,
+      "loss": 3.3153,
+      "step": 1750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009679783491675506,
+      "loss": 3.4239,
+      "step": 1751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009679323602151944,
+      "loss": 3.316,
+      "step": 1752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009678863393563233,
+      "loss": 3.3976,
+      "step": 1753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009678402865940754,
+      "loss": 3.0855,
+      "step": 1754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009677942019315906,
+      "loss": 3.448,
+      "step": 1755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009677480853720115,
+      "loss": 3.3691,
+      "step": 1756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009677019369184825,
+      "loss": 3.2604,
+      "step": 1757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676557565741502,
+      "loss": 3.3144,
+      "step": 1758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009676095443421635,
+      "loss": 3.2297,
+      "step": 1759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009675633002256735,
+      "loss": 3.334,
+      "step": 1760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009675170242278335,
+      "loss": 3.2184,
+      "step": 1761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009674707163517986,
+      "loss": 3.2212,
+      "step": 1762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009674243766007266,
+      "loss": 3.1461,
+      "step": 1763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673780049777772,
+      "loss": 3.0746,
+      "step": 1764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009673316014861121,
+      "loss": 3.2374,
+      "step": 1765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009672851661288956,
+      "loss": 3.0728,
+      "step": 1766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967238698909294,
+      "loss": 3.1113,
+      "step": 1767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009671921998304755,
+      "loss": 3.1603,
+      "step": 1768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009671456688956108,
+      "loss": 3.1446,
+      "step": 1769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009670991061078726,
+      "loss": 3.0979,
+      "step": 1770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967052511470436,
+      "loss": 3.2346,
+      "step": 1771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000967005884986478,
+      "loss": 3.23,
+      "step": 1772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009669592266591778,
+      "loss": 3.3819,
+      "step": 1773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966912536491717,
+      "loss": 3.1168,
+      "step": 1774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009668658144872793,
+      "loss": 3.0902,
+      "step": 1775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009668190606490501,
+      "loss": 3.0716,
+      "step": 1776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009667722749802179,
+      "loss": 3.1009,
+      "step": 1777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009667254574839723,
+      "loss": 3.3615,
+      "step": 1778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009666786081635061,
+      "loss": 3.2914,
+      "step": 1779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009666317270220134,
+      "loss": 3.1991,
+      "step": 1780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665848140626911,
+      "loss": 3.2237,
+      "step": 1781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009665378692887378,
+      "loss": 3.1497,
+      "step": 1782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009664908927033547,
+      "loss": 3.0491,
+      "step": 1783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009664438843097448,
+      "loss": 3.3805,
+      "step": 1784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663968441111133,
+      "loss": 3.3104,
+      "step": 1785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966349772110668,
+      "loss": 3.1722,
+      "step": 1786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009663026683116185,
+      "loss": 3.3388,
+      "step": 1787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009662555327171763,
+      "loss": 3.4329,
+      "step": 1788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009662083653305558,
+      "loss": 3.0766,
+      "step": 1789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966161166154973,
+      "loss": 3.025,
+      "step": 1790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009661139351936464,
+      "loss": 3.1197,
+      "step": 1791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966066672449796,
+      "loss": 3.1013,
+      "step": 1792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000966019377926645,
+      "loss": 3.1297,
+      "step": 1793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000965972051627418,
+      "loss": 3.2231,
+      "step": 1794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000965924693555342,
+      "loss": 3.029,
+      "step": 1795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658773037136461,
+      "loss": 3.2317,
+      "step": 1796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009658298821055619,
+      "loss": 3.1198,
+      "step": 1797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0009657824287343227,
+      "loss": 3.1121,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000965734943603164,
+      "loss": 3.1208,
+      "step": 1799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656874267153239,
+      "loss": 3.0999,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009656398780740424,
+      "loss": 3.344,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655922976825615,
+      "loss": 2.9138,
+      "step": 1802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009655446855441256,
+      "loss": 3.2872,
+      "step": 1803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009654970416619813,
+      "loss": 3.1446,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000965449366039377,
+      "loss": 3.307,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009654016586795638,
+      "loss": 3.2745,
+      "step": 1806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009653539195857943,
+      "loss": 3.3346,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009653061487613241,
+      "loss": 3.2415,
+      "step": 1808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009652583462094102,
+      "loss": 3.0225,
+      "step": 1809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009652105119333121,
+      "loss": 3.1914,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009651626459362916,
+      "loss": 3.2135,
+      "step": 1811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009651147482216124,
+      "loss": 3.279,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650668187925405,
+      "loss": 3.1159,
+      "step": 1813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009650188576523438,
+      "loss": 3.2354,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000964970864804293,
+      "loss": 3.1105,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009649228402516601,
+      "loss": 2.9502,
+      "step": 1816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648747839977202,
+      "loss": 3.2077,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009648266960457495,
+      "loss": 3.3416,
+      "step": 1818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009647785763990274,
+      "loss": 3.2266,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009647304250608349,
+      "loss": 3.2289,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000964682242034455,
+      "loss": 3.3515,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009646340273231735,
+      "loss": 3.336,
+      "step": 1822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009645857809302776,
+      "loss": 3.335,
+      "step": 1823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009645375028590572,
+      "loss": 3.3484,
+      "step": 1824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009644891931128044,
+      "loss": 3.1023,
+      "step": 1825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000964440851694813,
+      "loss": 3.076,
+      "step": 1826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009643924786083793,
+      "loss": 3.3027,
+      "step": 1827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009643440738568015,
+      "loss": 3.0316,
+      "step": 1828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642956374433804,
+      "loss": 3.0653,
+      "step": 1829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009642471693714188,
+      "loss": 3.2493,
+      "step": 1830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009641986696442211,
+      "loss": 3.2654,
+      "step": 1831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009641501382650946,
+      "loss": 3.2281,
+      "step": 1832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009641015752373483,
+      "loss": 3.3472,
+      "step": 1833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640529805642939,
+      "loss": 3.2739,
+      "step": 1834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009640043542492445,
+      "loss": 3.4072,
+      "step": 1835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009639556962955158,
+      "loss": 3.3196,
+      "step": 1836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009639070067064257,
+      "loss": 3.3802,
+      "step": 1837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009638582854852942,
+      "loss": 3.1861,
+      "step": 1838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009638095326354433,
+      "loss": 3.115,
+      "step": 1839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637607481601973,
+      "loss": 3.4813,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009637119320628826,
+      "loss": 3.2109,
+      "step": 1841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009636630843468279,
+      "loss": 3.2164,
+      "step": 1842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000963614205015364,
+      "loss": 3.1915,
+      "step": 1843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009635652940718233,
+      "loss": 3.2636,
+      "step": 1844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009635163515195415,
+      "loss": 3.0838,
+      "step": 1845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009634673773618554,
+      "loss": 3.3699,
+      "step": 1846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009634183716021045,
+      "loss": 3.1714,
+      "step": 1847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009633693342436302,
+      "loss": 3.1979,
+      "step": 1848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009633202652897763,
+      "loss": 3.1809,
+      "step": 1849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009632711647438888,
+      "loss": 3.2741,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009632220326093152,
+      "loss": 3.3397,
+      "step": 1851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631728688894059,
+      "loss": 3.2663,
+      "step": 1852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009631236735875133,
+      "loss": 3.3959,
+      "step": 1853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009630744467069917,
+      "loss": 3.1203,
+      "step": 1854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009630251882511976,
+      "loss": 3.2999,
+      "step": 1855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009629758982234898,
+      "loss": 3.2373,
+      "step": 1856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009629265766272293,
+      "loss": 3.2129,
+      "step": 1857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009628772234657792,
+      "loss": 3.1677,
+      "step": 1858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009628278387425045,
+      "loss": 3.0586,
+      "step": 1859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009627784224607727,
+      "loss": 3.2168,
+      "step": 1860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009627289746239533,
+      "loss": 3.3444,
+      "step": 1861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009626794952354179,
+      "loss": 3.1465,
+      "step": 1862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009626299842985402,
+      "loss": 3.1386,
+      "step": 1863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009625804418166965,
+      "loss": 3.2177,
+      "step": 1864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009625308677932646,
+      "loss": 3.2318,
+      "step": 1865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009624812622316249,
+      "loss": 3.2989,
+      "step": 1866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009624316251351597,
+      "loss": 3.2604,
+      "step": 1867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009623819565072538,
+      "loss": 3.0504,
+      "step": 1868
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009623322563512937,
+      "loss": 3.0328,
+      "step": 1869
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009622825246706682,
+      "loss": 3.3407,
+      "step": 1870
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009622327614687686,
+      "loss": 2.9267,
+      "step": 1871
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621829667489878,
+      "loss": 3.2,
+      "step": 1872
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009621331405147214,
+      "loss": 3.3523,
+      "step": 1873
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009620832827693664,
+      "loss": 3.2209,
+      "step": 1874
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009620333935163228,
+      "loss": 3.3408,
+      "step": 1875
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009619834727589922,
+      "loss": 3.3463,
+      "step": 1876
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009619335205007785,
+      "loss": 3.1301,
+      "step": 1877
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009618835367450879,
+      "loss": 3.2575,
+      "step": 1878
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009618335214953284,
+      "loss": 3.2849,
+      "step": 1879
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617834747549105,
+      "loss": 3.1376,
+      "step": 1880
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009617333965272467,
+      "loss": 3.1761,
+      "step": 1881
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009616832868157514,
+      "loss": 3.3687,
+      "step": 1882
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009616331456238417,
+      "loss": 3.1466,
+      "step": 1883
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009615829729549363,
+      "loss": 3.1441,
+      "step": 1884
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009615327688124567,
+      "loss": 3.2107,
+      "step": 1885
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009614825331998255,
+      "loss": 3.1253,
+      "step": 1886
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009614322661204685,
+      "loss": 3.0148,
+      "step": 1887
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613819675778131,
+      "loss": 3.2223,
+      "step": 1888
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009613316375752888,
+      "loss": 3.1837,
+      "step": 1889
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009612812761163279,
+      "loss": 3.1709,
+      "step": 1890
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009612308832043638,
+      "loss": 3.1305,
+      "step": 1891
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611804588428329,
+      "loss": 3.3257,
+      "step": 1892
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009611300030351734,
+      "loss": 3.2764,
+      "step": 1893
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009610795157848256,
+      "loss": 3.4286,
+      "step": 1894
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009610289970952322,
+      "loss": 3.2781,
+      "step": 1895
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009609784469698377,
+      "loss": 3.317,
+      "step": 1896
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009609278654120891,
+      "loss": 3.2835,
+      "step": 1897
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009608772524254352,
+      "loss": 2.9357,
+      "step": 1898
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009608266080133272,
+      "loss": 3.0122,
+      "step": 1899
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607759321792182,
+      "loss": 2.9593,
+      "step": 1900
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009607252249265639,
+      "loss": 3.1729,
+      "step": 1901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009606744862588215,
+      "loss": 3.2213,
+      "step": 1902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009606237161794508,
+      "loss": 3.3653,
+      "step": 1903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009605729146919136,
+      "loss": 3.1,
+      "step": 1904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000960522081799674,
+      "loss": 3.2394,
+      "step": 1905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000960471217506198,
+      "loss": 3.1116,
+      "step": 1906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009604203218149537,
+      "loss": 3.3207,
+      "step": 1907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603693947294118,
+      "loss": 3.3099,
+      "step": 1908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009603184362530445,
+      "loss": 3.1208,
+      "step": 1909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602674463893267,
+      "loss": 3.2202,
+      "step": 1910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009602164251417349,
+      "loss": 3.1562,
+      "step": 1911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009601653725137485,
+      "loss": 3.0821,
+      "step": 1912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009601142885088481,
+      "loss": 3.2021,
+      "step": 1913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009600631731305172,
+      "loss": 3.0207,
+      "step": 1914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009600120263822412,
+      "loss": 3.2962,
+      "step": 1915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009599608482675075,
+      "loss": 3.2027,
+      "step": 1916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009599096387898059,
+      "loss": 3.3121,
+      "step": 1917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009598583979526278,
+      "loss": 2.9573,
+      "step": 1918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009598071257594674,
+      "loss": 3.2779,
+      "step": 1919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009597558222138208,
+      "loss": 3.2902,
+      "step": 1920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000959704487319186,
+      "loss": 3.2589,
+      "step": 1921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0009596531210790634,
+      "loss": 3.3368,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009596017234969556,
+      "loss": 3.2316,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000959550294576367,
+      "loss": 3.1213,
+      "step": 1924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009594988343208044,
+      "loss": 3.335,
+      "step": 1925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009594473427337769,
+      "loss": 3.2162,
+      "step": 1926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009593958198187952,
+      "loss": 3.2825,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009593442655793726,
+      "loss": 3.2124,
+      "step": 1928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009592926800190243,
+      "loss": 2.9978,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009592410631412679,
+      "loss": 3.0089,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591894149496228,
+      "loss": 3.1951,
+      "step": 1931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009591377354476108,
+      "loss": 3.2483,
+      "step": 1932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009590860246387556,
+      "loss": 3.3589,
+      "step": 1933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009590342825265832,
+      "loss": 3.315,
+      "step": 1934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009589825091146218,
+      "loss": 3.2904,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009589307044064015,
+      "loss": 3.2917,
+      "step": 1936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009588788684054547,
+      "loss": 3.0928,
+      "step": 1937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009588270011153159,
+      "loss": 3.1635,
+      "step": 1938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009587751025395218,
+      "loss": 3.0604,
+      "step": 1939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009587231726816111,
+      "loss": 3.1417,
+      "step": 1940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009586712115451248,
+      "loss": 3.3055,
+      "step": 1941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009586192191336056,
+      "loss": 3.035,
+      "step": 1942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009585671954505991,
+      "loss": 3.0915,
+      "step": 1943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009585151404996523,
+      "loss": 2.884,
+      "step": 1944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009584630542843149,
+      "loss": 2.9714,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009584109368081382,
+      "loss": 3.1217,
+      "step": 1946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009583587880746758,
+      "loss": 3.3323,
+      "step": 1947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009583066080874839,
+      "loss": 3.3679,
+      "step": 1948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009582543968501203,
+      "loss": 3.2392,
+      "step": 1949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009582021543661449,
+      "loss": 3.2921,
+      "step": 1950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00095814988063912,
+      "loss": 3.2467,
+      "step": 1951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009580975756726103,
+      "loss": 3.2625,
+      "step": 1952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009580452394701817,
+      "loss": 3.1936,
+      "step": 1953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000957992872035403,
+      "loss": 3.1611,
+      "step": 1954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009579404733718453,
+      "loss": 3.0088,
+      "step": 1955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000957888043483081,
+      "loss": 3.1707,
+      "step": 1956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009578355823726854,
+      "loss": 3.2633,
+      "step": 1957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009577830900442353,
+      "loss": 3.0375,
+      "step": 1958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009577305665013103,
+      "loss": 3.1774,
+      "step": 1959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576780117474915,
+      "loss": 3.2254,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009576254257863627,
+      "loss": 3.3638,
+      "step": 1961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009575728086215092,
+      "loss": 3.2252,
+      "step": 1962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009575201602565191,
+      "loss": 3.0361,
+      "step": 1963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009574674806949819,
+      "loss": 3.3668,
+      "step": 1964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00095741476994049,
+      "loss": 3.2012,
+      "step": 1965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009573620279966372,
+      "loss": 2.8356,
+      "step": 1966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009573092548670202,
+      "loss": 3.0972,
+      "step": 1967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009572564505552372,
+      "loss": 3.3081,
+      "step": 1968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009572036150648883,
+      "loss": 3.3951,
+      "step": 1969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009571507483995769,
+      "loss": 3.2999,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009570978505629072,
+      "loss": 3.2196,
+      "step": 1971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009570449215584863,
+      "loss": 3.2089,
+      "step": 1972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569919613899233,
+      "loss": 3.126,
+      "step": 1973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009569389700608294,
+      "loss": 3.0943,
+      "step": 1974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009568859475748175,
+      "loss": 3.006,
+      "step": 1975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009568328939355033,
+      "loss": 3.2551,
+      "step": 1976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567798091465043,
+      "loss": 3.2414,
+      "step": 1977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009567266932114402,
+      "loss": 3.0517,
+      "step": 1978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009566735461339327,
+      "loss": 3.4973,
+      "step": 1979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009566203679176055,
+      "loss": 3.3305,
+      "step": 1980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009565671585660849,
+      "loss": 3.473,
+      "step": 1981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000956513918082999,
+      "loss": 3.3195,
+      "step": 1982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009564606464719781,
+      "loss": 3.2133,
+      "step": 1983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009564073437366544,
+      "loss": 3.0229,
+      "step": 1984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009563540098806624,
+      "loss": 3.075,
+      "step": 1985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000956300644907639,
+      "loss": 3.0509,
+      "step": 1986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009562472488212229,
+      "loss": 3.2273,
+      "step": 1987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009561938216250547,
+      "loss": 3.3815,
+      "step": 1988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009561403633227776,
+      "loss": 3.1535,
+      "step": 1989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009560868739180369,
+      "loss": 3.3851,
+      "step": 1990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009560333534144792,
+      "loss": 3.3264,
+      "step": 1991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009559798018157547,
+      "loss": 3.1984,
+      "step": 1992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009559262191255142,
+      "loss": 3.0302,
+      "step": 1993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009558726053474116,
+      "loss": 3.3028,
+      "step": 1994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009558189604851027,
+      "loss": 3.2093,
+      "step": 1995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009557652845422451,
+      "loss": 3.0887,
+      "step": 1996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009557115775224988,
+      "loss": 3.2619,
+      "step": 1997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009556578394295262,
+      "loss": 3.2297,
+      "step": 1998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009556040702669909,
+      "loss": 3.0762,
+      "step": 1999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009555502700385598,
+      "loss": 3.0805,
+      "step": 2000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009554964387479011,
+      "loss": 3.0824,
+      "step": 2001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009554425763986851,
+      "loss": 3.0962,
+      "step": 2002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009553886829945849,
+      "loss": 3.3124,
+      "step": 2003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009553347585392751,
+      "loss": 3.2999,
+      "step": 2004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552808030364325,
+      "loss": 3.0497,
+      "step": 2005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009552268164897362,
+      "loss": 3.3404,
+      "step": 2006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009551727989028675,
+      "loss": 3.0303,
+      "step": 2007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009551187502795093,
+      "loss": 3.1523,
+      "step": 2008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009550646706233473,
+      "loss": 3.0179,
+      "step": 2009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009550105599380687,
+      "loss": 3.0468,
+      "step": 2010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009549564182273633,
+      "loss": 3.0545,
+      "step": 2011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954902245494923,
+      "loss": 3.1886,
+      "step": 2012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009548480417444411,
+      "loss": 2.9708,
+      "step": 2013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954793806979614,
+      "loss": 2.8883,
+      "step": 2014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009547395412041396,
+      "loss": 3.2403,
+      "step": 2015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954685244421718,
+      "loss": 3.1068,
+      "step": 2016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009546309166360516,
+      "loss": 3.1324,
+      "step": 2017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009545765578508448,
+      "loss": 3.3272,
+      "step": 2018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954522168069804,
+      "loss": 3.395,
+      "step": 2019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009544677472966381,
+      "loss": 3.1107,
+      "step": 2020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009544132955350576,
+      "loss": 3.3126,
+      "step": 2021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009543588127887752,
+      "loss": 3.2899,
+      "step": 2022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009543042990615064,
+      "loss": 3.2199,
+      "step": 2023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009542497543569678,
+      "loss": 2.9919,
+      "step": 2024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009541951786788788,
+      "loss": 3.1932,
+      "step": 2025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009541405720309607,
+      "loss": 3.2645,
+      "step": 2026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000954085934416937,
+      "loss": 2.9986,
+      "step": 2027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009540312658405329,
+      "loss": 3.2697,
+      "step": 2028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009539765663054764,
+      "loss": 3.1019,
+      "step": 2029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000953921835815497,
+      "loss": 3.2649,
+      "step": 2030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009538670743743267,
+      "loss": 3.0138,
+      "step": 2031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009538122819856996,
+      "loss": 3.1323,
+      "step": 2032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537574586533515,
+      "loss": 3.1339,
+      "step": 2033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009537026043810209,
+      "loss": 3.2206,
+      "step": 2034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009536477191724478,
+      "loss": 3.3352,
+      "step": 2035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009535928030313745,
+      "loss": 3.2041,
+      "step": 2036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009535378559615462,
+      "loss": 3.0971,
+      "step": 2037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009534828779667088,
+      "loss": 3.2873,
+      "step": 2038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009534278690506114,
+      "loss": 3.2124,
+      "step": 2039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009533728292170047,
+      "loss": 3.1058,
+      "step": 2040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000953317758469642,
+      "loss": 3.1153,
+      "step": 2041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009532626568122778,
+      "loss": 3.1029,
+      "step": 2042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009532075242486697,
+      "loss": 3.2036,
+      "step": 2043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.000953152360782577,
+      "loss": 3.0137,
+      "step": 2044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009530971664177607,
+      "loss": 3.0901,
+      "step": 2045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0009530419411579845,
+      "loss": 3.2085,
+      "step": 2046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009529866850070142,
+      "loss": 3.4537,
+      "step": 2047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009529313979686171,
+      "loss": 3.3373,
+      "step": 2048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009528760800465634,
+      "loss": 3.2247,
+      "step": 2049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009528207312446249,
+      "loss": 3.1256,
+      "step": 2050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009527653515665755,
+      "loss": 3.2874,
+      "step": 2051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009527099410161914,
+      "loss": 3.3368,
+      "step": 2052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009526544995972509,
+      "loss": 3.1588,
+      "step": 2053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009525990273135343,
+      "loss": 3.0218,
+      "step": 2054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009525435241688239,
+      "loss": 3.1075,
+      "step": 2055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009524879901669045,
+      "loss": 3.1411,
+      "step": 2056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009524324253115626,
+      "loss": 3.3012,
+      "step": 2057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009523768296065869,
+      "loss": 3.4531,
+      "step": 2058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009523212030557684,
+      "loss": 3.1275,
+      "step": 2059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009522655456629,
+      "loss": 3.1831,
+      "step": 2060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009522098574317766,
+      "loss": 3.2139,
+      "step": 2061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009521541383661955,
+      "loss": 3.17,
+      "step": 2062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520983884699563,
+      "loss": 3.0697,
+      "step": 2063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009520426077468598,
+      "loss": 3.2677,
+      "step": 2064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009519867962007098,
+      "loss": 3.1949,
+      "step": 2065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009519309538353119,
+      "loss": 3.2912,
+      "step": 2066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009518750806544736,
+      "loss": 3.2938,
+      "step": 2067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009518191766620048,
+      "loss": 3.091,
+      "step": 2068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009517632418617174,
+      "loss": 3.3088,
+      "step": 2069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009517072762574252,
+      "loss": 3.3948,
+      "step": 2070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009516512798529446,
+      "loss": 3.218,
+      "step": 2071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009515952526520934,
+      "loss": 3.1247,
+      "step": 2072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009515391946586922,
+      "loss": 3.073,
+      "step": 2073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009514831058765632,
+      "loss": 3.1086,
+      "step": 2074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009514269863095308,
+      "loss": 3.2234,
+      "step": 2075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009513708359614219,
+      "loss": 3.1461,
+      "step": 2076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009513146548360649,
+      "loss": 3.065,
+      "step": 2077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009512584429372908,
+      "loss": 3.2251,
+      "step": 2078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009512022002689322,
+      "loss": 3.1872,
+      "step": 2079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009511459268348242,
+      "loss": 3.2846,
+      "step": 2080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009510896226388038,
+      "loss": 3.2267,
+      "step": 2081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009510332876847104,
+      "loss": 3.124,
+      "step": 2082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009509769219763851,
+      "loss": 3.0586,
+      "step": 2083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009509205255176712,
+      "loss": 3.0469,
+      "step": 2084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508640983124141,
+      "loss": 3.082,
+      "step": 2085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009508076403644617,
+      "loss": 3.2737,
+      "step": 2086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009507511516776634,
+      "loss": 3.1799,
+      "step": 2087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009506946322558709,
+      "loss": 3.2557,
+      "step": 2088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009506380821029381,
+      "loss": 3.2613,
+      "step": 2089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000950581501222721,
+      "loss": 3.3239,
+      "step": 2090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009505248896190776,
+      "loss": 3.2843,
+      "step": 2091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009504682472958681,
+      "loss": 2.906,
+      "step": 2092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009504115742569546,
+      "loss": 3.0781,
+      "step": 2093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009503548705062016,
+      "loss": 3.3469,
+      "step": 2094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009502981360474751,
+      "loss": 3.1326,
+      "step": 2095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009502413708846441,
+      "loss": 3.131,
+      "step": 2096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009501845750215787,
+      "loss": 3.107,
+      "step": 2097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009501277484621521,
+      "loss": 3.1059,
+      "step": 2098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000950070891210239,
+      "loss": 2.9858,
+      "step": 2099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009500140032697158,
+      "loss": 3.2474,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000949957084644462,
+      "loss": 3.2467,
+      "step": 2101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009499001353383584,
+      "loss": 3.1489,
+      "step": 2102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009498431553552882,
+      "loss": 3.188,
+      "step": 2103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009497861446991367,
+      "loss": 3.2187,
+      "step": 2104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009497291033737914,
+      "loss": 3.2026,
+      "step": 2105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009496720313831412,
+      "loss": 3.1515,
+      "step": 2106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009496149287310783,
+      "loss": 3.1563,
+      "step": 2107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009495577954214958,
+      "loss": 3.2564,
+      "step": 2108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009495006314582895,
+      "loss": 3.2817,
+      "step": 2109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009494434368453573,
+      "loss": 3.1272,
+      "step": 2110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009493862115865991,
+      "loss": 3.1531,
+      "step": 2111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009493289556859168,
+      "loss": 3.2988,
+      "step": 2112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009492716691472145,
+      "loss": 3.3516,
+      "step": 2113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009492143519743983,
+      "loss": 3.3178,
+      "step": 2114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009491570041713764,
+      "loss": 3.1628,
+      "step": 2115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009490996257420593,
+      "loss": 3.1987,
+      "step": 2116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009490422166903593,
+      "loss": 3.2913,
+      "step": 2117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000948984777020191,
+      "loss": 3.3297,
+      "step": 2118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009489273067354707,
+      "loss": 3.1199,
+      "step": 2119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009488698058401173,
+      "loss": 3.4189,
+      "step": 2120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009488122743380516,
+      "loss": 3.3099,
+      "step": 2121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009487547122331965,
+      "loss": 3.1392,
+      "step": 2122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009486971195294767,
+      "loss": 3.18,
+      "step": 2123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009486394962308194,
+      "loss": 3.0934,
+      "step": 2124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009485818423411536,
+      "loss": 3.2184,
+      "step": 2125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009485241578644107,
+      "loss": 3.4465,
+      "step": 2126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009484664428045238,
+      "loss": 3.3743,
+      "step": 2127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009484086971654283,
+      "loss": 3.1199,
+      "step": 2128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009483509209510617,
+      "loss": 3.1814,
+      "step": 2129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009482931141653636,
+      "loss": 3.2858,
+      "step": 2130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009482352768122755,
+      "loss": 3.1837,
+      "step": 2131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009481774088957411,
+      "loss": 3.1273,
+      "step": 2132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009481195104197063,
+      "loss": 3.1665,
+      "step": 2133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009480615813881188,
+      "loss": 3.1525,
+      "step": 2134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009480036218049288,
+      "loss": 3.3891,
+      "step": 2135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009479456316740882,
+      "loss": 3.0117,
+      "step": 2136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009478876109995511,
+      "loss": 3.2188,
+      "step": 2137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009478295597852738,
+      "loss": 3.1191,
+      "step": 2138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009477714780352144,
+      "loss": 2.9559,
+      "step": 2139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009477133657533335,
+      "loss": 3.1294,
+      "step": 2140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009476552229435935,
+      "loss": 3.229,
+      "step": 2141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009475970496099588,
+      "loss": 3.0811,
+      "step": 2142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009475388457563963,
+      "loss": 3.2221,
+      "step": 2143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009474806113868744,
+      "loss": 3.0086,
+      "step": 2144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000947422346505364,
+      "loss": 3.1209,
+      "step": 2145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009473640511158379,
+      "loss": 3.2367,
+      "step": 2146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009473057252222712,
+      "loss": 3.2952,
+      "step": 2147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009472473688286407,
+      "loss": 3.1478,
+      "step": 2148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009471889819389256,
+      "loss": 3.2385,
+      "step": 2149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009471305645571074,
+      "loss": 2.9362,
+      "step": 2150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009470721166871687,
+      "loss": 3.3845,
+      "step": 2151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009470136383330953,
+      "loss": 2.9816,
+      "step": 2152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009469551294988745,
+      "loss": 3.1488,
+      "step": 2153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009468965901884958,
+      "loss": 3.1905,
+      "step": 2154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009468380204059506,
+      "loss": 3.1982,
+      "step": 2155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009467794201552329,
+      "loss": 3.1735,
+      "step": 2156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009467207894403381,
+      "loss": 3.0648,
+      "step": 2157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009466621282652643,
+      "loss": 3.0836,
+      "step": 2158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.000946603436634011,
+      "loss": 2.9925,
+      "step": 2159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009465447145505806,
+      "loss": 3.0724,
+      "step": 2160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009464859620189769,
+      "loss": 3.2423,
+      "step": 2161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009464271790432059,
+      "loss": 3.2972,
+      "step": 2162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009463683656272759,
+      "loss": 3.0816,
+      "step": 2163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009463095217751974,
+      "loss": 3.1155,
+      "step": 2164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009462506474909822,
+      "loss": 3.2191,
+      "step": 2165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009461917427786452,
+      "loss": 3.2084,
+      "step": 2166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009461328076422027,
+      "loss": 3.2307,
+      "step": 2167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009460738420856732,
+      "loss": 3.2049,
+      "step": 2168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009460148461130774,
+      "loss": 3.2749,
+      "step": 2169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0009459558197284382,
+      "loss": 3.0261,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00094589676293578,
+      "loss": 3.0742,
+      "step": 2171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009458376757391298,
+      "loss": 2.9989,
+      "step": 2172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009457785581425167,
+      "loss": 3.2026,
+      "step": 2173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009457194101499715,
+      "loss": 3.5087,
+      "step": 2174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009456602317655274,
+      "loss": 3.1868,
+      "step": 2175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009456010229932194,
+      "loss": 3.2346,
+      "step": 2176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000945541783837085,
+      "loss": 3.3234,
+      "step": 2177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009454825143011634,
+      "loss": 3.281,
+      "step": 2178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009454232143894957,
+      "loss": 3.1875,
+      "step": 2179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009453638841061255,
+      "loss": 3.1973,
+      "step": 2180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009453045234550983,
+      "loss": 2.9596,
+      "step": 2181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009452451324404619,
+      "loss": 3.2102,
+      "step": 2182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009451857110662656,
+      "loss": 2.9704,
+      "step": 2183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009451262593365611,
+      "loss": 2.991,
+      "step": 2184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009450667772554026,
+      "loss": 3.2658,
+      "step": 2185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009450072648268456,
+      "loss": 3.247,
+      "step": 2186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009449477220549481,
+      "loss": 3.1001,
+      "step": 2187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00094488814894377,
+      "loss": 3.0482,
+      "step": 2188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009448285454973737,
+      "loss": 3.3143,
+      "step": 2189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009447689117198229,
+      "loss": 3.1369,
+      "step": 2190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009447092476151842,
+      "loss": 3.3253,
+      "step": 2191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009446495531875255,
+      "loss": 3.0635,
+      "step": 2192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009445898284409173,
+      "loss": 3.0983,
+      "step": 2193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009445300733794322,
+      "loss": 3.2155,
+      "step": 2194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009444702880071443,
+      "loss": 3.2898,
+      "step": 2195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009444104723281304,
+      "loss": 3.2233,
+      "step": 2196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000944350626346469,
+      "loss": 3.2684,
+      "step": 2197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009442907500662408,
+      "loss": 3.0472,
+      "step": 2198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009442308434915285,
+      "loss": 3.3139,
+      "step": 2199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009441709066264168,
+      "loss": 3.236,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009441109394749929,
+      "loss": 3.1856,
+      "step": 2201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009440509420413453,
+      "loss": 3.046,
+      "step": 2202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009439909143295653,
+      "loss": 3.2443,
+      "step": 2203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009439308563437459,
+      "loss": 3.0536,
+      "step": 2204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009438707680879821,
+      "loss": 3.1082,
+      "step": 2205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009438106495663712,
+      "loss": 3.0431,
+      "step": 2206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009437505007830125,
+      "loss": 3.203,
+      "step": 2207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009436903217420072,
+      "loss": 3.1079,
+      "step": 2208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009436301124474587,
+      "loss": 3.2425,
+      "step": 2209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009435698729034725,
+      "loss": 3.18,
+      "step": 2210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009435096031141561,
+      "loss": 3.3174,
+      "step": 2211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000943449303083619,
+      "loss": 3.0308,
+      "step": 2212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009433889728159728,
+      "loss": 3.3028,
+      "step": 2213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009433286123153313,
+      "loss": 3.0997,
+      "step": 2214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009432682215858102,
+      "loss": 3.388,
+      "step": 2215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009432078006315274,
+      "loss": 3.0324,
+      "step": 2216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009431473494566025,
+      "loss": 3.2841,
+      "step": 2217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009430868680651578,
+      "loss": 3.2278,
+      "step": 2218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009430263564613171,
+      "loss": 3.1714,
+      "step": 2219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009429658146492064,
+      "loss": 3.1577,
+      "step": 2220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009429052426329538,
+      "loss": 3.2466,
+      "step": 2221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009428446404166897,
+      "loss": 3.1538,
+      "step": 2222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009427840080045462,
+      "loss": 3.0916,
+      "step": 2223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009427233454006575,
+      "loss": 3.0335,
+      "step": 2224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009426626526091602,
+      "loss": 3.2336,
+      "step": 2225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009426019296341924,
+      "loss": 3.1106,
+      "step": 2226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009425411764798947,
+      "loss": 3.1763,
+      "step": 2227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009424803931504096,
+      "loss": 3.2919,
+      "step": 2228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009424195796498818,
+      "loss": 3.1936,
+      "step": 2229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009423587359824579,
+      "loss": 3.1162,
+      "step": 2230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009422978621522864,
+      "loss": 3.1212,
+      "step": 2231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009422369581635184,
+      "loss": 3.1941,
+      "step": 2232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421760240203063,
+      "loss": 3.0961,
+      "step": 2233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009421150597268053,
+      "loss": 3.1644,
+      "step": 2234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009420540652871723,
+      "loss": 3.0689,
+      "step": 2235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009419930407055661,
+      "loss": 3.1853,
+      "step": 2236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009419319859861478,
+      "loss": 2.8803,
+      "step": 2237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009418709011330805,
+      "loss": 3.2155,
+      "step": 2238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009418097861505294,
+      "loss": 3.2355,
+      "step": 2239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009417486410426615,
+      "loss": 3.2073,
+      "step": 2240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009416874658136463,
+      "loss": 2.9378,
+      "step": 2241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000941626260467655,
+      "loss": 3.0808,
+      "step": 2242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000941565025008861,
+      "loss": 3.2581,
+      "step": 2243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009415037594414395,
+      "loss": 3.21,
+      "step": 2244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009414424637695682,
+      "loss": 3.3306,
+      "step": 2245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009413811379974267,
+      "loss": 3.1531,
+      "step": 2246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009413197821291961,
+      "loss": 3.0834,
+      "step": 2247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009412583961690606,
+      "loss": 3.1437,
+      "step": 2248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009411969801212055,
+      "loss": 2.872,
+      "step": 2249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009411355339898187,
+      "loss": 3.2629,
+      "step": 2250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00094107405777909,
+      "loss": 2.924,
+      "step": 2251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009410125514932109,
+      "loss": 3.1421,
+      "step": 2252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009409510151363756,
+      "loss": 3.1724,
+      "step": 2253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00094088944871278,
+      "loss": 3.2712,
+      "step": 2254
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000940827852226622,
+      "loss": 3.2815,
+      "step": 2255
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009407662256821017,
+      "loss": 3.0395,
+      "step": 2256
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000940704569083421,
+      "loss": 2.9422,
+      "step": 2257
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009406428824347843,
+      "loss": 3.2695,
+      "step": 2258
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009405811657403975,
+      "loss": 3.4097,
+      "step": 2259
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009405194190044692,
+      "loss": 3.2336,
+      "step": 2260
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009404576422312092,
+      "loss": 3.2969,
+      "step": 2261
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009403958354248301,
+      "loss": 3.3763,
+      "step": 2262
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009403339985895462,
+      "loss": 3.1906,
+      "step": 2263
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000940272131729574,
+      "loss": 3.1171,
+      "step": 2264
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009402102348491319,
+      "loss": 3.1792,
+      "step": 2265
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009401483079524405,
+      "loss": 3.3078,
+      "step": 2266
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009400863510437221,
+      "loss": 3.4001,
+      "step": 2267
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009400243641272015,
+      "loss": 2.9742,
+      "step": 2268
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009399623472071055,
+      "loss": 3.3219,
+      "step": 2269
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009399003002876625,
+      "loss": 2.9704,
+      "step": 2270
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009398382233731033,
+      "loss": 3.1529,
+      "step": 2271
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009397761164676608,
+      "loss": 3.4692,
+      "step": 2272
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009397139795755699,
+      "loss": 3.328,
+      "step": 2273
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009396518127010671,
+      "loss": 3.0782,
+      "step": 2274
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009395896158483918,
+      "loss": 3.1604,
+      "step": 2275
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009395273890217848,
+      "loss": 3.117,
+      "step": 2276
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009394651322254889,
+      "loss": 3.3231,
+      "step": 2277
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009394028454637493,
+      "loss": 3.0909,
+      "step": 2278
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009393405287408132,
+      "loss": 3.2088,
+      "step": 2279
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009392781820609296,
+      "loss": 3.0075,
+      "step": 2280
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009392158054283497,
+      "loss": 3.0462,
+      "step": 2281
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009391533988473269,
+      "loss": 3.1292,
+      "step": 2282
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009390909623221163,
+      "loss": 3.0785,
+      "step": 2283
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000939028495856975,
+      "loss": 3.0526,
+      "step": 2284
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009389659994561625,
+      "loss": 3.1643,
+      "step": 2285
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009389034731239406,
+      "loss": 3.1908,
+      "step": 2286
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009388409168645721,
+      "loss": 3.2059,
+      "step": 2287
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.000938778330682323,
+      "loss": 3.2098,
+      "step": 2288
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009387157145814604,
+      "loss": 3.0468,
+      "step": 2289
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009386530685662543,
+      "loss": 3.1142,
+      "step": 2290
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009385903926409757,
+      "loss": 2.9802,
+      "step": 2291
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009385276868098986,
+      "loss": 3.2787,
+      "step": 2292
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009384649510772987,
+      "loss": 3.1307,
+      "step": 2293
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0009384021854474536,
+      "loss": 3.1637,
+      "step": 2294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009383393899246431,
+      "loss": 3.0011,
+      "step": 2295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000938276564513149,
+      "loss": 3.1586,
+      "step": 2296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000938213709217255,
+      "loss": 3.2847,
+      "step": 2297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000938150824041247,
+      "loss": 3.1018,
+      "step": 2298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009380879089894132,
+      "loss": 2.9724,
+      "step": 2299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000938024964066043,
+      "loss": 3.2297,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009379619892754288,
+      "loss": 3.343,
+      "step": 2301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009378989846218645,
+      "loss": 3.1114,
+      "step": 2302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000937835950109646,
+      "loss": 3.0323,
+      "step": 2303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009377728857430716,
+      "loss": 3.1714,
+      "step": 2304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009377097915264412,
+      "loss": 3.3014,
+      "step": 2305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009376466674640571,
+      "loss": 3.3893,
+      "step": 2306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009375835135602235,
+      "loss": 3.1768,
+      "step": 2307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009375203298192466,
+      "loss": 3.1404,
+      "step": 2308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009374571162454347,
+      "loss": 3.0931,
+      "step": 2309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009373938728430978,
+      "loss": 2.9164,
+      "step": 2310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009373305996165487,
+      "loss": 3.1437,
+      "step": 2311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009372672965701013,
+      "loss": 2.9222,
+      "step": 2312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009372039637080723,
+      "loss": 3.1259,
+      "step": 2313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00093714060103478,
+      "loss": 3.1783,
+      "step": 2314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000937077208554545,
+      "loss": 2.9697,
+      "step": 2315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009370137862716895,
+      "loss": 3.2392,
+      "step": 2316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009369503341905382,
+      "loss": 3.088,
+      "step": 2317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009368868523154177,
+      "loss": 3.1425,
+      "step": 2318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009368233406506566,
+      "loss": 3.1299,
+      "step": 2319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009367597992005854,
+      "loss": 3.4432,
+      "step": 2320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009366962279695368,
+      "loss": 3.0167,
+      "step": 2321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009366326269618455,
+      "loss": 3.1098,
+      "step": 2322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009365689961818482,
+      "loss": 3.185,
+      "step": 2323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009365053356338836,
+      "loss": 2.992,
+      "step": 2324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009364416453222924,
+      "loss": 3.1714,
+      "step": 2325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009363779252514178,
+      "loss": 3.1217,
+      "step": 2326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009363141754256039,
+      "loss": 3.1038,
+      "step": 2327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009362503958491984,
+      "loss": 3.4535,
+      "step": 2328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009361865865265495,
+      "loss": 2.9863,
+      "step": 2329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009361227474620085,
+      "loss": 3.0006,
+      "step": 2330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009360588786599282,
+      "loss": 3.0507,
+      "step": 2331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009359949801246634,
+      "loss": 2.9297,
+      "step": 2332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009359310518605716,
+      "loss": 3.2237,
+      "step": 2333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009358670938720113,
+      "loss": 2.9901,
+      "step": 2334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009358031061633439,
+      "loss": 2.9746,
+      "step": 2335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009357390887389322,
+      "loss": 3.0403,
+      "step": 2336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009356750416031415,
+      "loss": 3.1434,
+      "step": 2337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009356109647603388,
+      "loss": 3.1438,
+      "step": 2338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009355468582148933,
+      "loss": 3.34,
+      "step": 2339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009354827219711764,
+      "loss": 3.4392,
+      "step": 2340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000935418556033561,
+      "loss": 3.1858,
+      "step": 2341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009353543604064224,
+      "loss": 3.2417,
+      "step": 2342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009352901350941379,
+      "loss": 3.142,
+      "step": 2343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009352258801010869,
+      "loss": 3.2089,
+      "step": 2344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009351615954316504,
+      "loss": 3.1945,
+      "step": 2345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009350972810902118,
+      "loss": 3.2388,
+      "step": 2346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009350329370811567,
+      "loss": 3.2004,
+      "step": 2347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009349685634088725,
+      "loss": 3.2258,
+      "step": 2348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009349041600777481,
+      "loss": 3.2965,
+      "step": 2349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009348397270921753,
+      "loss": 3.1925,
+      "step": 2350
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009347752644565476,
+      "loss": 3.0239,
+      "step": 2351
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009347107721752601,
+      "loss": 2.9811,
+      "step": 2352
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009346462502527106,
+      "loss": 3.116,
+      "step": 2353
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009345816986932987,
+      "loss": 3.261,
+      "step": 2354
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009345171175014255,
+      "loss": 3.2478,
+      "step": 2355
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009344525066814948,
+      "loss": 3.085,
+      "step": 2356
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009343878662379122,
+      "loss": 3.2377,
+      "step": 2357
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009343231961750853,
+      "loss": 3.2978,
+      "step": 2358
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009342584964974235,
+      "loss": 3.1119,
+      "step": 2359
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009341937672093385,
+      "loss": 3.1321,
+      "step": 2360
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009341290083152441,
+      "loss": 2.9889,
+      "step": 2361
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009340642198195559,
+      "loss": 3.2011,
+      "step": 2362
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009339994017266915,
+      "loss": 3.2141,
+      "step": 2363
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009339345540410706,
+      "loss": 3.093,
+      "step": 2364
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009338696767671148,
+      "loss": 3.0869,
+      "step": 2365
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009338047699092482,
+      "loss": 3.1262,
+      "step": 2366
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009337398334718963,
+      "loss": 2.8798,
+      "step": 2367
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009336748674594868,
+      "loss": 3.2454,
+      "step": 2368
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009336098718764497,
+      "loss": 3.1772,
+      "step": 2369
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009335448467272166,
+      "loss": 3.364,
+      "step": 2370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009334797920162215,
+      "loss": 3.2504,
+      "step": 2371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009334147077478999,
+      "loss": 3.0734,
+      "step": 2372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00093334959392669,
+      "loss": 3.0681,
+      "step": 2373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009332844505570316,
+      "loss": 3.0705,
+      "step": 2374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009332192776433664,
+      "loss": 3.0381,
+      "step": 2375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009331540751901384,
+      "loss": 3.2863,
+      "step": 2376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009330888432017936,
+      "loss": 3.1424,
+      "step": 2377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009330235816827799,
+      "loss": 2.9309,
+      "step": 2378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009329582906375469,
+      "loss": 3.0702,
+      "step": 2379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009328929700705471,
+      "loss": 3.1784,
+      "step": 2380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000932827619986234,
+      "loss": 3.0246,
+      "step": 2381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009327622403890636,
+      "loss": 3.0585,
+      "step": 2382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009326968312834943,
+      "loss": 3.1678,
+      "step": 2383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009326313926739857,
+      "loss": 3.1848,
+      "step": 2384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000932565924565,
+      "loss": 3.3003,
+      "step": 2385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000932500426961001,
+      "loss": 3.1943,
+      "step": 2386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009324348998664549,
+      "loss": 3.1117,
+      "step": 2387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009323693432858298,
+      "loss": 3.0022,
+      "step": 2388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009323037572235958,
+      "loss": 3.109,
+      "step": 2389
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009322381416842245,
+      "loss": 3.0688,
+      "step": 2390
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009321724966721905,
+      "loss": 2.9652,
+      "step": 2391
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009321068221919697,
+      "loss": 3.0488,
+      "step": 2392
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009320411182480401,
+      "loss": 3.2258,
+      "step": 2393
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931975384844882,
+      "loss": 3.1573,
+      "step": 2394
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009319096219869774,
+      "loss": 3.1968,
+      "step": 2395
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009318438296788102,
+      "loss": 2.9929,
+      "step": 2396
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009317780079248669,
+      "loss": 3.153,
+      "step": 2397
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009317121567296354,
+      "loss": 2.971,
+      "step": 2398
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009316462760976059,
+      "loss": 3.1818,
+      "step": 2399
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009315803660332706,
+      "loss": 3.2512,
+      "step": 2400
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009315144265411235,
+      "loss": 3.0899,
+      "step": 2401
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931448457625661,
+      "loss": 3.0546,
+      "step": 2402
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009313824592913811,
+      "loss": 3.3175,
+      "step": 2403
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000931316431542784,
+      "loss": 3.1401,
+      "step": 2404
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009312503743843719,
+      "loss": 3.0965,
+      "step": 2405
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009311842878206489,
+      "loss": 3.0802,
+      "step": 2406
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009311181718561212,
+      "loss": 3.2071,
+      "step": 2407
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009310520264952973,
+      "loss": 3.5132,
+      "step": 2408
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.000930985851742687,
+      "loss": 3.0145,
+      "step": 2409
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009309196476028027,
+      "loss": 3.1463,
+      "step": 2410
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009308534140801585,
+      "loss": 3.2715,
+      "step": 2411
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009307871511792708,
+      "loss": 3.0768,
+      "step": 2412
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009307208589046576,
+      "loss": 3.3277,
+      "step": 2413
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009306545372608393,
+      "loss": 3.1434,
+      "step": 2414
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009305881862523379,
+      "loss": 3.1428,
+      "step": 2415
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009305218058836777,
+      "loss": 3.162,
+      "step": 2416
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009304553961593851,
+      "loss": 3.018,
+      "step": 2417
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0009303889570839882,
+      "loss": 3.1793,
+      "step": 2418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009303224886620172,
+      "loss": 3.211,
+      "step": 2419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009302559908980043,
+      "loss": 3.2954,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301894637964839,
+      "loss": 3.1827,
+      "step": 2421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009301229073619919,
+      "loss": 3.2274,
+      "step": 2422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009300563215990667,
+      "loss": 3.1652,
+      "step": 2423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009299897065122488,
+      "loss": 3.2129,
+      "step": 2424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00092992306210608,
+      "loss": 3.0883,
+      "step": 2425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009298563883851047,
+      "loss": 3.0114,
+      "step": 2426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009297896853538693,
+      "loss": 3.115,
+      "step": 2427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009297229530169218,
+      "loss": 3.3499,
+      "step": 2428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009296561913788124,
+      "loss": 3.1908,
+      "step": 2429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009295894004440934,
+      "loss": 3.0245,
+      "step": 2430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009295225802173193,
+      "loss": 3.0303,
+      "step": 2431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009294557307030457,
+      "loss": 2.9386,
+      "step": 2432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009293888519058313,
+      "loss": 2.9733,
+      "step": 2433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009293219438302362,
+      "loss": 3.117,
+      "step": 2434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009292550064808225,
+      "loss": 3.0383,
+      "step": 2435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009291880398621544,
+      "loss": 3.1715,
+      "step": 2436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009291210439787982,
+      "loss": 3.1472,
+      "step": 2437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009290540188353221,
+      "loss": 2.983,
+      "step": 2438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009289869644362963,
+      "loss": 3.1058,
+      "step": 2439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009289198807862928,
+      "loss": 2.8452,
+      "step": 2440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009288527678898859,
+      "loss": 3.0262,
+      "step": 2441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009287856257516519,
+      "loss": 3.095,
+      "step": 2442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009287184543761689,
+      "loss": 3.178,
+      "step": 2443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009286512537680168,
+      "loss": 3.3525,
+      "step": 2444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000928584023931778,
+      "loss": 3.0869,
+      "step": 2445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009285167648720366,
+      "loss": 3.3294,
+      "step": 2446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009284494765933787,
+      "loss": 3.0679,
+      "step": 2447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009283821591003926,
+      "loss": 3.0138,
+      "step": 2448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009283148123976681,
+      "loss": 3.3478,
+      "step": 2449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009282474364897977,
+      "loss": 3.0176,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009281800313813752,
+      "loss": 3.2475,
+      "step": 2451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009281125970769968,
+      "loss": 3.0026,
+      "step": 2452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009280451335812605,
+      "loss": 3.2676,
+      "step": 2453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009279776408987665,
+      "loss": 2.9836,
+      "step": 2454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009279101190341171,
+      "loss": 3.1312,
+      "step": 2455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009278425679919158,
+      "loss": 3.1196,
+      "step": 2456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009277749877767689,
+      "loss": 3.1918,
+      "step": 2457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009277073783932846,
+      "loss": 3.0923,
+      "step": 2458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009276397398460727,
+      "loss": 3.0234,
+      "step": 2459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009275720721397454,
+      "loss": 3.2287,
+      "step": 2460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009275043752789168,
+      "loss": 3.1114,
+      "step": 2461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009274366492682024,
+      "loss": 3.1025,
+      "step": 2462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009273688941122206,
+      "loss": 3.2576,
+      "step": 2463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009273011098155912,
+      "loss": 3.26,
+      "step": 2464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009272332963829362,
+      "loss": 3.1026,
+      "step": 2465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009271654538188795,
+      "loss": 2.9262,
+      "step": 2466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009270975821280471,
+      "loss": 3.1313,
+      "step": 2467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000927029681315067,
+      "loss": 3.1148,
+      "step": 2468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009269617513845687,
+      "loss": 3.1495,
+      "step": 2469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009268937923411844,
+      "loss": 3.2707,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009268258041895481,
+      "loss": 3.3275,
+      "step": 2471
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009267577869342953,
+      "loss": 3.1979,
+      "step": 2472
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926689740580064,
+      "loss": 3.2242,
+      "step": 2473
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926621665131494,
+      "loss": 3.153,
+      "step": 2474
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009265535605932273,
+      "loss": 3.2349,
+      "step": 2475
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009264854269699075,
+      "loss": 3.1799,
+      "step": 2476
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009264172642661801,
+      "loss": 3.1622,
+      "step": 2477
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009263490724866934,
+      "loss": 3.2143,
+      "step": 2478
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009262808516360967,
+      "loss": 3.2972,
+      "step": 2479
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000926212601719042,
+      "loss": 3.2101,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009261443227401829,
+      "loss": 3.1944,
+      "step": 2481
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009260760147041751,
+      "loss": 3.163,
+      "step": 2482
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009260076776156762,
+      "loss": 2.9409,
+      "step": 2483
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009259393114793459,
+      "loss": 3.1669,
+      "step": 2484
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009258709162998456,
+      "loss": 3.2077,
+      "step": 2485
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009258024920818393,
+      "loss": 3.0132,
+      "step": 2486
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009257340388299924,
+      "loss": 3.2972,
+      "step": 2487
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009256655565489725,
+      "loss": 3.2143,
+      "step": 2488
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000925597045243449,
+      "loss": 2.9439,
+      "step": 2489
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009255285049180935,
+      "loss": 3.3409,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009254599355775797,
+      "loss": 3.2025,
+      "step": 2491
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009253913372265827,
+      "loss": 3.1063,
+      "step": 2492
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009253227098697803,
+      "loss": 3.1622,
+      "step": 2493
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009252540535118519,
+      "loss": 3.1381,
+      "step": 2494
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009251853681574787,
+      "loss": 3.2768,
+      "step": 2495
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009251166538113442,
+      "loss": 3.289,
+      "step": 2496
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009250479104781338,
+      "loss": 3.0819,
+      "step": 2497
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009249791381625347,
+      "loss": 3.1811,
+      "step": 2498
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009249103368692365,
+      "loss": 3.3121,
+      "step": 2499
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009248415066029303,
+      "loss": 3.1045,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009247726473683095,
+      "loss": 3.1379,
+      "step": 2501
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009247037591700691,
+      "loss": 3.1503,
+      "step": 2502
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009246348420129066,
+      "loss": 3.0157,
+      "step": 2503
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009245658959015211,
+      "loss": 2.9729,
+      "step": 2504
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009244969208406135,
+      "loss": 3.4003,
+      "step": 2505
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009244279168348874,
+      "loss": 3.1512,
+      "step": 2506
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009243588838890478,
+      "loss": 3.2324,
+      "step": 2507
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009242898220078016,
+      "loss": 3.0703,
+      "step": 2508
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000924220731195858,
+      "loss": 3.1401,
+      "step": 2509
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000924151611457928,
+      "loss": 3.0567,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009240824627987246,
+      "loss": 3.1776,
+      "step": 2511
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009240132852229628,
+      "loss": 3.1503,
+      "step": 2512
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009239440787353597,
+      "loss": 3.196,
+      "step": 2513
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009238748433406339,
+      "loss": 3.0041,
+      "step": 2514
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009238055790435065,
+      "loss": 3.1083,
+      "step": 2515
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009237362858487004,
+      "loss": 3.1357,
+      "step": 2516
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009236669637609404,
+      "loss": 3.1913,
+      "step": 2517
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009235976127849533,
+      "loss": 2.9897,
+      "step": 2518
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009235282329254677,
+      "loss": 3.2725,
+      "step": 2519
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009234588241872146,
+      "loss": 3.1853,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009233893865749267,
+      "loss": 3.1806,
+      "step": 2521
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009233199200933387,
+      "loss": 3.4171,
+      "step": 2522
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009232504247471869,
+      "loss": 3.0851,
+      "step": 2523
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009231809005412104,
+      "loss": 3.1921,
+      "step": 2524
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009231113474801496,
+      "loss": 3.1868,
+      "step": 2525
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009230417655687468,
+      "loss": 3.2254,
+      "step": 2526
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000922972154811747,
+      "loss": 3.1026,
+      "step": 2527
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009229025152138964,
+      "loss": 3.2927,
+      "step": 2528
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009228328467799434,
+      "loss": 3.2321,
+      "step": 2529
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009227631495146387,
+      "loss": 2.9357,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009226934234227344,
+      "loss": 3.4061,
+      "step": 2531
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009226236685089851,
+      "loss": 3.0904,
+      "step": 2532
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009225538847781467,
+      "loss": 3.2379,
+      "step": 2533
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009224840722349781,
+      "loss": 2.9318,
+      "step": 2534
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009224142308842391,
+      "loss": 3.0754,
+      "step": 2535
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000922344360730692,
+      "loss": 3.2366,
+      "step": 2536
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009222744617791011,
+      "loss": 3.1453,
+      "step": 2537
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009222045340342325,
+      "loss": 3.119,
+      "step": 2538
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009221345775008541,
+      "loss": 2.9965,
+      "step": 2539
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009220645921837363,
+      "loss": 3.1454,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0009219945780876509,
+      "loss": 3.1573,
+      "step": 2541
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.000921924535217372,
+      "loss": 3.3024,
+      "step": 2542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009218544635776755,
+      "loss": 2.9111,
+      "step": 2543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009217843631733393,
+      "loss": 3.1036,
+      "step": 2544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009217142340091434,
+      "loss": 3.3439,
+      "step": 2545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009216440760898694,
+      "loss": 3.0889,
+      "step": 2546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009215738894203014,
+      "loss": 3.0026,
+      "step": 2547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000921503674005225,
+      "loss": 3.0763,
+      "step": 2548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009214334298494279,
+      "loss": 3.3115,
+      "step": 2549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009213631569576998,
+      "loss": 3.1549,
+      "step": 2550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009212928553348324,
+      "loss": 3.1019,
+      "step": 2551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009212225249856192,
+      "loss": 3.2652,
+      "step": 2552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009211521659148557,
+      "loss": 3.0991,
+      "step": 2553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009210817781273397,
+      "loss": 3.3027,
+      "step": 2554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009210113616278703,
+      "loss": 3.1036,
+      "step": 2555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009209409164212493,
+      "loss": 3.0371,
+      "step": 2556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009208704425122798,
+      "loss": 3.1513,
+      "step": 2557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009207999399057674,
+      "loss": 3.1953,
+      "step": 2558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009207294086065192,
+      "loss": 3.2845,
+      "step": 2559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009206588486193445,
+      "loss": 3.1092,
+      "step": 2560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009205882599490544,
+      "loss": 3.1072,
+      "step": 2561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009205176426004625,
+      "loss": 3.004,
+      "step": 2562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009204469965783835,
+      "loss": 3.2235,
+      "step": 2563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009203763218876346,
+      "loss": 3.2808,
+      "step": 2564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009203056185330347,
+      "loss": 3.3219,
+      "step": 2565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009202348865194052,
+      "loss": 3.1728,
+      "step": 2566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009201641258515688,
+      "loss": 3.2248,
+      "step": 2567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009200933365343502,
+      "loss": 3.1902,
+      "step": 2568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009200225185725764,
+      "loss": 2.898,
+      "step": 2569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009199516719710763,
+      "loss": 3.1973,
+      "step": 2570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009198807967346807,
+      "loss": 3.2185,
+      "step": 2571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009198098928682221,
+      "loss": 3.2691,
+      "step": 2572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009197389603765355,
+      "loss": 3.2989,
+      "step": 2573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009196679992644571,
+      "loss": 3.0587,
+      "step": 2574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009195970095368257,
+      "loss": 3.1163,
+      "step": 2575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009195259911984818,
+      "loss": 3.2458,
+      "step": 2576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009194549442542678,
+      "loss": 3.3557,
+      "step": 2577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009193838687090282,
+      "loss": 2.9733,
+      "step": 2578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009193127645676093,
+      "loss": 3.2433,
+      "step": 2579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009192416318348595,
+      "loss": 3.0923,
+      "step": 2580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009191704705156289,
+      "loss": 3.1148,
+      "step": 2581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00091909928061477,
+      "loss": 3.2172,
+      "step": 2582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009190280621371367,
+      "loss": 3.0779,
+      "step": 2583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009189568150875852,
+      "loss": 3.14,
+      "step": 2584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009188855394709737,
+      "loss": 3.0261,
+      "step": 2585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009188142352921621,
+      "loss": 3.4133,
+      "step": 2586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009187429025560121,
+      "loss": 3.2328,
+      "step": 2587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000918671541267388,
+      "loss": 3.1,
+      "step": 2588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009186001514311557,
+      "loss": 3.0348,
+      "step": 2589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009185287330521826,
+      "loss": 3.0534,
+      "step": 2590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009184572861353386,
+      "loss": 3.3346,
+      "step": 2591
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009183858106854957,
+      "loss": 3.0746,
+      "step": 2592
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009183143067075272,
+      "loss": 2.9533,
+      "step": 2593
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009182427742063086,
+      "loss": 3.0342,
+      "step": 2594
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009181712131867177,
+      "loss": 3.023,
+      "step": 2595
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000918099623653634,
+      "loss": 3.2681,
+      "step": 2596
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009180280056119385,
+      "loss": 3.0701,
+      "step": 2597
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009179563590665151,
+      "loss": 3.2323,
+      "step": 2598
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009178846840222488,
+      "loss": 3.1007,
+      "step": 2599
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009178129804840267,
+      "loss": 3.0857,
+      "step": 2600
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009177412484567385,
+      "loss": 3.2872,
+      "step": 2601
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009176694879452748,
+      "loss": 3.1885,
+      "step": 2602
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009175976989545289,
+      "loss": 3.1463,
+      "step": 2603
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009175258814893959,
+      "loss": 3.0006,
+      "step": 2604
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009174540355547725,
+      "loss": 3.1439,
+      "step": 2605
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009173821611555579,
+      "loss": 2.9288,
+      "step": 2606
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009173102582966527,
+      "loss": 3.0997,
+      "step": 2607
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009172383269829598,
+      "loss": 2.9189,
+      "step": 2608
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000917166367219384,
+      "loss": 3.3521,
+      "step": 2609
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009170943790108318,
+      "loss": 3.0383,
+      "step": 2610
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009170223623622118,
+      "loss": 3.2154,
+      "step": 2611
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009169503172784348,
+      "loss": 3.1433,
+      "step": 2612
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009168782437644128,
+      "loss": 3.0021,
+      "step": 2613
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009168061418250607,
+      "loss": 3.2075,
+      "step": 2614
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009167340114652946,
+      "loss": 2.939,
+      "step": 2615
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009166618526900328,
+      "loss": 3.1036,
+      "step": 2616
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009165896655041957,
+      "loss": 3.2638,
+      "step": 2617
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009165174499127052,
+      "loss": 3.1819,
+      "step": 2618
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009164452059204857,
+      "loss": 3.2134,
+      "step": 2619
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009163729335324629,
+      "loss": 3.3522,
+      "step": 2620
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009163006327535651,
+      "loss": 3.205,
+      "step": 2621
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000916228303588722,
+      "loss": 3.1675,
+      "step": 2622
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009161559460428657,
+      "loss": 3.3189,
+      "step": 2623
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009160835601209296,
+      "loss": 3.0541,
+      "step": 2624
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009160111458278497,
+      "loss": 3.2172,
+      "step": 2625
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009159387031685638,
+      "loss": 3.2127,
+      "step": 2626
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000915866232148011,
+      "loss": 3.0939,
+      "step": 2627
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009157937327711332,
+      "loss": 3.0807,
+      "step": 2628
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009157212050428736,
+      "loss": 2.9352,
+      "step": 2629
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009156486489681779,
+      "loss": 3.3506,
+      "step": 2630
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009155760645519933,
+      "loss": 3.267,
+      "step": 2631
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009155034517992688,
+      "loss": 3.2128,
+      "step": 2632
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000915430810714956,
+      "loss": 3.0474,
+      "step": 2633
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009153581413040078,
+      "loss": 3.0277,
+      "step": 2634
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009152854435713791,
+      "loss": 2.9765,
+      "step": 2635
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000915212717522027,
+      "loss": 3.0792,
+      "step": 2636
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009151399631609108,
+      "loss": 3.0124,
+      "step": 2637
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009150671804929907,
+      "loss": 3.188,
+      "step": 2638
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009149943695232298,
+      "loss": 3.0331,
+      "step": 2639
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009149215302565927,
+      "loss": 3.1284,
+      "step": 2640
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009148486626980463,
+      "loss": 3.1607,
+      "step": 2641
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009147757668525588,
+      "loss": 3.4232,
+      "step": 2642
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914702842725101,
+      "loss": 3.06,
+      "step": 2643
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009146298903206449,
+      "loss": 2.9116,
+      "step": 2644
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009145569096441654,
+      "loss": 2.9208,
+      "step": 2645
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009144839007006382,
+      "loss": 3.3027,
+      "step": 2646
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000914410863495042,
+      "loss": 3.0028,
+      "step": 2647
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009143377980323567,
+      "loss": 3.1069,
+      "step": 2648
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009142647043175643,
+      "loss": 3.3535,
+      "step": 2649
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009141915823556487,
+      "loss": 3.2236,
+      "step": 2650
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009141184321515961,
+      "loss": 3.0938,
+      "step": 2651
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009140452537103941,
+      "loss": 3.2569,
+      "step": 2652
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009139720470370327,
+      "loss": 3.077,
+      "step": 2653
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009138988121365031,
+      "loss": 3.1049,
+      "step": 2654
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009138255490137995,
+      "loss": 3.181,
+      "step": 2655
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009137522576739171,
+      "loss": 3.0566,
+      "step": 2656
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009136789381218534,
+      "loss": 2.9127,
+      "step": 2657
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009136055903626077,
+      "loss": 3.3082,
+      "step": 2658
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009135322144011815,
+      "loss": 3.1825,
+      "step": 2659
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000913458810242578,
+      "loss": 2.8214,
+      "step": 2660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009133853778918022,
+      "loss": 3.1356,
+      "step": 2661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009133119173538613,
+      "loss": 3.0353,
+      "step": 2662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000913238428633764,
+      "loss": 3.0529,
+      "step": 2663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009131649117365216,
+      "loss": 3.0476,
+      "step": 2664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009130913666671467,
+      "loss": 2.9795,
+      "step": 2665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0009130177934306543,
+      "loss": 3.3053,
+      "step": 2666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009129441920320609,
+      "loss": 3.1764,
+      "step": 2667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000912870562476385,
+      "loss": 2.9578,
+      "step": 2668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009127969047686473,
+      "loss": 3.1255,
+      "step": 2669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009127232189138702,
+      "loss": 2.9132,
+      "step": 2670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009126495049170778,
+      "loss": 3.367,
+      "step": 2671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009125757627832969,
+      "loss": 3.2143,
+      "step": 2672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009125019925175551,
+      "loss": 3.2442,
+      "step": 2673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009124281941248831,
+      "loss": 3.0875,
+      "step": 2674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009123543676103124,
+      "loss": 3.3248,
+      "step": 2675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009122805129788774,
+      "loss": 3.0706,
+      "step": 2676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009122066302356135,
+      "loss": 3.0575,
+      "step": 2677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009121327193855589,
+      "loss": 3.1953,
+      "step": 2678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000912058780433753,
+      "loss": 3.3081,
+      "step": 2679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009119848133852377,
+      "loss": 3.0316,
+      "step": 2680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009119108182450562,
+      "loss": 3.1396,
+      "step": 2681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009118367950182542,
+      "loss": 3.2255,
+      "step": 2682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000911762743709879,
+      "loss": 3.407,
+      "step": 2683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009116886643249798,
+      "loss": 3.2212,
+      "step": 2684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009116145568686078,
+      "loss": 3.102,
+      "step": 2685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009115404213458163,
+      "loss": 3.004,
+      "step": 2686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009114662577616599,
+      "loss": 3.1291,
+      "step": 2687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009113920661211961,
+      "loss": 2.9853,
+      "step": 2688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009113178464294832,
+      "loss": 3.0747,
+      "step": 2689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009112435986915823,
+      "loss": 3.1191,
+      "step": 2690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009111693229125559,
+      "loss": 3.1217,
+      "step": 2691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009110950190974686,
+      "loss": 2.8085,
+      "step": 2692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009110206872513871,
+      "loss": 3.114,
+      "step": 2693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009109463273793794,
+      "loss": 3.2463,
+      "step": 2694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009108719394865161,
+      "loss": 3.1483,
+      "step": 2695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009107975235778696,
+      "loss": 3.0946,
+      "step": 2696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009107230796585136,
+      "loss": 3.0495,
+      "step": 2697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009106486077335245,
+      "loss": 3.0301,
+      "step": 2698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009105741078079799,
+      "loss": 3.3269,
+      "step": 2699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009104995798869601,
+      "loss": 3.2434,
+      "step": 2700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009104250239755465,
+      "loss": 3.2816,
+      "step": 2701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000910350440078823,
+      "loss": 3.0781,
+      "step": 2702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000910275828201875,
+      "loss": 3.1657,
+      "step": 2703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009102011883497902,
+      "loss": 3.3926,
+      "step": 2704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009101265205276581,
+      "loss": 3.0422,
+      "step": 2705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009100518247405697,
+      "loss": 3.1349,
+      "step": 2706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009099771009936183,
+      "loss": 3.1312,
+      "step": 2707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009099023492918992,
+      "loss": 3.2288,
+      "step": 2708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009098275696405091,
+      "loss": 3.0626,
+      "step": 2709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009097527620445474,
+      "loss": 3.2841,
+      "step": 2710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009096779265091147,
+      "loss": 2.8662,
+      "step": 2711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009096030630393135,
+      "loss": 3.2014,
+      "step": 2712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009095281716402489,
+      "loss": 3.3468,
+      "step": 2713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009094532523170271,
+      "loss": 3.0726,
+      "step": 2714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000909378305074757,
+      "loss": 3.1911,
+      "step": 2715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009093033299185485,
+      "loss": 3.1275,
+      "step": 2716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000909228326853514,
+      "loss": 3.2741,
+      "step": 2717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009091532958847676,
+      "loss": 3.4935,
+      "step": 2718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009090782370174258,
+      "loss": 3.1433,
+      "step": 2719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009090031502566062,
+      "loss": 3.344,
+      "step": 2720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009089280356074287,
+      "loss": 3.0418,
+      "step": 2721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000908852893075015,
+      "loss": 3.1331,
+      "step": 2722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009087777226644891,
+      "loss": 2.9864,
+      "step": 2723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009087025243809763,
+      "loss": 2.9925,
+      "step": 2724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009086272982296043,
+      "loss": 2.9021,
+      "step": 2725
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009085520442155024,
+      "loss": 3.2508,
+      "step": 2726
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009084767623438018,
+      "loss": 3.3903,
+      "step": 2727
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009084014526196355,
+      "loss": 3.1508,
+      "step": 2728
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000908326115048139,
+      "loss": 3.2067,
+      "step": 2729
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009082507496344491,
+      "loss": 3.1155,
+      "step": 2730
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009081753563837047,
+      "loss": 3.0776,
+      "step": 2731
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009080999353010463,
+      "loss": 3.0681,
+      "step": 2732
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009080244863916172,
+      "loss": 3.2966,
+      "step": 2733
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009079490096605612,
+      "loss": 3.123,
+      "step": 2734
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009078735051130254,
+      "loss": 3.0621,
+      "step": 2735
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009077979727541578,
+      "loss": 3.0659,
+      "step": 2736
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009077224125891091,
+      "loss": 3.264,
+      "step": 2737
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009076468246230308,
+      "loss": 3.2684,
+      "step": 2738
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009075712088610774,
+      "loss": 3.0383,
+      "step": 2739
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009074955653084048,
+      "loss": 3.1712,
+      "step": 2740
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009074198939701708,
+      "loss": 3.0439,
+      "step": 2741
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009073441948515351,
+      "loss": 3.2927,
+      "step": 2742
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009072684679576594,
+      "loss": 3.1339,
+      "step": 2743
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009071927132937071,
+      "loss": 3.049,
+      "step": 2744
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009071169308648438,
+      "loss": 2.9741,
+      "step": 2745
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009070411206762368,
+      "loss": 3.079,
+      "step": 2746
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000906965282733055,
+      "loss": 3.169,
+      "step": 2747
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009068894170404699,
+      "loss": 3.2501,
+      "step": 2748
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009068135236036543,
+      "loss": 3.3881,
+      "step": 2749
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009067376024277831,
+      "loss": 3.1517,
+      "step": 2750
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009066616535180331,
+      "loss": 3.1335,
+      "step": 2751
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009065856768795828,
+      "loss": 3.2278,
+      "step": 2752
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009065096725176131,
+      "loss": 3.0018,
+      "step": 2753
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009064336404373062,
+      "loss": 3.3142,
+      "step": 2754
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009063575806438465,
+      "loss": 3.1713,
+      "step": 2755
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009062814931424201,
+      "loss": 3.2861,
+      "step": 2756
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009062053779382153,
+      "loss": 3.0204,
+      "step": 2757
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009061292350364221,
+      "loss": 3.1146,
+      "step": 2758
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009060530644422321,
+      "loss": 3.0152,
+      "step": 2759
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009059768661608394,
+      "loss": 2.938,
+      "step": 2760
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009059006401974397,
+      "loss": 3.2662,
+      "step": 2761
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009058243865572303,
+      "loss": 3.2333,
+      "step": 2762
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009057481052454109,
+      "loss": 3.1525,
+      "step": 2763
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009056717962671825,
+      "loss": 3.385,
+      "step": 2764
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009055954596277487,
+      "loss": 3.1394,
+      "step": 2765
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009055190953323143,
+      "loss": 3.2823,
+      "step": 2766
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009054427033860865,
+      "loss": 3.1427,
+      "step": 2767
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000905366283794274,
+      "loss": 3.1493,
+      "step": 2768
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009052898365620877,
+      "loss": 3.3365,
+      "step": 2769
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009052133616947402,
+      "loss": 3.2229,
+      "step": 2770
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000905136859197446,
+      "loss": 3.1767,
+      "step": 2771
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009050603290754216,
+      "loss": 3.1681,
+      "step": 2772
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000904983771333885,
+      "loss": 3.0401,
+      "step": 2773
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009049071859780569,
+      "loss": 3.3561,
+      "step": 2774
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009048305730131589,
+      "loss": 3.1169,
+      "step": 2775
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009047539324444151,
+      "loss": 3.2094,
+      "step": 2776
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009046772642770513,
+      "loss": 3.0739,
+      "step": 2777
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009046005685162952,
+      "loss": 3.2027,
+      "step": 2778
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009045238451673767,
+      "loss": 3.0689,
+      "step": 2779
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009044470942355268,
+      "loss": 3.2892,
+      "step": 2780
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000904370315725979,
+      "loss": 3.1076,
+      "step": 2781
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009042935096439689,
+      "loss": 3.1457,
+      "step": 2782
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000904216675994733,
+      "loss": 3.2382,
+      "step": 2783
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009041398147835107,
+      "loss": 3.0334,
+      "step": 2784
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009040629260155427,
+      "loss": 3.1084,
+      "step": 2785
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009039860096960719,
+      "loss": 3.0657,
+      "step": 2786
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009039090658303429,
+      "loss": 3.0758,
+      "step": 2787
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.000903832094423602,
+      "loss": 3.1131,
+      "step": 2788
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009037550954810978,
+      "loss": 3.4273,
+      "step": 2789
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0009036780690080806,
+      "loss": 2.9404,
+      "step": 2790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009036010150098023,
+      "loss": 3.2673,
+      "step": 2791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009035239334915171,
+      "loss": 2.9733,
+      "step": 2792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009034468244584808,
+      "loss": 2.9781,
+      "step": 2793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009033696879159514,
+      "loss": 3.0989,
+      "step": 2794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009032925238691882,
+      "loss": 3.3033,
+      "step": 2795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000903215332323453,
+      "loss": 3.325,
+      "step": 2796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009031381132840091,
+      "loss": 3.0267,
+      "step": 2797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009030608667561215,
+      "loss": 3.0919,
+      "step": 2798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009029835927450579,
+      "loss": 3.3316,
+      "step": 2799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009029062912560869,
+      "loss": 3.247,
+      "step": 2800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009028289622944794,
+      "loss": 3.1706,
+      "step": 2801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009027516058655085,
+      "loss": 3.2575,
+      "step": 2802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009026742219744485,
+      "loss": 3.0418,
+      "step": 2803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000902596810626576,
+      "loss": 3.0984,
+      "step": 2804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009025193718271694,
+      "loss": 3.1634,
+      "step": 2805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009024419055815091,
+      "loss": 2.9691,
+      "step": 2806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009023644118948769,
+      "loss": 3.0823,
+      "step": 2807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009022868907725573,
+      "loss": 3.1488,
+      "step": 2808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009022093422198354,
+      "loss": 3.2131,
+      "step": 2809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009021317662419997,
+      "loss": 3.0936,
+      "step": 2810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009020541628443395,
+      "loss": 3.3198,
+      "step": 2811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009019765320321462,
+      "loss": 3.2414,
+      "step": 2812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009018988738107132,
+      "loss": 3.0296,
+      "step": 2813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009018211881853357,
+      "loss": 3.115,
+      "step": 2814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009017434751613109,
+      "loss": 3.1091,
+      "step": 2815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009016657347439378,
+      "loss": 3.1884,
+      "step": 2816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000901587966938517,
+      "loss": 3.3141,
+      "step": 2817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009015101717503514,
+      "loss": 3.1498,
+      "step": 2818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009014323491847453,
+      "loss": 3.0824,
+      "step": 2819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009013544992470053,
+      "loss": 3.1374,
+      "step": 2820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009012766219424397,
+      "loss": 3.2084,
+      "step": 2821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009011987172763586,
+      "loss": 3.1992,
+      "step": 2822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009011207852540742,
+      "loss": 3.1082,
+      "step": 2823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009010428258809,
+      "loss": 3.1813,
+      "step": 2824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000900964839162152,
+      "loss": 3.1226,
+      "step": 2825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009008868251031479,
+      "loss": 3.1502,
+      "step": 2826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009008087837092069,
+      "loss": 3.2757,
+      "step": 2827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009007307149856507,
+      "loss": 3.2093,
+      "step": 2828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009006526189378021,
+      "loss": 3.1338,
+      "step": 2829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009005744955709867,
+      "loss": 2.9761,
+      "step": 2830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009004963448905309,
+      "loss": 3.3093,
+      "step": 2831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009004181669017637,
+      "loss": 3.1917,
+      "step": 2832
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009003399616100159,
+      "loss": 3.304,
+      "step": 2833
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009002617290206199,
+      "loss": 3.2885,
+      "step": 2834
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009001834691389099,
+      "loss": 3.031,
+      "step": 2835
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009001051819702226,
+      "loss": 3.0327,
+      "step": 2836
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0009000268675198956,
+      "loss": 3.2158,
+      "step": 2837
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008999485257932691,
+      "loss": 2.9754,
+      "step": 2838
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899870156795685,
+      "loss": 3.2028,
+      "step": 2839
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008997917605324869,
+      "loss": 3.2006,
+      "step": 2840
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008997133370090202,
+      "loss": 3.2531,
+      "step": 2841
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008996348862306325,
+      "loss": 3.1655,
+      "step": 2842
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008995564082026731,
+      "loss": 3.1649,
+      "step": 2843
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008994779029304928,
+      "loss": 3.0669,
+      "step": 2844
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899399370419445,
+      "loss": 3.2592,
+      "step": 2845
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899320810674884,
+      "loss": 3.1399,
+      "step": 2846
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000899242223702167,
+      "loss": 3.1632,
+      "step": 2847
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008991636095066523,
+      "loss": 2.8546,
+      "step": 2848
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008990849680937003,
+      "loss": 3.1786,
+      "step": 2849
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008990062994686734,
+      "loss": 3.2327,
+      "step": 2850
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008989276036369356,
+      "loss": 3.2626,
+      "step": 2851
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008988488806038528,
+      "loss": 3.0066,
+      "step": 2852
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008987701303747928,
+      "loss": 3.1652,
+      "step": 2853
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008986913529551255,
+      "loss": 3.4161,
+      "step": 2854
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008986125483502223,
+      "loss": 3.1031,
+      "step": 2855
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008985337165654566,
+      "loss": 3.0786,
+      "step": 2856
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008984548576062037,
+      "loss": 3.0928,
+      "step": 2857
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008983759714778405,
+      "loss": 3.2825,
+      "step": 2858
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008982970581857461,
+      "loss": 3.1403,
+      "step": 2859
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008982181177353012,
+      "loss": 3.1101,
+      "step": 2860
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008981391501318885,
+      "loss": 3.1462,
+      "step": 2861
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008980601553808927,
+      "loss": 3.1263,
+      "step": 2862
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008979811334876997,
+      "loss": 3.265,
+      "step": 2863
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897902084457698,
+      "loss": 2.9788,
+      "step": 2864
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008978230082962777,
+      "loss": 3.0529,
+      "step": 2865
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008977439050088306,
+      "loss": 3.1081,
+      "step": 2866
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008976647746007504,
+      "loss": 3.0568,
+      "step": 2867
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008975856170774327,
+      "loss": 3.4331,
+      "step": 2868
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897506432444275,
+      "loss": 3.0784,
+      "step": 2869
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008974272207066767,
+      "loss": 3.0655,
+      "step": 2870
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008973479818700388,
+      "loss": 3.0651,
+      "step": 2871
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008972687159397642,
+      "loss": 3.1118,
+      "step": 2872
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008971894229212581,
+      "loss": 3.156,
+      "step": 2873
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008971101028199267,
+      "loss": 2.9,
+      "step": 2874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000897030755641179,
+      "loss": 3.1198,
+      "step": 2875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008969513813904251,
+      "loss": 3.2255,
+      "step": 2876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008968719800730773,
+      "loss": 3.157,
+      "step": 2877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008967925516945496,
+      "loss": 2.9854,
+      "step": 2878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008967130962602581,
+      "loss": 2.8699,
+      "step": 2879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008966336137756204,
+      "loss": 3.1302,
+      "step": 2880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008965541042460562,
+      "loss": 3.1091,
+      "step": 2881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008964745676769869,
+      "loss": 3.1206,
+      "step": 2882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008963950040738358,
+      "loss": 3.2383,
+      "step": 2883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000896315413442028,
+      "loss": 2.9663,
+      "step": 2884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008962357957869904,
+      "loss": 3.2107,
+      "step": 2885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008961561511141522,
+      "loss": 2.8944,
+      "step": 2886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008960764794289437,
+      "loss": 3.1268,
+      "step": 2887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008959967807367975,
+      "loss": 3.0588,
+      "step": 2888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008959170550431479,
+      "loss": 3.1155,
+      "step": 2889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008958373023534312,
+      "loss": 3.2491,
+      "step": 2890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008957575226730854,
+      "loss": 3.0004,
+      "step": 2891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008956777160075503,
+      "loss": 3.4451,
+      "step": 2892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008955978823622677,
+      "loss": 3.076,
+      "step": 2893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000895518021742681,
+      "loss": 3.1677,
+      "step": 2894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008954381341542359,
+      "loss": 3.2016,
+      "step": 2895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008953582196023793,
+      "loss": 3.0931,
+      "step": 2896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008952782780925606,
+      "loss": 3.2536,
+      "step": 2897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008951983096302304,
+      "loss": 3.3198,
+      "step": 2898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008951183142208413,
+      "loss": 2.9925,
+      "step": 2899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008950382918698483,
+      "loss": 3.0629,
+      "step": 2900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008949582425827077,
+      "loss": 3.0192,
+      "step": 2901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008948781663648777,
+      "loss": 3.2223,
+      "step": 2902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008947980632218182,
+      "loss": 3.2877,
+      "step": 2903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008947179331589915,
+      "loss": 2.7457,
+      "step": 2904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008946377761818612,
+      "loss": 2.9115,
+      "step": 2905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008945575922958928,
+      "loss": 3.065,
+      "step": 2906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008944773815065539,
+      "loss": 3.0713,
+      "step": 2907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008943971438193135,
+      "loss": 2.9516,
+      "step": 2908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.000894316879239643,
+      "loss": 3.2133,
+      "step": 2909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008942365877730152,
+      "loss": 3.067,
+      "step": 2910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008941562694249048,
+      "loss": 2.9593,
+      "step": 2911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008940759242007885,
+      "loss": 2.8827,
+      "step": 2912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008939955521061448,
+      "loss": 3.1186,
+      "step": 2913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0008939151531464538,
+      "loss": 3.0052,
+      "step": 2914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008938347273271976,
+      "loss": 3.0655,
+      "step": 2915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008937542746538604,
+      "loss": 2.8686,
+      "step": 2916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008936737951319276,
+      "loss": 3.2694,
+      "step": 2917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000893593288766887,
+      "loss": 2.9958,
+      "step": 2918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000893512755564228,
+      "loss": 2.9454,
+      "step": 2919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008934321955294418,
+      "loss": 2.9534,
+      "step": 2920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008933516086680215,
+      "loss": 3.36,
+      "step": 2921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008932709949854619,
+      "loss": 3.3553,
+      "step": 2922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008931903544872599,
+      "loss": 2.9973,
+      "step": 2923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008931096871789141,
+      "loss": 3.2614,
+      "step": 2924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008930289930659248,
+      "loss": 3.2586,
+      "step": 2925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008929482721537943,
+      "loss": 3.1402,
+      "step": 2926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008928675244480264,
+      "loss": 3.0883,
+      "step": 2927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008927867499541272,
+      "loss": 3.2601,
+      "step": 2928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008927059486776044,
+      "loss": 3.0622,
+      "step": 2929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008926251206239675,
+      "loss": 3.1737,
+      "step": 2930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008925442657987279,
+      "loss": 3.0665,
+      "step": 2931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008924633842073986,
+      "loss": 3.0239,
+      "step": 2932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008923824758554948,
+      "loss": 3.0019,
+      "step": 2933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008923015407485333,
+      "loss": 3.3198,
+      "step": 2934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008922205788920326,
+      "loss": 3.2599,
+      "step": 2935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008921395902915135,
+      "loss": 3.2923,
+      "step": 2936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008920585749524979,
+      "loss": 3.3228,
+      "step": 2937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008919775328805102,
+      "loss": 2.9657,
+      "step": 2938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008918964640810763,
+      "loss": 3.009,
+      "step": 2939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008918153685597239,
+      "loss": 3.2578,
+      "step": 2940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008917342463219826,
+      "loss": 3.0882,
+      "step": 2941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008916530973733839,
+      "loss": 3.1818,
+      "step": 2942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008915719217194611,
+      "loss": 2.8467,
+      "step": 2943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008914907193657489,
+      "loss": 3.195,
+      "step": 2944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008914094903177847,
+      "loss": 3.1045,
+      "step": 2945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008913282345811066,
+      "loss": 3.0188,
+      "step": 2946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008912469521612557,
+      "loss": 3.0129,
+      "step": 2947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000891165643063774,
+      "loss": 3.1791,
+      "step": 2948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008910843072942057,
+      "loss": 3.2754,
+      "step": 2949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008910029448580968,
+      "loss": 3.3069,
+      "step": 2950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008909215557609952,
+      "loss": 3.2824,
+      "step": 2951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008908401400084503,
+      "loss": 3.0092,
+      "step": 2952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008907586976060137,
+      "loss": 3.083,
+      "step": 2953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008906772285592386,
+      "loss": 3.0506,
+      "step": 2954
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00089059573287368,
+      "loss": 3.0996,
+      "step": 2955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008905142105548948,
+      "loss": 3.0766,
+      "step": 2956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008904326616084417,
+      "loss": 2.7787,
+      "step": 2957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008903510860398813,
+      "loss": 3.1037,
+      "step": 2958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008902694838547758,
+      "loss": 3.0831,
+      "step": 2959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008901878550586894,
+      "loss": 2.902,
+      "step": 2960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000890106199657188,
+      "loss": 3.1452,
+      "step": 2961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008900245176558394,
+      "loss": 3.4081,
+      "step": 2962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008899428090602133,
+      "loss": 2.8637,
+      "step": 2963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000889861073875881,
+      "loss": 3.1344,
+      "step": 2964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008897793121084157,
+      "loss": 3.204,
+      "step": 2965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008896975237633924,
+      "loss": 2.8405,
+      "step": 2966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008896157088463879,
+      "loss": 3.1072,
+      "step": 2967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008895338673629809,
+      "loss": 3.1385,
+      "step": 2968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000889451999318752,
+      "loss": 3.0033,
+      "step": 2969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008893701047192833,
+      "loss": 3.0674,
+      "step": 2970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008892881835701589,
+      "loss": 3.083,
+      "step": 2971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008892062358769646,
+      "loss": 3.3045,
+      "step": 2972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008891242616452884,
+      "loss": 2.9887,
+      "step": 2973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008890422608807194,
+      "loss": 2.9608,
+      "step": 2974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008889602335888492,
+      "loss": 3.3166,
+      "step": 2975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008888781797752708,
+      "loss": 3.1167,
+      "step": 2976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008887960994455792,
+      "loss": 3.0023,
+      "step": 2977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008887139926053711,
+      "loss": 3.183,
+      "step": 2978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008886318592602453,
+      "loss": 3.1622,
+      "step": 2979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008885496994158017,
+      "loss": 3.1529,
+      "step": 2980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008884675130776429,
+      "loss": 3.0416,
+      "step": 2981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008883853002513725,
+      "loss": 3.1358,
+      "step": 2982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008883030609425966,
+      "loss": 3.087,
+      "step": 2983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008882207951569225,
+      "loss": 3.1339,
+      "step": 2984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008881385028999598,
+      "loss": 3.3613,
+      "step": 2985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008880561841773196,
+      "loss": 2.8803,
+      "step": 2986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008879738389946149,
+      "loss": 3.1218,
+      "step": 2987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008878914673574606,
+      "loss": 3.0835,
+      "step": 2988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008878090692714731,
+      "loss": 3.0655,
+      "step": 2989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000887726644742271,
+      "loss": 3.0805,
+      "step": 2990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008876441937754745,
+      "loss": 3.0747,
+      "step": 2991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008875617163767056,
+      "loss": 2.9782,
+      "step": 2992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000887479212551588,
+      "loss": 3.1175,
+      "step": 2993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008873966823057473,
+      "loss": 3.191,
+      "step": 2994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008873141256448112,
+      "loss": 3.1716,
+      "step": 2995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008872315425744087,
+      "loss": 3.1681,
+      "step": 2996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008871489331001708,
+      "loss": 3.218,
+      "step": 2997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008870662972277305,
+      "loss": 3.031,
+      "step": 2998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008869836349627223,
+      "loss": 3.0927,
+      "step": 2999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008869009463107827,
+      "loss": 3.0991,
+      "step": 3000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008868182312775497,
+      "loss": 3.2858,
+      "step": 3001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008867354898686635,
+      "loss": 3.0731,
+      "step": 3002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008866527220897659,
+      "loss": 3.1681,
+      "step": 3003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008865699279465007,
+      "loss": 2.8693,
+      "step": 3004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008864871074445129,
+      "loss": 3.0692,
+      "step": 3005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008864042605894501,
+      "loss": 3.008,
+      "step": 3006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008863213873869611,
+      "loss": 3.2316,
+      "step": 3007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008862384878426968,
+      "loss": 3.2006,
+      "step": 3008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008861555619623098,
+      "loss": 3.1613,
+      "step": 3009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008860726097514544,
+      "loss": 3.1631,
+      "step": 3010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008859896312157869,
+      "loss": 3.1685,
+      "step": 3011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008859066263609652,
+      "loss": 3.0797,
+      "step": 3012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008858235951926492,
+      "loss": 3.114,
+      "step": 3013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008857405377165003,
+      "loss": 3.2269,
+      "step": 3014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008856574539381821,
+      "loss": 3.0623,
+      "step": 3015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008855743438633594,
+      "loss": 3.153,
+      "step": 3016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008854912074976998,
+      "loss": 3.1313,
+      "step": 3017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008854080448468713,
+      "loss": 3.1416,
+      "step": 3018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008853248559165449,
+      "loss": 3.1362,
+      "step": 3019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008852416407123929,
+      "loss": 3.0789,
+      "step": 3020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008851583992400894,
+      "loss": 2.9544,
+      "step": 3021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008850751315053101,
+      "loss": 3.1684,
+      "step": 3022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008849918375137331,
+      "loss": 3.1236,
+      "step": 3023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008849085172710374,
+      "loss": 3.0443,
+      "step": 3024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008848251707829047,
+      "loss": 2.9498,
+      "step": 3025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008847417980550181,
+      "loss": 3.1607,
+      "step": 3026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008846583990930621,
+      "loss": 3.2983,
+      "step": 3027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008845749739027236,
+      "loss": 3.3101,
+      "step": 3028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000884491522489691,
+      "loss": 2.8666,
+      "step": 3029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008844080448596546,
+      "loss": 2.9151,
+      "step": 3030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008843245410183062,
+      "loss": 3.0621,
+      "step": 3031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008842410109713401,
+      "loss": 3.1308,
+      "step": 3032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008841574547244513,
+      "loss": 3.1879,
+      "step": 3033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008840738722833374,
+      "loss": 2.7873,
+      "step": 3034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008839902636536978,
+      "loss": 3.0224,
+      "step": 3035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008839066288412332,
+      "loss": 3.2976,
+      "step": 3036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008838229678516465,
+      "loss": 3.5348,
+      "step": 3037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0008837392806906419,
+      "loss": 3.21,
+      "step": 3038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000883655567363926,
+      "loss": 2.9974,
+      "step": 3039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008835718278772069,
+      "loss": 3.0346,
+      "step": 3040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008834880622361944,
+      "loss": 3.2013,
+      "step": 3041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008834042704466001,
+      "loss": 3.2967,
+      "step": 3042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008833204525141374,
+      "loss": 3.0355,
+      "step": 3043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008832366084445218,
+      "loss": 3.1404,
+      "step": 3044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008831527382434699,
+      "loss": 2.9342,
+      "step": 3045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008830688419167009,
+      "loss": 3.0563,
+      "step": 3046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000882984919469935,
+      "loss": 3.0864,
+      "step": 3047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000882900970908895,
+      "loss": 3.1017,
+      "step": 3048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008828169962393046,
+      "loss": 3.0575,
+      "step": 3049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008827329954668899,
+      "loss": 3.0174,
+      "step": 3050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008826489685973784,
+      "loss": 3.3227,
+      "step": 3051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008825649156365,
+      "loss": 3.14,
+      "step": 3052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008824808365899856,
+      "loss": 3.0795,
+      "step": 3053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008823967314635681,
+      "loss": 3.2836,
+      "step": 3054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008823126002629829,
+      "loss": 3.1274,
+      "step": 3055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000882228442993966,
+      "loss": 3.0267,
+      "step": 3056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008821442596622557,
+      "loss": 3.1232,
+      "step": 3057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008820600502735927,
+      "loss": 3.066,
+      "step": 3058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008819758148337187,
+      "loss": 2.9973,
+      "step": 3059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008818915533483771,
+      "loss": 3.1015,
+      "step": 3060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008818072658233136,
+      "loss": 3.2211,
+      "step": 3061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008817229522642756,
+      "loss": 3.0697,
+      "step": 3062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008816386126770117,
+      "loss": 3.0218,
+      "step": 3063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008815542470672731,
+      "loss": 3.0693,
+      "step": 3064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000881469855440812,
+      "loss": 3.1792,
+      "step": 3065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008813854378033832,
+      "loss": 3.2799,
+      "step": 3066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008813009941607423,
+      "loss": 3.3275,
+      "step": 3067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008812165245186478,
+      "loss": 3.2162,
+      "step": 3068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008811320288828587,
+      "loss": 3.258,
+      "step": 3069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008810475072591369,
+      "loss": 3.2509,
+      "step": 3070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008809629596532453,
+      "loss": 3.0319,
+      "step": 3071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008808783860709491,
+      "loss": 3.0744,
+      "step": 3072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000880793786518015,
+      "loss": 3.082,
+      "step": 3073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008807091610002115,
+      "loss": 3.0073,
+      "step": 3074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008806245095233087,
+      "loss": 3.2171,
+      "step": 3075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008805398320930791,
+      "loss": 3.0183,
+      "step": 3076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008804551287152962,
+      "loss": 3.1876,
+      "step": 3077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008803703993957355,
+      "loss": 3.13,
+      "step": 3078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008802856441401748,
+      "loss": 3.3502,
+      "step": 3079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008802008629543928,
+      "loss": 2.9539,
+      "step": 3080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008801160558441706,
+      "loss": 2.9855,
+      "step": 3081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008800312228152908,
+      "loss": 3.252,
+      "step": 3082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000879946363873538,
+      "loss": 3.0423,
+      "step": 3083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008798614790246982,
+      "loss": 3.238,
+      "step": 3084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008797765682745595,
+      "loss": 3.1608,
+      "step": 3085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008796916316289117,
+      "loss": 3.2352,
+      "step": 3086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000879606669093546,
+      "loss": 3.1806,
+      "step": 3087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008795216806742561,
+      "loss": 3.1467,
+      "step": 3088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008794366663768367,
+      "loss": 3.2224,
+      "step": 3089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008793516262070848,
+      "loss": 3.1924,
+      "step": 3090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008792665601707987,
+      "loss": 3.1323,
+      "step": 3091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008791814682737791,
+      "loss": 3.037,
+      "step": 3092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008790963505218278,
+      "loss": 3.1799,
+      "step": 3093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008790112069207487,
+      "loss": 2.9593,
+      "step": 3094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008789260374763475,
+      "loss": 3.3574,
+      "step": 3095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008788408421944316,
+      "loss": 3.2436,
+      "step": 3096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008787556210808101,
+      "loss": 3.1368,
+      "step": 3097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008786703741412938,
+      "loss": 3.052,
+      "step": 3098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008785851013816954,
+      "loss": 3.2211,
+      "step": 3099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008784998028078295,
+      "loss": 3.2585,
+      "step": 3100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008784144784255121,
+      "loss": 3.1874,
+      "step": 3101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008783291282405612,
+      "loss": 3.2544,
+      "step": 3102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008782437522587965,
+      "loss": 3.0833,
+      "step": 3103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008781583504860396,
+      "loss": 3.1065,
+      "step": 3104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008780729229281132,
+      "loss": 3.1256,
+      "step": 3105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000877987469590843,
+      "loss": 3.1781,
+      "step": 3106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008779019904800551,
+      "loss": 3.1992,
+      "step": 3107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008778164856015786,
+      "loss": 3.0764,
+      "step": 3108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008777309549612433,
+      "loss": 3.2297,
+      "step": 3109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008776453985648813,
+      "loss": 3.0494,
+      "step": 3110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008775598164183265,
+      "loss": 2.8773,
+      "step": 3111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008774742085274141,
+      "loss": 3.2228,
+      "step": 3112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008773885748979819,
+      "loss": 3.2141,
+      "step": 3113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008773029155358684,
+      "loss": 3.2244,
+      "step": 3114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008772172304469147,
+      "loss": 3.113,
+      "step": 3115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008771315196369631,
+      "loss": 3.0855,
+      "step": 3116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008770457831118584,
+      "loss": 3.1175,
+      "step": 3117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000876960020877446,
+      "loss": 3.1611,
+      "step": 3118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000876874232939574,
+      "loss": 3.0799,
+      "step": 3119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008767884193040921,
+      "loss": 3.0225,
+      "step": 3120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008767025799768511,
+      "loss": 3.2563,
+      "step": 3121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008766167149637047,
+      "loss": 3.0692,
+      "step": 3122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008765308242705072,
+      "loss": 3.1401,
+      "step": 3123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008764449079031156,
+      "loss": 2.9322,
+      "step": 3124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008763589658673877,
+      "loss": 3.3173,
+      "step": 3125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008762729981691839,
+      "loss": 2.9927,
+      "step": 3126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008761870048143661,
+      "loss": 3.1644,
+      "step": 3127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008761009858087974,
+      "loss": 3.1468,
+      "step": 3128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008760149411583436,
+      "loss": 3.0546,
+      "step": 3129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008759288708688715,
+      "loss": 2.9864,
+      "step": 3130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008758427749462499,
+      "loss": 3.1119,
+      "step": 3131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008757566533963495,
+      "loss": 3.1153,
+      "step": 3132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008756705062250423,
+      "loss": 3.2316,
+      "step": 3133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008755843334382027,
+      "loss": 3.0084,
+      "step": 3134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008754981350417064,
+      "loss": 3.0929,
+      "step": 3135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008754119110414308,
+      "loss": 2.9232,
+      "step": 3136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008753256614432552,
+      "loss": 2.9676,
+      "step": 3137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008752393862530608,
+      "loss": 3.1289,
+      "step": 3138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008751530854767302,
+      "loss": 3.0349,
+      "step": 3139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000875066759120148,
+      "loss": 3.1395,
+      "step": 3140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008749804071892005,
+      "loss": 2.9309,
+      "step": 3141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008748940296897756,
+      "loss": 3.2968,
+      "step": 3142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008748076266277631,
+      "loss": 3.1885,
+      "step": 3143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008747211980090544,
+      "loss": 3.2478,
+      "step": 3144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000874634743839543,
+      "loss": 3.1306,
+      "step": 3145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008745482641251236,
+      "loss": 3.0835,
+      "step": 3146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008744617588716932,
+      "loss": 2.9758,
+      "step": 3147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00087437522808515,
+      "loss": 3.1722,
+      "step": 3148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008742886717713942,
+      "loss": 3.209,
+      "step": 3149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000874202089936328,
+      "loss": 3.2237,
+      "step": 3150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008741154825858548,
+      "loss": 3.1175,
+      "step": 3151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008740288497258804,
+      "loss": 3.0136,
+      "step": 3152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008739421913623115,
+      "loss": 3.1275,
+      "step": 3153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008738555075010572,
+      "loss": 3.0525,
+      "step": 3154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008737687981480282,
+      "loss": 3.1495,
+      "step": 3155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008736820633091368,
+      "loss": 3.0678,
+      "step": 3156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008735953029902973,
+      "loss": 3.0448,
+      "step": 3157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008735085171974251,
+      "loss": 3.1478,
+      "step": 3158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008734217059364383,
+      "loss": 3.2121,
+      "step": 3159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000873334869213256,
+      "loss": 3.1762,
+      "step": 3160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008732480070337994,
+      "loss": 3.1276,
+      "step": 3161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.000873161119403991,
+      "loss": 3.0911,
+      "step": 3162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0008730742063297555,
+      "loss": 2.8608,
+      "step": 3163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008729872678170193,
+      "loss": 3.0815,
+      "step": 3164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008729003038717103,
+      "loss": 3.1886,
+      "step": 3165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008728133144997583,
+      "loss": 3.0718,
+      "step": 3166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008727262997070945,
+      "loss": 2.9869,
+      "step": 3167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008726392594996524,
+      "loss": 3.1648,
+      "step": 3168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008725521938833669,
+      "loss": 3.3562,
+      "step": 3169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008724651028641745,
+      "loss": 3.2157,
+      "step": 3170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008723779864480139,
+      "loss": 3.0206,
+      "step": 3171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000872290844640825,
+      "loss": 3.2869,
+      "step": 3172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008722036774485496,
+      "loss": 3.1174,
+      "step": 3173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008721164848771314,
+      "loss": 3.05,
+      "step": 3174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000872029266932516,
+      "loss": 2.8579,
+      "step": 3175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00087194202362065,
+      "loss": 2.9666,
+      "step": 3176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008718547549474825,
+      "loss": 3.004,
+      "step": 3177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008717674609189638,
+      "loss": 3.0781,
+      "step": 3178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008716801415410462,
+      "loss": 3.2884,
+      "step": 3179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008715927968196839,
+      "loss": 3.1761,
+      "step": 3180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008715054267608324,
+      "loss": 3.2851,
+      "step": 3181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000871418031370449,
+      "loss": 3.1024,
+      "step": 3182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008713306106544931,
+      "loss": 2.9815,
+      "step": 3183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008712431646189256,
+      "loss": 3.2126,
+      "step": 3184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008711556932697089,
+      "loss": 3.065,
+      "step": 3185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008710681966128075,
+      "loss": 3.0862,
+      "step": 3186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008709806746541873,
+      "loss": 3.168,
+      "step": 3187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008708931273998164,
+      "loss": 3.0985,
+      "step": 3188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008708055548556639,
+      "loss": 3.2444,
+      "step": 3189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008707179570277012,
+      "loss": 2.9873,
+      "step": 3190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008706303339219015,
+      "loss": 3.2136,
+      "step": 3191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008705426855442393,
+      "loss": 3.3012,
+      "step": 3192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008704550119006908,
+      "loss": 3.1986,
+      "step": 3193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008703673129972345,
+      "loss": 2.8943,
+      "step": 3194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008702795888398499,
+      "loss": 3.0501,
+      "step": 3195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008701918394345188,
+      "loss": 2.9192,
+      "step": 3196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008701040647872244,
+      "loss": 2.8305,
+      "step": 3197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000870016264903952,
+      "loss": 3.1053,
+      "step": 3198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008699284397906879,
+      "loss": 3.0452,
+      "step": 3199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000869840589453421,
+      "loss": 3.3174,
+      "step": 3200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008697527138981409,
+      "loss": 3.3322,
+      "step": 3201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00086966481313084,
+      "loss": 3.1401,
+      "step": 3202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008695768871575117,
+      "loss": 3.2756,
+      "step": 3203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008694889359841514,
+      "loss": 3.2205,
+      "step": 3204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008694009596167562,
+      "loss": 3.0175,
+      "step": 3205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008693129580613246,
+      "loss": 3.2761,
+      "step": 3206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008692249313238576,
+      "loss": 3.1521,
+      "step": 3207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008691368794103569,
+      "loss": 3.2312,
+      "step": 3208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008690488023268266,
+      "loss": 3.2038,
+      "step": 3209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008689607000792725,
+      "loss": 3.046,
+      "step": 3210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008688725726737018,
+      "loss": 3.0384,
+      "step": 3211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008687844201161235,
+      "loss": 3.2055,
+      "step": 3212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008686962424125485,
+      "loss": 2.9457,
+      "step": 3213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008686080395689893,
+      "loss": 3.0764,
+      "step": 3214
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00086851981159146,
+      "loss": 3.0903,
+      "step": 3215
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008684315584859768,
+      "loss": 3.0442,
+      "step": 3216
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008683432802585572,
+      "loss": 3.1515,
+      "step": 3217
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008682549769152203,
+      "loss": 3.1065,
+      "step": 3218
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008681666484619874,
+      "loss": 2.9897,
+      "step": 3219
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008680782949048814,
+      "loss": 3.1724,
+      "step": 3220
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008679899162499266,
+      "loss": 2.9987,
+      "step": 3221
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008679015125031493,
+      "loss": 3.2566,
+      "step": 3222
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008678130836705772,
+      "loss": 3.0083,
+      "step": 3223
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008677246297582402,
+      "loss": 2.7997,
+      "step": 3224
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008676361507721695,
+      "loss": 3.021,
+      "step": 3225
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008675476467183981,
+      "loss": 2.8843,
+      "step": 3226
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008674591176029609,
+      "loss": 2.9375,
+      "step": 3227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008673705634318943,
+      "loss": 3.3234,
+      "step": 3228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008672819842112363,
+      "loss": 3.1237,
+      "step": 3229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008671933799470271,
+      "loss": 3.1174,
+      "step": 3230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000867104750645308,
+      "loss": 2.9786,
+      "step": 3231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008670160963121223,
+      "loss": 3.1414,
+      "step": 3232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008669274169535153,
+      "loss": 3.0719,
+      "step": 3233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008668387125755335,
+      "loss": 2.995,
+      "step": 3234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008667499831842251,
+      "loss": 3.2548,
+      "step": 3235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008666612287856408,
+      "loss": 2.9946,
+      "step": 3236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008665724493858318,
+      "loss": 3.0855,
+      "step": 3237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000866483644990852,
+      "loss": 2.8983,
+      "step": 3238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008663948156067563,
+      "loss": 2.9935,
+      "step": 3239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008663059612396021,
+      "loss": 2.924,
+      "step": 3240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008662170818954476,
+      "loss": 2.8567,
+      "step": 3241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008661281775803535,
+      "loss": 3.1063,
+      "step": 3242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008660392483003816,
+      "loss": 3.202,
+      "step": 3243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008659502940615958,
+      "loss": 2.9803,
+      "step": 3244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008658613148700613,
+      "loss": 3.1121,
+      "step": 3245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008657723107318456,
+      "loss": 3.0878,
+      "step": 3246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008656832816530172,
+      "loss": 2.9399,
+      "step": 3247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008655942276396471,
+      "loss": 3.2023,
+      "step": 3248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008655051486978069,
+      "loss": 3.1465,
+      "step": 3249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008654160448335712,
+      "loss": 3.1691,
+      "step": 3250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008653269160530154,
+      "loss": 3.2893,
+      "step": 3251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008652377623622167,
+      "loss": 3.3298,
+      "step": 3252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008651485837672543,
+      "loss": 3.0958,
+      "step": 3253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008650593802742091,
+      "loss": 3.2378,
+      "step": 3254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008649701518891632,
+      "loss": 3.2038,
+      "step": 3255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008648808986182008,
+      "loss": 3.0995,
+      "step": 3256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000864791620467408,
+      "loss": 2.9655,
+      "step": 3257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008647023174428723,
+      "loss": 3.1565,
+      "step": 3258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008646129895506827,
+      "loss": 3.1335,
+      "step": 3259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008645236367969303,
+      "loss": 3.003,
+      "step": 3260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008644342591877075,
+      "loss": 3.1206,
+      "step": 3261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000864344856729109,
+      "loss": 2.9777,
+      "step": 3262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008642554294272305,
+      "loss": 3.4362,
+      "step": 3263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008641659772881697,
+      "loss": 3.2379,
+      "step": 3264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008640765003180262,
+      "loss": 2.9817,
+      "step": 3265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008639869985229009,
+      "loss": 2.9738,
+      "step": 3266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008638974719088968,
+      "loss": 2.9365,
+      "step": 3267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008638079204821182,
+      "loss": 3.0138,
+      "step": 3268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008637183442486713,
+      "loss": 3.0513,
+      "step": 3269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000863628743214664,
+      "loss": 3.1059,
+      "step": 3270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008635391173862059,
+      "loss": 3.0995,
+      "step": 3271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000863449466769408,
+      "loss": 3.2064,
+      "step": 3272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008633597913703836,
+      "loss": 3.0375,
+      "step": 3273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008632700911952471,
+      "loss": 3.0067,
+      "step": 3274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008631803662501147,
+      "loss": 3.4056,
+      "step": 3275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008630906165411046,
+      "loss": 3.1526,
+      "step": 3276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008630008420743365,
+      "loss": 2.987,
+      "step": 3277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008629110428559315,
+      "loss": 2.9395,
+      "step": 3278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008628212188920131,
+      "loss": 3.3496,
+      "step": 3279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008627313701887056,
+      "loss": 3.0452,
+      "step": 3280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008626414967521359,
+      "loss": 3.1838,
+      "step": 3281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008625515985884318,
+      "loss": 2.9234,
+      "step": 3282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008624616757037231,
+      "loss": 3.0374,
+      "step": 3283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008623717281041416,
+      "loss": 2.9636,
+      "step": 3284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008622817557958201,
+      "loss": 3.1054,
+      "step": 3285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008621917587848937,
+      "loss": 2.9475,
+      "step": 3286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0008621017370774989,
+      "loss": 3.0113,
+      "step": 3287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008620116906797739,
+      "loss": 3.3128,
+      "step": 3288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008619216195978587,
+      "loss": 2.9689,
+      "step": 3289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008618315238378949,
+      "loss": 2.8392,
+      "step": 3290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008617414034060256,
+      "loss": 3.3131,
+      "step": 3291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008616512583083959,
+      "loss": 2.8163,
+      "step": 3292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008615610885511525,
+      "loss": 2.9811,
+      "step": 3293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008614708941404437,
+      "loss": 3.1633,
+      "step": 3294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008613806750824194,
+      "loss": 3.1589,
+      "step": 3295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008612904313832315,
+      "loss": 3.1412,
+      "step": 3296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008612001630490332,
+      "loss": 3.3601,
+      "step": 3297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008611098700859796,
+      "loss": 3.0015,
+      "step": 3298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008610195525002275,
+      "loss": 3.128,
+      "step": 3299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008609292102979353,
+      "loss": 3.1376,
+      "step": 3300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000860838843485263,
+      "loss": 3.0701,
+      "step": 3301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008607484520683721,
+      "loss": 3.1714,
+      "step": 3302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008606580360534268,
+      "loss": 3.05,
+      "step": 3303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008605675954465915,
+      "loss": 3.07,
+      "step": 3304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008604771302540334,
+      "loss": 3.3049,
+      "step": 3305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008603866404819207,
+      "loss": 3.0469,
+      "step": 3306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008602961261364239,
+      "loss": 3.0283,
+      "step": 3307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008602055872237145,
+      "loss": 3.1171,
+      "step": 3308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008601150237499662,
+      "loss": 2.9025,
+      "step": 3309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008600244357213539,
+      "loss": 3.1631,
+      "step": 3310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008599338231440548,
+      "loss": 3.3124,
+      "step": 3311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008598431860242472,
+      "loss": 2.9903,
+      "step": 3312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008597525243681115,
+      "loss": 3.2168,
+      "step": 3313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008596618381818293,
+      "loss": 2.8143,
+      "step": 3314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008595711274715844,
+      "loss": 3.1095,
+      "step": 3315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008594803922435619,
+      "loss": 3.2628,
+      "step": 3316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008593896325039487,
+      "loss": 3.2295,
+      "step": 3317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008592988482589332,
+      "loss": 3.3919,
+      "step": 3318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008592080395147059,
+      "loss": 3.1176,
+      "step": 3319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008591172062774587,
+      "loss": 3.0268,
+      "step": 3320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008590263485533849,
+      "loss": 3.0781,
+      "step": 3321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00085893546634868,
+      "loss": 3.0884,
+      "step": 3322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008588445596695406,
+      "loss": 2.9437,
+      "step": 3323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008587536285221655,
+      "loss": 3.0975,
+      "step": 3324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008586626729127551,
+      "loss": 3.1345,
+      "step": 3325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008585716928475112,
+      "loss": 3.0531,
+      "step": 3326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000858480688332637,
+      "loss": 3.1057,
+      "step": 3327
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008583896593743383,
+      "loss": 3.0588,
+      "step": 3328
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008582986059788217,
+      "loss": 2.9175,
+      "step": 3329
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008582075281522958,
+      "loss": 3.0162,
+      "step": 3330
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008581164259009711,
+      "loss": 3.0842,
+      "step": 3331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008580252992310593,
+      "loss": 3.0936,
+      "step": 3332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008579341481487739,
+      "loss": 3.292,
+      "step": 3333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008578429726603303,
+      "loss": 2.9745,
+      "step": 3334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008577517727719455,
+      "loss": 3.0091,
+      "step": 3335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008576605484898376,
+      "loss": 3.0988,
+      "step": 3336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008575692998202274,
+      "loss": 3.2114,
+      "step": 3337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008574780267693366,
+      "loss": 3.0996,
+      "step": 3338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008573867293433885,
+      "loss": 3.1837,
+      "step": 3339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008572954075486088,
+      "loss": 3.3928,
+      "step": 3340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000857204061391224,
+      "loss": 2.9842,
+      "step": 3341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008571126908774628,
+      "loss": 3.2271,
+      "step": 3342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008570212960135554,
+      "loss": 3.1606,
+      "step": 3343
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008569298768057336,
+      "loss": 3.0617,
+      "step": 3344
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008568384332602309,
+      "loss": 2.9411,
+      "step": 3345
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008567469653832827,
+      "loss": 3.2336,
+      "step": 3346
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008566554731811256,
+      "loss": 2.9593,
+      "step": 3347
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008565639566599983,
+      "loss": 3.2726,
+      "step": 3348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008564724158261407,
+      "loss": 3.1183,
+      "step": 3349
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008563808506857948,
+      "loss": 3.0389,
+      "step": 3350
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008562892612452042,
+      "loss": 2.9303,
+      "step": 3351
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008561976475106137,
+      "loss": 3.1202,
+      "step": 3352
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008561060094882704,
+      "loss": 3.0615,
+      "step": 3353
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008560143471844226,
+      "loss": 3.1235,
+      "step": 3354
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008559226606053204,
+      "loss": 3.1715,
+      "step": 3355
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008558309497572156,
+      "loss": 3.0752,
+      "step": 3356
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008557392146463617,
+      "loss": 2.9929,
+      "step": 3357
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008556474552790135,
+      "loss": 3.2041,
+      "step": 3358
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008555556716614279,
+      "loss": 2.9231,
+      "step": 3359
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008554638637998633,
+      "loss": 3.1884,
+      "step": 3360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008553720317005796,
+      "loss": 3.0322,
+      "step": 3361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008552801753698387,
+      "loss": 2.8677,
+      "step": 3362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008551882948139035,
+      "loss": 3.1402,
+      "step": 3363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008550963900390397,
+      "loss": 3.0957,
+      "step": 3364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008550044610515133,
+      "loss": 3.0563,
+      "step": 3365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008549125078575927,
+      "loss": 3.1659,
+      "step": 3366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008548205304635479,
+      "loss": 2.9273,
+      "step": 3367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008547285288756508,
+      "loss": 3.1425,
+      "step": 3368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000854636503100174,
+      "loss": 3.2335,
+      "step": 3369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008545444531433929,
+      "loss": 2.9326,
+      "step": 3370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000854452379011584,
+      "loss": 3.2936,
+      "step": 3371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008543602807110251,
+      "loss": 3.3695,
+      "step": 3372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008542681582479965,
+      "loss": 3.0436,
+      "step": 3373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008541760116287792,
+      "loss": 2.9318,
+      "step": 3374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008540838408596567,
+      "loss": 3.1351,
+      "step": 3375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008539916459469137,
+      "loss": 2.8687,
+      "step": 3376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008538994268968367,
+      "loss": 3.2854,
+      "step": 3377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008538071837157134,
+      "loss": 2.9603,
+      "step": 3378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008537149164098339,
+      "loss": 3.0781,
+      "step": 3379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008536226249854893,
+      "loss": 3.2192,
+      "step": 3380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008535303094489728,
+      "loss": 2.8423,
+      "step": 3381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008534379698065789,
+      "loss": 3.1072,
+      "step": 3382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008533456060646039,
+      "loss": 2.9625,
+      "step": 3383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008532532182293457,
+      "loss": 3.1399,
+      "step": 3384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008531608063071039,
+      "loss": 3.0653,
+      "step": 3385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008530683703041799,
+      "loss": 3.3069,
+      "step": 3386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008529759102268764,
+      "loss": 3.0606,
+      "step": 3387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008528834260814978,
+      "loss": 2.9593,
+      "step": 3388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008527909178743504,
+      "loss": 3.1052,
+      "step": 3389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008526983856117418,
+      "loss": 2.9288,
+      "step": 3390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008526058292999816,
+      "loss": 3.193,
+      "step": 3391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000852513248945381,
+      "loss": 3.0596,
+      "step": 3392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008524206445542522,
+      "loss": 3.0502,
+      "step": 3393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00085232801613291,
+      "loss": 3.0821,
+      "step": 3394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008522353636876703,
+      "loss": 2.9905,
+      "step": 3395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008521426872248504,
+      "loss": 3.1863,
+      "step": 3396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00085204998675077,
+      "loss": 3.285,
+      "step": 3397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008519572622717496,
+      "loss": 3.144,
+      "step": 3398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.000851864513794112,
+      "loss": 3.214,
+      "step": 3399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008517717413241813,
+      "loss": 3.0974,
+      "step": 3400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008516789448682831,
+      "loss": 3.2269,
+      "step": 3401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008515861244327452,
+      "loss": 3.117,
+      "step": 3402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008514932800238964,
+      "loss": 3.1357,
+      "step": 3403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008514004116480674,
+      "loss": 3.1118,
+      "step": 3404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008513075193115906,
+      "loss": 3.0845,
+      "step": 3405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008512146030208002,
+      "loss": 3.1574,
+      "step": 3406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008511216627820313,
+      "loss": 3.0129,
+      "step": 3407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008510286986016215,
+      "loss": 3.2126,
+      "step": 3408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008509357104859098,
+      "loss": 3.2554,
+      "step": 3409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008508426984412364,
+      "loss": 3.1761,
+      "step": 3410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.0008507496624739435,
+      "loss": 3.0433,
+      "step": 3411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008506566025903748,
+      "loss": 3.0124,
+      "step": 3412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008505635187968759,
+      "loss": 3.0743,
+      "step": 3413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008504704110997938,
+      "loss": 2.9223,
+      "step": 3414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008503772795054771,
+      "loss": 3.0884,
+      "step": 3415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008502841240202759,
+      "loss": 2.9493,
+      "step": 3416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008501909446505425,
+      "loss": 2.9928,
+      "step": 3417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008500977414026301,
+      "loss": 3.2787,
+      "step": 3418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008500045142828941,
+      "loss": 3.2457,
+      "step": 3419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008499112632976911,
+      "loss": 3.213,
+      "step": 3420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008498179884533797,
+      "loss": 3.2108,
+      "step": 3421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008497246897563198,
+      "loss": 2.9728,
+      "step": 3422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008496313672128732,
+      "loss": 3.1907,
+      "step": 3423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008495380208294032,
+      "loss": 3.1345,
+      "step": 3424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008494446506122746,
+      "loss": 3.2184,
+      "step": 3425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008493512565678542,
+      "loss": 3.0743,
+      "step": 3426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008492578387025101,
+      "loss": 3.0973,
+      "step": 3427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000849164397022612,
+      "loss": 3.1461,
+      "step": 3428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008490709315345313,
+      "loss": 3.1634,
+      "step": 3429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008489774422446411,
+      "loss": 3.2203,
+      "step": 3430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008488839291593163,
+      "loss": 2.9784,
+      "step": 3431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008487903922849328,
+      "loss": 3.1033,
+      "step": 3432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008486968316278689,
+      "loss": 2.9704,
+      "step": 3433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008486032471945039,
+      "loss": 3.1234,
+      "step": 3434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000848509638991219,
+      "loss": 2.8412,
+      "step": 3435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008484160070243972,
+      "loss": 3.3678,
+      "step": 3436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008483223513004225,
+      "loss": 2.9128,
+      "step": 3437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008482286718256813,
+      "loss": 3.3005,
+      "step": 3438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008481349686065611,
+      "loss": 3.093,
+      "step": 3439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008480412416494512,
+      "loss": 3.0946,
+      "step": 3440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008479474909607423,
+      "loss": 3.1623,
+      "step": 3441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008478537165468271,
+      "loss": 2.8897,
+      "step": 3442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008477599184140996,
+      "loss": 2.9188,
+      "step": 3443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008476660965689558,
+      "loss": 2.9846,
+      "step": 3444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008475722510177926,
+      "loss": 3.1226,
+      "step": 3445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008474783817670093,
+      "loss": 3.0523,
+      "step": 3446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008473844888230064,
+      "loss": 2.918,
+      "step": 3447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008472905721921862,
+      "loss": 2.9599,
+      "step": 3448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008471966318809524,
+      "loss": 3.0982,
+      "step": 3449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008471026678957105,
+      "loss": 3.1241,
+      "step": 3450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008470086802428674,
+      "loss": 3.0527,
+      "step": 3451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008469146689288319,
+      "loss": 3.201,
+      "step": 3452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008468206339600141,
+      "loss": 3.0883,
+      "step": 3453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000846726575342826,
+      "loss": 2.9515,
+      "step": 3454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008466324930836812,
+      "loss": 2.8478,
+      "step": 3455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008465383871889948,
+      "loss": 3.1591,
+      "step": 3456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008464442576651833,
+      "loss": 3.1518,
+      "step": 3457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008463501045186653,
+      "loss": 3.135,
+      "step": 3458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008462559277558604,
+      "loss": 3.0727,
+      "step": 3459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008461617273831906,
+      "loss": 3.1544,
+      "step": 3460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008460675034070785,
+      "loss": 3.0654,
+      "step": 3461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008459732558339495,
+      "loss": 3.2039,
+      "step": 3462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008458789846702295,
+      "loss": 3.0529,
+      "step": 3463
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008457846899223467,
+      "loss": 3.2071,
+      "step": 3464
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008456903715967306,
+      "loss": 3.15,
+      "step": 3465
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008455960296998126,
+      "loss": 3.0103,
+      "step": 3466
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008455016642380253,
+      "loss": 3.1139,
+      "step": 3467
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008454072752178033,
+      "loss": 3.1298,
+      "step": 3468
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008453128626455823,
+      "loss": 3.0475,
+      "step": 3469
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008452184265278004,
+      "loss": 3.2792,
+      "step": 3470
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008451239668708965,
+      "loss": 3.2148,
+      "step": 3471
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008450294836813115,
+      "loss": 3.2531,
+      "step": 3472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000844934976965488,
+      "loss": 2.9355,
+      "step": 3473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008448404467298698,
+      "loss": 2.9635,
+      "step": 3474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008447458929809027,
+      "loss": 2.8271,
+      "step": 3475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000844651315725034,
+      "loss": 3.0309,
+      "step": 3476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008445567149687126,
+      "loss": 2.9175,
+      "step": 3477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008444620907183887,
+      "loss": 3.1037,
+      "step": 3478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008443674429805145,
+      "loss": 3.1014,
+      "step": 3479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008442727717615437,
+      "loss": 3.1498,
+      "step": 3480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008441780770679317,
+      "loss": 3.2068,
+      "step": 3481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008440833589061352,
+      "loss": 3.1032,
+      "step": 3482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008439886172826128,
+      "loss": 3.1778,
+      "step": 3483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008438938522038243,
+      "loss": 3.4083,
+      "step": 3484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008437990636762316,
+      "loss": 3.1565,
+      "step": 3485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000843704251706298,
+      "loss": 3.2522,
+      "step": 3486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008436094163004882,
+      "loss": 2.9164,
+      "step": 3487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008435145574652686,
+      "loss": 3.1139,
+      "step": 3488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008434196752071075,
+      "loss": 3.1301,
+      "step": 3489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008433247695324746,
+      "loss": 2.8451,
+      "step": 3490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008432298404478408,
+      "loss": 3.1048,
+      "step": 3491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008431348879596791,
+      "loss": 3.304,
+      "step": 3492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008430399120744643,
+      "loss": 2.9339,
+      "step": 3493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008429449127986719,
+      "loss": 3.2113,
+      "step": 3494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00084284989013878,
+      "loss": 3.1538,
+      "step": 3495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008427548441012674,
+      "loss": 3.1148,
+      "step": 3496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008426597746926152,
+      "loss": 3.1064,
+      "step": 3497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008425646819193059,
+      "loss": 3.1302,
+      "step": 3498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008424695657878232,
+      "loss": 2.9607,
+      "step": 3499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008423744263046529,
+      "loss": 2.964,
+      "step": 3500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008422792634762824,
+      "loss": 2.9758,
+      "step": 3501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008421840773092,
+      "loss": 3.3513,
+      "step": 3502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008420888678098964,
+      "loss": 2.943,
+      "step": 3503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008419936349848636,
+      "loss": 2.8941,
+      "step": 3504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000841898378840595,
+      "loss": 2.9621,
+      "step": 3505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008418030993835857,
+      "loss": 3.1509,
+      "step": 3506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008417077966203328,
+      "loss": 3.225,
+      "step": 3507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008416124705573342,
+      "loss": 3.3333,
+      "step": 3508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008415171212010901,
+      "loss": 3.0826,
+      "step": 3509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.000841421748558102,
+      "loss": 3.0964,
+      "step": 3510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008413263526348728,
+      "loss": 3.1891,
+      "step": 3511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008412309334379074,
+      "loss": 3.1104,
+      "step": 3512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008411354909737119,
+      "loss": 3.0835,
+      "step": 3513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008410400252487942,
+      "loss": 3.0621,
+      "step": 3514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008409445362696637,
+      "loss": 3.002,
+      "step": 3515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008408490240428317,
+      "loss": 3.2897,
+      "step": 3516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008407534885748105,
+      "loss": 3.0471,
+      "step": 3517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008406579298721144,
+      "loss": 2.9055,
+      "step": 3518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008405623479412592,
+      "loss": 3.1134,
+      "step": 3519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008404667427887622,
+      "loss": 2.9753,
+      "step": 3520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008403711144211424,
+      "loss": 3.1257,
+      "step": 3521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008402754628449205,
+      "loss": 2.9484,
+      "step": 3522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008401797880666183,
+      "loss": 2.9873,
+      "step": 3523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008400840900927598,
+      "loss": 3.2027,
+      "step": 3524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00083998836892987,
+      "loss": 3.0944,
+      "step": 3525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008398926245844758,
+      "loss": 3.0269,
+      "step": 3526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008397968570631058,
+      "loss": 3.2798,
+      "step": 3527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008397010663722899,
+      "loss": 3.0597,
+      "step": 3528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008396052525185598,
+      "loss": 3.0638,
+      "step": 3529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008395094155084486,
+      "loss": 2.9048,
+      "step": 3530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008394135553484912,
+      "loss": 3.319,
+      "step": 3531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008393176720452238,
+      "loss": 2.8354,
+      "step": 3532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008392217656051842,
+      "loss": 3.2436,
+      "step": 3533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008391258360349121,
+      "loss": 2.8055,
+      "step": 3534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0008390298833409484,
+      "loss": 3.1204,
+      "step": 3535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008389339075298359,
+      "loss": 3.3077,
+      "step": 3536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008388379086081189,
+      "loss": 3.0766,
+      "step": 3537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000838741886582343,
+      "loss": 2.9562,
+      "step": 3538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008386458414590555,
+      "loss": 3.2047,
+      "step": 3539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008385497732448056,
+      "loss": 3.2515,
+      "step": 3540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008384536819461437,
+      "loss": 3.1384,
+      "step": 3541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000838357567569622,
+      "loss": 3.0429,
+      "step": 3542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000838261430121794,
+      "loss": 2.9679,
+      "step": 3543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008381652696092149,
+      "loss": 2.9819,
+      "step": 3544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008380690860384418,
+      "loss": 3.2728,
+      "step": 3545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008379728794160328,
+      "loss": 2.9471,
+      "step": 3546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008378766497485481,
+      "loss": 2.9003,
+      "step": 3547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837780397042549,
+      "loss": 3.0275,
+      "step": 3548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000837684121304599,
+      "loss": 2.9544,
+      "step": 3549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008375878225412623,
+      "loss": 2.9775,
+      "step": 3550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008374915007591053,
+      "loss": 3.1217,
+      "step": 3551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008373951559646959,
+      "loss": 3.1611,
+      "step": 3552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008372987881646035,
+      "loss": 3.0836,
+      "step": 3553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008372023973653989,
+      "loss": 3.1096,
+      "step": 3554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008371059835736548,
+      "loss": 3.1404,
+      "step": 3555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008370095467959451,
+      "loss": 3.2313,
+      "step": 3556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008369130870388458,
+      "loss": 3.1095,
+      "step": 3557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008368166043089336,
+      "loss": 2.9336,
+      "step": 3558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008367200986127876,
+      "loss": 3.0946,
+      "step": 3559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008366235699569882,
+      "loss": 2.9741,
+      "step": 3560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008365270183481172,
+      "loss": 3.0519,
+      "step": 3561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008364304437927582,
+      "loss": 2.92,
+      "step": 3562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008363338462974958,
+      "loss": 3.0974,
+      "step": 3563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008362372258689175,
+      "loss": 3.0981,
+      "step": 3564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008361405825136106,
+      "loss": 3.2421,
+      "step": 3565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008360439162381654,
+      "loss": 3.1417,
+      "step": 3566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008359472270491728,
+      "loss": 3.043,
+      "step": 3567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008358505149532261,
+      "loss": 3.2181,
+      "step": 3568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008357537799569192,
+      "loss": 2.9844,
+      "step": 3569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008356570220668486,
+      "loss": 3.171,
+      "step": 3570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008355602412896117,
+      "loss": 2.81,
+      "step": 3571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008354634376318075,
+      "loss": 2.9543,
+      "step": 3572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008353666111000366,
+      "loss": 3.1999,
+      "step": 3573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008352697617009013,
+      "loss": 3.1091,
+      "step": 3574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008351728894410056,
+      "loss": 3.1694,
+      "step": 3575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008350759943269546,
+      "loss": 3.0236,
+      "step": 3576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008349790763653554,
+      "loss": 3.1999,
+      "step": 3577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000834882135562816,
+      "loss": 3.1601,
+      "step": 3578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008347851719259471,
+      "loss": 3.1918,
+      "step": 3579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008346881854613599,
+      "loss": 3.0433,
+      "step": 3580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008345911761756675,
+      "loss": 3.2057,
+      "step": 3581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008344941440754847,
+      "loss": 3.2022,
+      "step": 3582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008343970891674277,
+      "loss": 3.0054,
+      "step": 3583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008343000114581145,
+      "loss": 3.2814,
+      "step": 3584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008342029109541641,
+      "loss": 3.1328,
+      "step": 3585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008341057876621976,
+      "loss": 3.3463,
+      "step": 3586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008340086415888376,
+      "loss": 3.0426,
+      "step": 3587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000833911472740708,
+      "loss": 2.8472,
+      "step": 3588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008338142811244343,
+      "loss": 2.9624,
+      "step": 3589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008337170667466435,
+      "loss": 3.0474,
+      "step": 3590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008336198296139645,
+      "loss": 3.0503,
+      "step": 3591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008335225697330277,
+      "loss": 3.1039,
+      "step": 3592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008334252871104645,
+      "loss": 3.0416,
+      "step": 3593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008333279817529085,
+      "loss": 2.9818,
+      "step": 3594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008332306536669942,
+      "loss": 2.9344,
+      "step": 3595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008331333028593586,
+      "loss": 3.19,
+      "step": 3596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008330359293366392,
+      "loss": 3.0364,
+      "step": 3597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008329385331054758,
+      "loss": 3.0748,
+      "step": 3598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008328411141725094,
+      "loss": 3.2171,
+      "step": 3599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008327436725443824,
+      "loss": 3.0996,
+      "step": 3600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008326462082277393,
+      "loss": 3.1277,
+      "step": 3601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008325487212292258,
+      "loss": 3.0934,
+      "step": 3602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000832451211555489,
+      "loss": 3.1076,
+      "step": 3603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008323536792131778,
+      "loss": 2.8597,
+      "step": 3604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008322561242089424,
+      "loss": 3.2035,
+      "step": 3605
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008321585465494349,
+      "loss": 2.8628,
+      "step": 3606
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008320609462413086,
+      "loss": 2.9862,
+      "step": 3607
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008319633232912187,
+      "loss": 3.0663,
+      "step": 3608
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008318656777058215,
+      "loss": 3.1956,
+      "step": 3609
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008317680094917752,
+      "loss": 3.18,
+      "step": 3610
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008316703186557392,
+      "loss": 2.8794,
+      "step": 3611
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008315726052043753,
+      "loss": 3.2131,
+      "step": 3612
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008314748691443455,
+      "loss": 3.025,
+      "step": 3613
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008313771104823143,
+      "loss": 3.4408,
+      "step": 3614
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008312793292249475,
+      "loss": 3.3121,
+      "step": 3615
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008311815253789125,
+      "loss": 3.1804,
+      "step": 3616
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008310836989508781,
+      "loss": 3.2691,
+      "step": 3617
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008309858499475147,
+      "loss": 2.9268,
+      "step": 3618
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008308879783754942,
+      "loss": 3.0395,
+      "step": 3619
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008307900842414903,
+      "loss": 2.9492,
+      "step": 3620
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008306921675521778,
+      "loss": 3.0399,
+      "step": 3621
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008305942283142332,
+      "loss": 3.08,
+      "step": 3622
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000830496266534335,
+      "loss": 3.2189,
+      "step": 3623
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008303982822191625,
+      "loss": 3.1572,
+      "step": 3624
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008303002753753967,
+      "loss": 3.0575,
+      "step": 3625
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000830202246009721,
+      "loss": 3.102,
+      "step": 3626
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008301041941288188,
+      "loss": 3.0649,
+      "step": 3627
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008300061197393766,
+      "loss": 3.0242,
+      "step": 3628
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008299080228480814,
+      "loss": 3.0629,
+      "step": 3629
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008298099034616217,
+      "loss": 3.0329,
+      "step": 3630
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008297117615866885,
+      "loss": 3.2351,
+      "step": 3631
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008296135972299736,
+      "loss": 2.9364,
+      "step": 3632
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008295154103981702,
+      "loss": 3.1041,
+      "step": 3633
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008294172010979733,
+      "loss": 3.1286,
+      "step": 3634
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008293189693360796,
+      "loss": 3.4136,
+      "step": 3635
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008292207151191872,
+      "loss": 3.0591,
+      "step": 3636
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008291224384539953,
+      "loss": 3.1115,
+      "step": 3637
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008290241393472054,
+      "loss": 3.0748,
+      "step": 3638
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008289258178055199,
+      "loss": 3.1144,
+      "step": 3639
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008288274738356431,
+      "loss": 2.9512,
+      "step": 3640
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008287291074442806,
+      "loss": 3.216,
+      "step": 3641
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008286307186381399,
+      "loss": 3.242,
+      "step": 3642
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008285323074239295,
+      "loss": 3.1564,
+      "step": 3643
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008284338738083596,
+      "loss": 2.9127,
+      "step": 3644
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008283354177981421,
+      "loss": 3.0803,
+      "step": 3645
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008282369393999904,
+      "loss": 3.0675,
+      "step": 3646
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008281384386206194,
+      "loss": 2.9807,
+      "step": 3647
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008280399154667455,
+      "loss": 3.249,
+      "step": 3648
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008279413699450864,
+      "loss": 2.8039,
+      "step": 3649
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008278428020623618,
+      "loss": 3.1345,
+      "step": 3650
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008277442118252925,
+      "loss": 3.0719,
+      "step": 3651
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000827645599240601,
+      "loss": 3.0847,
+      "step": 3652
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008275469643150115,
+      "loss": 2.9612,
+      "step": 3653
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008274483070552492,
+      "loss": 3.0707,
+      "step": 3654
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008273496274680415,
+      "loss": 3.0897,
+      "step": 3655
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008272509255601168,
+      "loss": 3.0966,
+      "step": 3656
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008271522013382054,
+      "loss": 3.0037,
+      "step": 3657
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008270534548090386,
+      "loss": 3.0463,
+      "step": 3658
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0008269546859793498,
+      "loss": 2.9404,
+      "step": 3659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008268558948558736,
+      "loss": 3.103,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000826757081445346,
+      "loss": 3.029,
+      "step": 3661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000826658245754505,
+      "loss": 3.0208,
+      "step": 3662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008265593877900897,
+      "loss": 3.228,
+      "step": 3663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000826460507558841,
+      "loss": 3.1242,
+      "step": 3664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008263616050675008,
+      "loss": 3.3307,
+      "step": 3665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008262626803228133,
+      "loss": 3.0581,
+      "step": 3666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008261637333315235,
+      "loss": 3.1353,
+      "step": 3667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008260647641003782,
+      "loss": 3.1709,
+      "step": 3668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008259657726361261,
+      "loss": 3.1119,
+      "step": 3669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008258667589455166,
+      "loss": 2.9866,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008257677230353014,
+      "loss": 3.0753,
+      "step": 3671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008256686649122331,
+      "loss": 3.2281,
+      "step": 3672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008255695845830664,
+      "loss": 3.0611,
+      "step": 3673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008254704820545571,
+      "loss": 3.3555,
+      "step": 3674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008253713573334624,
+      "loss": 3.2971,
+      "step": 3675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008252722104265415,
+      "loss": 2.8823,
+      "step": 3676
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008251730413405546,
+      "loss": 3.0137,
+      "step": 3677
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008250738500822639,
+      "loss": 3.2936,
+      "step": 3678
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008249746366584328,
+      "loss": 2.9868,
+      "step": 3679
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008248754010758263,
+      "loss": 3.322,
+      "step": 3680
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008247761433412109,
+      "loss": 3.201,
+      "step": 3681
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008246768634613544,
+      "loss": 3.1704,
+      "step": 3682
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008245775614430266,
+      "loss": 3.1981,
+      "step": 3683
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008244782372929983,
+      "loss": 3.2489,
+      "step": 3684
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008243788910180422,
+      "loss": 3.1453,
+      "step": 3685
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008242795226249322,
+      "loss": 3.1504,
+      "step": 3686
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000824180132120444,
+      "loss": 3.1895,
+      "step": 3687
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008240807195113546,
+      "loss": 3.1444,
+      "step": 3688
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008239812848044427,
+      "loss": 3.1707,
+      "step": 3689
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008238818280064881,
+      "loss": 3.107,
+      "step": 3690
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008237823491242724,
+      "loss": 2.9954,
+      "step": 3691
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008236828481645789,
+      "loss": 3.2777,
+      "step": 3692
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008235833251341919,
+      "loss": 3.0603,
+      "step": 3693
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008234837800398979,
+      "loss": 3.153,
+      "step": 3694
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008233842128884841,
+      "loss": 3.1765,
+      "step": 3695
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008232846236867399,
+      "loss": 3.1273,
+      "step": 3696
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008231850124414557,
+      "loss": 3.1371,
+      "step": 3697
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008230853791594237,
+      "loss": 2.7911,
+      "step": 3698
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008229857238474375,
+      "loss": 3.0684,
+      "step": 3699
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008228860465122923,
+      "loss": 3.128,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008227863471607846,
+      "loss": 3.1771,
+      "step": 3701
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008226866257997124,
+      "loss": 3.2488,
+      "step": 3702
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008225868824358755,
+      "loss": 3.1878,
+      "step": 3703
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008224871170760751,
+      "loss": 3.1763,
+      "step": 3704
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008223873297271136,
+      "loss": 2.9887,
+      "step": 3705
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008222875203957952,
+      "loss": 2.9108,
+      "step": 3706
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008221876890889256,
+      "loss": 3.1014,
+      "step": 3707
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008220878358133118,
+      "loss": 2.7824,
+      "step": 3708
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008219879605757624,
+      "loss": 3.0936,
+      "step": 3709
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008218880633830876,
+      "loss": 3.1295,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008217881442420991,
+      "loss": 3.0104,
+      "step": 3711
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008216882031596097,
+      "loss": 3.2407,
+      "step": 3712
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008215882401424343,
+      "loss": 2.8145,
+      "step": 3713
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008214882551973888,
+      "loss": 3.126,
+      "step": 3714
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008213882483312909,
+      "loss": 3.102,
+      "step": 3715
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008212882195509595,
+      "loss": 3.0772,
+      "step": 3716
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008211881688632154,
+      "loss": 2.9765,
+      "step": 3717
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008210880962748808,
+      "loss": 3.0091,
+      "step": 3718
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008209880017927789,
+      "loss": 3.3341,
+      "step": 3719
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008208878854237349,
+      "loss": 3.2045,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008207877471745755,
+      "loss": 2.9603,
+      "step": 3721
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008206875870521284,
+      "loss": 3.0109,
+      "step": 3722
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008205874050632236,
+      "loss": 3.2528,
+      "step": 3723
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008204872012146918,
+      "loss": 2.9235,
+      "step": 3724
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008203869755133655,
+      "loss": 3.0742,
+      "step": 3725
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008202867279660788,
+      "loss": 3.0067,
+      "step": 3726
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008201864585796674,
+      "loss": 3.1868,
+      "step": 3727
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008200861673609679,
+      "loss": 2.9255,
+      "step": 3728
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008199858543168191,
+      "loss": 3.103,
+      "step": 3729
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008198855194540606,
+      "loss": 3.2054,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008197851627795343,
+      "loss": 2.9913,
+      "step": 3731
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008196847843000828,
+      "loss": 2.9769,
+      "step": 3732
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008195843840225506,
+      "loss": 3.1295,
+      "step": 3733
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008194839619537837,
+      "loss": 3.0082,
+      "step": 3734
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008193835181006295,
+      "loss": 3.0512,
+      "step": 3735
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008192830524699366,
+      "loss": 3.0965,
+      "step": 3736
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008191825650685559,
+      "loss": 3.2025,
+      "step": 3737
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008190820559033388,
+      "loss": 3.0901,
+      "step": 3738
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008189815249811385,
+      "loss": 3.0134,
+      "step": 3739
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008188809723088104,
+      "loss": 3.0737,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008187803978932104,
+      "loss": 3.0241,
+      "step": 3741
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008186798017411965,
+      "loss": 2.8924,
+      "step": 3742
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008185791838596277,
+      "loss": 3.1963,
+      "step": 3743
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008184785442553649,
+      "loss": 3.1237,
+      "step": 3744
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008183778829352704,
+      "loss": 3.1759,
+      "step": 3745
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008182771999062078,
+      "loss": 3.0774,
+      "step": 3746
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008181764951750423,
+      "loss": 3.1224,
+      "step": 3747
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008180757687486405,
+      "loss": 2.9167,
+      "step": 3748
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008179750206338708,
+      "loss": 3.08,
+      "step": 3749
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008178742508376024,
+      "loss": 3.158,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000817773459366707,
+      "loss": 3.3602,
+      "step": 3751
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008176726462280568,
+      "loss": 3.1761,
+      "step": 3752
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008175718114285257,
+      "loss": 2.8653,
+      "step": 3753
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008174709549749896,
+      "loss": 2.919,
+      "step": 3754
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008173700768743255,
+      "loss": 3.0444,
+      "step": 3755
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008172691771334116,
+      "loss": 3.0852,
+      "step": 3756
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.000817168255759128,
+      "loss": 2.9378,
+      "step": 3757
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008170673127583563,
+      "loss": 3.0608,
+      "step": 3758
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008169663481379792,
+      "loss": 3.2201,
+      "step": 3759
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008168653619048811,
+      "loss": 3.1146,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008167643540659482,
+      "loss": 3.0154,
+      "step": 3761
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008166633246280674,
+      "loss": 3.3024,
+      "step": 3762
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008165622735981277,
+      "loss": 3.0773,
+      "step": 3763
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008164612009830194,
+      "loss": 3.0534,
+      "step": 3764
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008163601067896342,
+      "loss": 3.1676,
+      "step": 3765
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008162589910248654,
+      "loss": 3.2049,
+      "step": 3766
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008161578536956076,
+      "loss": 3.1444,
+      "step": 3767
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008160566948087571,
+      "loss": 3.0372,
+      "step": 3768
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008159555143712114,
+      "loss": 3.2572,
+      "step": 3769
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008158543123898696,
+      "loss": 3.0543,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008157530888716325,
+      "loss": 3.262,
+      "step": 3771
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008156518438234017,
+      "loss": 3.0077,
+      "step": 3772
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008155505772520813,
+      "loss": 3.1982,
+      "step": 3773
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008154492891645757,
+      "loss": 2.9078,
+      "step": 3774
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008153479795677918,
+      "loss": 3.1945,
+      "step": 3775
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008152466484686373,
+      "loss": 3.2719,
+      "step": 3776
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008151452958740217,
+      "loss": 3.163,
+      "step": 3777
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008150439217908557,
+      "loss": 2.8988,
+      "step": 3778
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008149425262260515,
+      "loss": 3.0919,
+      "step": 3779
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008148411091865231,
+      "loss": 2.818,
+      "step": 3780
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008147396706791859,
+      "loss": 3.1548,
+      "step": 3781
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008146382107109561,
+      "loss": 3.0895,
+      "step": 3782
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0008145367292887524,
+      "loss": 3.0678,
+      "step": 3783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008144352264194939,
+      "loss": 3.1558,
+      "step": 3784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008143337021101021,
+      "loss": 3.3313,
+      "step": 3785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008142321563674995,
+      "loss": 3.2747,
+      "step": 3786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008141305891986101,
+      "loss": 3.3815,
+      "step": 3787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008140290006103592,
+      "loss": 3.1849,
+      "step": 3788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008139273906096738,
+      "loss": 3.182,
+      "step": 3789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008138257592034824,
+      "loss": 3.0742,
+      "step": 3790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008137241063987149,
+      "loss": 2.9712,
+      "step": 3791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008136224322023023,
+      "loss": 3.0091,
+      "step": 3792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008135207366211779,
+      "loss": 2.8164,
+      "step": 3793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008134190196622754,
+      "loss": 2.966,
+      "step": 3794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008133172813325308,
+      "loss": 3.0768,
+      "step": 3795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008132155216388811,
+      "loss": 2.888,
+      "step": 3796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000813113740588265,
+      "loss": 2.9177,
+      "step": 3797
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008130119381876226,
+      "loss": 2.9101,
+      "step": 3798
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008129101144438952,
+      "loss": 2.9456,
+      "step": 3799
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008128082693640261,
+      "loss": 3.1456,
+      "step": 3800
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008127064029549596,
+      "loss": 3.0061,
+      "step": 3801
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008126045152236412,
+      "loss": 2.9919,
+      "step": 3802
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008125026061770187,
+      "loss": 3.0086,
+      "step": 3803
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008124006758220408,
+      "loss": 3.1784,
+      "step": 3804
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008122987241656576,
+      "loss": 3.0995,
+      "step": 3805
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000812196751214821,
+      "loss": 3.168,
+      "step": 3806
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008120947569764838,
+      "loss": 3.1889,
+      "step": 3807
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000811992741457601,
+      "loss": 3.0223,
+      "step": 3808
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008118907046651284,
+      "loss": 3.0398,
+      "step": 3809
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008117886466060236,
+      "loss": 3.2743,
+      "step": 3810
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008116865672872455,
+      "loss": 3.1345,
+      "step": 3811
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008115844667157545,
+      "loss": 3.0185,
+      "step": 3812
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008114823448985125,
+      "loss": 2.9293,
+      "step": 3813
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008113802018424827,
+      "loss": 3.1734,
+      "step": 3814
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008112780375546298,
+      "loss": 3.0323,
+      "step": 3815
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008111758520419202,
+      "loss": 3.0719,
+      "step": 3816
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008110736453113215,
+      "loss": 3.0855,
+      "step": 3817
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008109714173698026,
+      "loss": 3.1203,
+      "step": 3818
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008108691682243341,
+      "loss": 2.9701,
+      "step": 3819
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008107668978818881,
+      "loss": 3.1157,
+      "step": 3820
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008106646063494378,
+      "loss": 3.0216,
+      "step": 3821
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008105622936339584,
+      "loss": 3.1443,
+      "step": 3822
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000810459959742426,
+      "loss": 3.233,
+      "step": 3823
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008103576046818182,
+      "loss": 3.1089,
+      "step": 3824
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008102552284591145,
+      "loss": 3.1644,
+      "step": 3825
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008101528310812954,
+      "loss": 2.9752,
+      "step": 3826
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008100504125553431,
+      "loss": 3.0086,
+      "step": 3827
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008099479728882409,
+      "loss": 3.0729,
+      "step": 3828
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000809845512086974,
+      "loss": 2.8101,
+      "step": 3829
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008097430301585286,
+      "loss": 2.8688,
+      "step": 3830
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008096405271098928,
+      "loss": 3.2133,
+      "step": 3831
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008095380029480557,
+      "loss": 2.953,
+      "step": 3832
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000809435457680008,
+      "loss": 3.0959,
+      "step": 3833
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000809332891312742,
+      "loss": 3.1695,
+      "step": 3834
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008092303038532511,
+      "loss": 2.9651,
+      "step": 3835
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008091276953085307,
+      "loss": 2.8645,
+      "step": 3836
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008090250656855771,
+      "loss": 2.8691,
+      "step": 3837
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000808922414991388,
+      "loss": 3.0997,
+      "step": 3838
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008088197432329631,
+      "loss": 3.0351,
+      "step": 3839
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008087170504173032,
+      "loss": 3.1687,
+      "step": 3840
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00080861433655141,
+      "loss": 3.0171,
+      "step": 3841
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008085116016422879,
+      "loss": 2.8405,
+      "step": 3842
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008084088456969417,
+      "loss": 3.187,
+      "step": 3843
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008083060687223776,
+      "loss": 3.0741,
+      "step": 3844
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000808203270725604,
+      "loss": 3.1301,
+      "step": 3845
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008081004517136301,
+      "loss": 3.1088,
+      "step": 3846
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008079976116934668,
+      "loss": 2.9408,
+      "step": 3847
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008078947506721263,
+      "loss": 3.0219,
+      "step": 3848
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008077918686566224,
+      "loss": 2.9574,
+      "step": 3849
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008076889656539703,
+      "loss": 3.1242,
+      "step": 3850
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008075860416711864,
+      "loss": 3.1167,
+      "step": 3851
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008074830967152887,
+      "loss": 2.9743,
+      "step": 3852
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008073801307932967,
+      "loss": 3.1097,
+      "step": 3853
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008072771439122311,
+      "loss": 2.914,
+      "step": 3854
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008071741360791143,
+      "loss": 3.1436,
+      "step": 3855
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00080707110730097,
+      "loss": 3.065,
+      "step": 3856
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008069680575848235,
+      "loss": 3.1521,
+      "step": 3857
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008068649869377012,
+      "loss": 3.127,
+      "step": 3858
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008067618953666309,
+      "loss": 3.2844,
+      "step": 3859
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008066587828786424,
+      "loss": 2.939,
+      "step": 3860
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008065556494807662,
+      "loss": 3.2443,
+      "step": 3861
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008064524951800349,
+      "loss": 3.001,
+      "step": 3862
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008063493199834819,
+      "loss": 3.2706,
+      "step": 3863
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008062461238981426,
+      "loss": 3.161,
+      "step": 3864
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008061429069310533,
+      "loss": 3.1613,
+      "step": 3865
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008060396690892521,
+      "loss": 3.1699,
+      "step": 3866
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008059364103797784,
+      "loss": 3.0414,
+      "step": 3867
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008058331308096729,
+      "loss": 3.1208,
+      "step": 3868
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008057298303859779,
+      "loss": 3.2783,
+      "step": 3869
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008056265091157371,
+      "loss": 3.0091,
+      "step": 3870
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008055231670059957,
+      "loss": 3.257,
+      "step": 3871
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008054198040637998,
+      "loss": 3.0404,
+      "step": 3872
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008053164202961977,
+      "loss": 3.0004,
+      "step": 3873
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008052130157102386,
+      "loss": 2.9255,
+      "step": 3874
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008051095903129734,
+      "loss": 2.9539,
+      "step": 3875
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008050061441114543,
+      "loss": 3.0717,
+      "step": 3876
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008049026771127344,
+      "loss": 3.1104,
+      "step": 3877
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008047991893238693,
+      "loss": 2.8242,
+      "step": 3878
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008046956807519151,
+      "loss": 3.144,
+      "step": 3879
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008045921514039298,
+      "loss": 3.2512,
+      "step": 3880
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008044886012869728,
+      "loss": 2.746,
+      "step": 3881
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008043850304081045,
+      "loss": 3.0487,
+      "step": 3882
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008042814387743872,
+      "loss": 3.1322,
+      "step": 3883
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008041778263928842,
+      "loss": 2.9978,
+      "step": 3884
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008040741932706608,
+      "loss": 3.2647,
+      "step": 3885
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000803970539414783,
+      "loss": 2.9117,
+      "step": 3886
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008038668648323187,
+      "loss": 3.2481,
+      "step": 3887
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008037631695303371,
+      "loss": 2.8571,
+      "step": 3888
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008036594535159087,
+      "loss": 3.1069,
+      "step": 3889
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008035557167961056,
+      "loss": 2.8271,
+      "step": 3890
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000803451959378001,
+      "loss": 3.1863,
+      "step": 3891
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008033481812686701,
+      "loss": 2.7918,
+      "step": 3892
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008032443824751888,
+      "loss": 2.995,
+      "step": 3893
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008031405630046348,
+      "loss": 2.9076,
+      "step": 3894
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008030367228640872,
+      "loss": 3.0477,
+      "step": 3895
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008029328620606266,
+      "loss": 3.0703,
+      "step": 3896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008028289806013346,
+      "loss": 2.9599,
+      "step": 3897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008027250784932947,
+      "loss": 2.7825,
+      "step": 3898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008026211557435914,
+      "loss": 3.0937,
+      "step": 3899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008025172123593109,
+      "loss": 3.1364,
+      "step": 3900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008024132483475408,
+      "loss": 3.2329,
+      "step": 3901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008023092637153697,
+      "loss": 3.0542,
+      "step": 3902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008022052584698882,
+      "loss": 3.0173,
+      "step": 3903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000802101232618188,
+      "loss": 3.129,
+      "step": 3904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008019971861673621,
+      "loss": 3.0888,
+      "step": 3905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.000801893119124505,
+      "loss": 3.083,
+      "step": 3906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0008017890314967129,
+      "loss": 3.0357,
+      "step": 3907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008016849232910828,
+      "loss": 2.987,
+      "step": 3908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008015807945147135,
+      "loss": 3.0747,
+      "step": 3909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008014766451747055,
+      "loss": 3.0658,
+      "step": 3910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008013724752781599,
+      "loss": 3.0803,
+      "step": 3911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008012682848321799,
+      "loss": 3.0407,
+      "step": 3912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008011640738438698,
+      "loss": 3.1905,
+      "step": 3913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008010598423203352,
+      "loss": 2.8477,
+      "step": 3914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008009555902686835,
+      "loss": 3.0204,
+      "step": 3915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008008513176960231,
+      "loss": 3.0901,
+      "step": 3916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008007470246094639,
+      "loss": 3.1378,
+      "step": 3917
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008006427110161174,
+      "loss": 2.8906,
+      "step": 3918
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008005383769230963,
+      "loss": 3.0565,
+      "step": 3919
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008004340223375145,
+      "loss": 3.3377,
+      "step": 3920
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008003296472664878,
+      "loss": 3.0885,
+      "step": 3921
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008002252517171331,
+      "loss": 3.0967,
+      "step": 3922
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008001208356965688,
+      "loss": 3.0642,
+      "step": 3923
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0008000163992119145,
+      "loss": 3.0393,
+      "step": 3924
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007999119422702911,
+      "loss": 3.106,
+      "step": 3925
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007998074648788218,
+      "loss": 2.9772,
+      "step": 3926
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007997029670446298,
+      "loss": 3.312,
+      "step": 3927
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007995984487748407,
+      "loss": 3.1706,
+      "step": 3928
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007994939100765813,
+      "loss": 2.9484,
+      "step": 3929
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007993893509569796,
+      "loss": 2.9963,
+      "step": 3930
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007992847714231652,
+      "loss": 3.0967,
+      "step": 3931
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007991801714822686,
+      "loss": 2.9378,
+      "step": 3932
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007990755511414225,
+      "loss": 3.0577,
+      "step": 3933
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007989709104077602,
+      "loss": 3.1189,
+      "step": 3934
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007988662492884171,
+      "loss": 3.0598,
+      "step": 3935
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007987615677905294,
+      "loss": 3.2104,
+      "step": 3936
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000798656865921235,
+      "loss": 3.1114,
+      "step": 3937
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000798552143687673,
+      "loss": 3.012,
+      "step": 3938
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007984474010969845,
+      "loss": 3.1555,
+      "step": 3939
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007983426381563108,
+      "loss": 2.8193,
+      "step": 3940
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007982378548727956,
+      "loss": 2.8131,
+      "step": 3941
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007981330512535839,
+      "loss": 3.1613,
+      "step": 3942
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007980282273058215,
+      "loss": 2.9005,
+      "step": 3943
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007979233830366561,
+      "loss": 3.1139,
+      "step": 3944
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007978185184532366,
+      "loss": 3.3098,
+      "step": 3945
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007977136335627133,
+      "loss": 3.0866,
+      "step": 3946
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000797608728372238,
+      "loss": 3.1562,
+      "step": 3947
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007975038028889637,
+      "loss": 3.1657,
+      "step": 3948
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007973988571200449,
+      "loss": 3.0807,
+      "step": 3949
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007972938910726373,
+      "loss": 2.8688,
+      "step": 3950
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007971889047538984,
+      "loss": 3.1446,
+      "step": 3951
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007970838981709866,
+      "loss": 3.0269,
+      "step": 3952
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007969788713310619,
+      "loss": 3.1207,
+      "step": 3953
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007968738242412858,
+      "loss": 3.2031,
+      "step": 3954
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007967687569088211,
+      "loss": 3.1194,
+      "step": 3955
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007966636693408318,
+      "loss": 2.9986,
+      "step": 3956
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007965585615444835,
+      "loss": 2.9521,
+      "step": 3957
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007964534335269429,
+      "loss": 2.919,
+      "step": 3958
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007963482852953785,
+      "loss": 2.8689,
+      "step": 3959
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007962431168569601,
+      "loss": 3.0599,
+      "step": 3960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007961379282188584,
+      "loss": 3.1026,
+      "step": 3961
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007960327193882457,
+      "loss": 2.86,
+      "step": 3962
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007959274903722964,
+      "loss": 3.174,
+      "step": 3963
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000795822241178185,
+      "loss": 2.6933,
+      "step": 3964
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007957169718130885,
+      "loss": 3.0574,
+      "step": 3965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007956116822841845,
+      "loss": 3.067,
+      "step": 3966
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007955063725986525,
+      "loss": 3.0017,
+      "step": 3967
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007954010427636731,
+      "loss": 3.024,
+      "step": 3968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007952956927864283,
+      "loss": 2.951,
+      "step": 3969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007951903226741016,
+      "loss": 3.0083,
+      "step": 3970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007950849324338777,
+      "loss": 3.0294,
+      "step": 3971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007949795220729427,
+      "loss": 3.0714,
+      "step": 3972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007948740915984843,
+      "loss": 2.8241,
+      "step": 3973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007947686410176914,
+      "loss": 3.1801,
+      "step": 3974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000794663170337754,
+      "loss": 3.2648,
+      "step": 3975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000794557679565864,
+      "loss": 3.143,
+      "step": 3976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007944521687092143,
+      "loss": 2.8574,
+      "step": 3977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007943466377749995,
+      "loss": 3.0277,
+      "step": 3978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007942410867704151,
+      "loss": 3.1106,
+      "step": 3979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007941355157026582,
+      "loss": 3.1247,
+      "step": 3980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007940299245789273,
+      "loss": 2.9217,
+      "step": 3981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007939243134064224,
+      "loss": 3.1761,
+      "step": 3982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007938186821923448,
+      "loss": 3.0046,
+      "step": 3983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007937130309438967,
+      "loss": 3.0012,
+      "step": 3984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007936073596682825,
+      "loss": 3.0966,
+      "step": 3985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007935016683727072,
+      "loss": 3.1192,
+      "step": 3986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007933959570643776,
+      "loss": 2.8518,
+      "step": 3987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007932902257505017,
+      "loss": 3.1414,
+      "step": 3988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007931844744382891,
+      "loss": 3.0753,
+      "step": 3989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007930787031349502,
+      "loss": 3.0831,
+      "step": 3990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007929729118476976,
+      "loss": 3.1532,
+      "step": 3991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007928671005837444,
+      "loss": 3.0044,
+      "step": 3992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007927612693503056,
+      "loss": 2.8691,
+      "step": 3993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007926554181545977,
+      "loss": 3.0426,
+      "step": 3994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000792549547003838,
+      "loss": 2.9648,
+      "step": 3995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007924436559052455,
+      "loss": 3.1067,
+      "step": 3996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007923377448660405,
+      "loss": 3.1739,
+      "step": 3997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007922318138934445,
+      "loss": 3.0369,
+      "step": 3998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000792125862994681,
+      "loss": 3.1603,
+      "step": 3999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007920198921769739,
+      "loss": 2.9861,
+      "step": 4000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007919139014475492,
+      "loss": 3.0845,
+      "step": 4001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000791807890813634,
+      "loss": 3.1677,
+      "step": 4002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007917018602824566,
+      "loss": 3.3696,
+      "step": 4003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007915958098612468,
+      "loss": 2.9564,
+      "step": 4004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000791489739557236,
+      "loss": 3.2202,
+      "step": 4005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007913836493776567,
+      "loss": 2.9363,
+      "step": 4006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007912775393297427,
+      "loss": 3.1669,
+      "step": 4007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007911714094207291,
+      "loss": 3.2448,
+      "step": 4008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007910652596578527,
+      "loss": 2.8464,
+      "step": 4009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007909590900483513,
+      "loss": 2.9593,
+      "step": 4010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007908529005994642,
+      "loss": 3.0034,
+      "step": 4011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007907466913184323,
+      "loss": 3.1505,
+      "step": 4012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007906404622124974,
+      "loss": 3.0411,
+      "step": 4013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007905342132889029,
+      "loss": 3.1828,
+      "step": 4014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007904279445548934,
+      "loss": 3.1232,
+      "step": 4015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007903216560177151,
+      "loss": 2.9725,
+      "step": 4016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007902153476846153,
+      "loss": 3.1022,
+      "step": 4017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007901090195628429,
+      "loss": 3.1602,
+      "step": 4018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007900026716596478,
+      "loss": 2.9674,
+      "step": 4019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007898963039822817,
+      "loss": 3.005,
+      "step": 4020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007897899165379971,
+      "loss": 3.3223,
+      "step": 4021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007896835093340486,
+      "loss": 2.9976,
+      "step": 4022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007895770823776912,
+      "loss": 2.969,
+      "step": 4023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007894706356761821,
+      "loss": 3.1956,
+      "step": 4024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007893641692367793,
+      "loss": 3.3823,
+      "step": 4025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007892576830667424,
+      "loss": 3.1527,
+      "step": 4026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007891511771733324,
+      "loss": 2.9389,
+      "step": 4027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007890446515638113,
+      "loss": 3.2834,
+      "step": 4028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.000788938106245443,
+      "loss": 3.1123,
+      "step": 4029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007888315412254921,
+      "loss": 3.0538,
+      "step": 4030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0007887249565112249,
+      "loss": 3.1272,
+      "step": 4031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007886183521099091,
+      "loss": 3.3537,
+      "step": 4032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007885117280288137,
+      "loss": 2.9351,
+      "step": 4033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007884050842752087,
+      "loss": 3.2337,
+      "step": 4034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007882984208563661,
+      "loss": 3.1746,
+      "step": 4035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007881917377795586,
+      "loss": 3.1344,
+      "step": 4036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007880850350520608,
+      "loss": 3.0233,
+      "step": 4037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007879783126811481,
+      "loss": 2.8069,
+      "step": 4038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007878715706740975,
+      "loss": 2.9631,
+      "step": 4039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007877648090381875,
+      "loss": 3.0492,
+      "step": 4040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007876580277806976,
+      "loss": 3.0336,
+      "step": 4041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007875512269089087,
+      "loss": 3.0391,
+      "step": 4042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007874444064301034,
+      "loss": 3.0683,
+      "step": 4043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007873375663515654,
+      "loss": 2.9665,
+      "step": 4044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007872307066805795,
+      "loss": 3.0698,
+      "step": 4045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007871238274244321,
+      "loss": 3.2284,
+      "step": 4046
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007870169285904109,
+      "loss": 3.1421,
+      "step": 4047
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007869100101858051,
+      "loss": 3.1653,
+      "step": 4048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007868030722179049,
+      "loss": 3.1597,
+      "step": 4049
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007866961146940019,
+      "loss": 3.1238,
+      "step": 4050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007865891376213894,
+      "loss": 2.9357,
+      "step": 4051
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007864821410073615,
+      "loss": 2.8097,
+      "step": 4052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000786375124859214,
+      "loss": 3.3073,
+      "step": 4053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007862680891842438,
+      "loss": 3.0211,
+      "step": 4054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007861610339897493,
+      "loss": 3.253,
+      "step": 4055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007860539592830303,
+      "loss": 3.4049,
+      "step": 4056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007859468650713877,
+      "loss": 3.1218,
+      "step": 4057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007858397513621241,
+      "loss": 3.0382,
+      "step": 4058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007857326181625427,
+      "loss": 3.0728,
+      "step": 4059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007856254654799488,
+      "loss": 3.2297,
+      "step": 4060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007855182933216489,
+      "loss": 2.9615,
+      "step": 4061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007854111016949504,
+      "loss": 3.1314,
+      "step": 4062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007853038906071621,
+      "loss": 3.0305,
+      "step": 4063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007851966600655949,
+      "loss": 3.3808,
+      "step": 4064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007850894100775599,
+      "loss": 3.0113,
+      "step": 4065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007849821406503701,
+      "loss": 3.0771,
+      "step": 4066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007848748517913402,
+      "loss": 2.8775,
+      "step": 4067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007847675435077855,
+      "loss": 3.0714,
+      "step": 4068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007846602158070229,
+      "loss": 3.075,
+      "step": 4069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007845528686963709,
+      "loss": 2.9898,
+      "step": 4070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000784445502183149,
+      "loss": 2.9622,
+      "step": 4071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007843381162746779,
+      "loss": 3.0446,
+      "step": 4072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007842307109782801,
+      "loss": 3.2922,
+      "step": 4073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007841232863012791,
+      "loss": 3.3306,
+      "step": 4074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007840158422509995,
+      "loss": 3.1239,
+      "step": 4075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007839083788347681,
+      "loss": 2.9066,
+      "step": 4076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007838008960599119,
+      "loss": 3.2085,
+      "step": 4077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007836933939337599,
+      "loss": 3.2066,
+      "step": 4078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007835858724636421,
+      "loss": 3.1082,
+      "step": 4079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007834783316568903,
+      "loss": 2.9398,
+      "step": 4080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007833707715208372,
+      "loss": 2.9657,
+      "step": 4081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007832631920628166,
+      "loss": 3.0437,
+      "step": 4082
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007831555932901642,
+      "loss": 3.1124,
+      "step": 4083
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007830479752102169,
+      "loss": 2.7983,
+      "step": 4084
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007829403378303124,
+      "loss": 3.173,
+      "step": 4085
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007828326811577903,
+      "loss": 2.9062,
+      "step": 4086
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007827250051999913,
+      "loss": 3.1744,
+      "step": 4087
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007826173099642575,
+      "loss": 2.961,
+      "step": 4088
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000782509595457932,
+      "loss": 3.0611,
+      "step": 4089
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007824018616883594,
+      "loss": 2.9509,
+      "step": 4090
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007822941086628859,
+      "loss": 3.1162,
+      "step": 4091
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007821863363888586,
+      "loss": 3.1392,
+      "step": 4092
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007820785448736263,
+      "loss": 3.1012,
+      "step": 4093
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007819707341245386,
+      "loss": 2.9671,
+      "step": 4094
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007818629041489468,
+      "loss": 2.9482,
+      "step": 4095
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007817550549542034,
+      "loss": 3.1523,
+      "step": 4096
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007816471865476623,
+      "loss": 3.1548,
+      "step": 4097
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007815392989366786,
+      "loss": 3.0171,
+      "step": 4098
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007814313921286088,
+      "loss": 3.2095,
+      "step": 4099
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007813234661308105,
+      "loss": 3.3425,
+      "step": 4100
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007812155209506429,
+      "loss": 2.8184,
+      "step": 4101
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007811075565954663,
+      "loss": 2.856,
+      "step": 4102
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007809995730726425,
+      "loss": 3.0801,
+      "step": 4103
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000780891570389534,
+      "loss": 3.0438,
+      "step": 4104
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007807835485535059,
+      "loss": 3.2846,
+      "step": 4105
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007806755075719232,
+      "loss": 3.0017,
+      "step": 4106
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007805674474521529,
+      "loss": 2.9694,
+      "step": 4107
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007804593682015633,
+      "loss": 3.2034,
+      "step": 4108
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007803512698275239,
+      "loss": 3.0751,
+      "step": 4109
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007802431523374055,
+      "loss": 3.0703,
+      "step": 4110
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007801350157385803,
+      "loss": 3.0458,
+      "step": 4111
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007800268600384213,
+      "loss": 2.9705,
+      "step": 4112
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007799186852443039,
+      "loss": 3.3371,
+      "step": 4113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007798104913636035,
+      "loss": 2.9455,
+      "step": 4114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007797022784036981,
+      "loss": 3.2088,
+      "step": 4115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007795940463719655,
+      "loss": 3.2431,
+      "step": 4116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007794857952757861,
+      "loss": 2.9384,
+      "step": 4117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007793775251225412,
+      "loss": 2.9495,
+      "step": 4118
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007792692359196135,
+      "loss": 3.3405,
+      "step": 4119
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007791609276743861,
+      "loss": 3.0579,
+      "step": 4120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007790526003942447,
+      "loss": 3.2847,
+      "step": 4121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007789442540865756,
+      "loss": 3.07,
+      "step": 4122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007788358887587665,
+      "loss": 3.1441,
+      "step": 4123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007787275044182064,
+      "loss": 3.0688,
+      "step": 4124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007786191010722857,
+      "loss": 3.0817,
+      "step": 4125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007785106787283959,
+      "loss": 2.981,
+      "step": 4126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007784022373939299,
+      "loss": 3.1931,
+      "step": 4127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007782937770762821,
+      "loss": 3.1205,
+      "step": 4128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007781852977828477,
+      "loss": 3.0518,
+      "step": 4129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007780767995210237,
+      "loss": 2.8169,
+      "step": 4130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007779682822982079,
+      "loss": 3.0404,
+      "step": 4131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007778597461218001,
+      "loss": 3.0786,
+      "step": 4132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007777511909992006,
+      "loss": 3.1853,
+      "step": 4133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007776426169378116,
+      "loss": 2.8965,
+      "step": 4134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007775340239450362,
+      "loss": 3.1088,
+      "step": 4135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007774254120282791,
+      "loss": 3.1164,
+      "step": 4136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000777316781194946,
+      "loss": 3.2464,
+      "step": 4137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007772081314524439,
+      "loss": 3.0435,
+      "step": 4138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007770994628081816,
+      "loss": 3.0553,
+      "step": 4139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007769907752695682,
+      "loss": 2.8383,
+      "step": 4140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007768820688440153,
+      "loss": 3.0087,
+      "step": 4141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007767733435389348,
+      "loss": 3.1844,
+      "step": 4142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007766645993617405,
+      "loss": 3.0742,
+      "step": 4143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007765558363198471,
+      "loss": 2.9238,
+      "step": 4144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007764470544206706,
+      "loss": 2.9602,
+      "step": 4145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007763382536716288,
+      "loss": 3.1073,
+      "step": 4146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00077622943408014,
+      "loss": 2.9791,
+      "step": 4147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007761205956536244,
+      "loss": 2.9513,
+      "step": 4148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007760117383995034,
+      "loss": 3.1465,
+      "step": 4149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007759028623251994,
+      "loss": 3.2168,
+      "step": 4150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000775793967438136,
+      "loss": 3.1756,
+      "step": 4151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007756850537457388,
+      "loss": 3.0799,
+      "step": 4152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.000775576121255434,
+      "loss": 2.991,
+      "step": 4153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007754671699746492,
+      "loss": 3.0432,
+      "step": 4154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0007753581999108135,
+      "loss": 3.0522,
+      "step": 4155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000775249211071357,
+      "loss": 2.9664,
+      "step": 4156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007751402034637114,
+      "loss": 3.1256,
+      "step": 4157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007750311770953095,
+      "loss": 3.1899,
+      "step": 4158
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007749221319735852,
+      "loss": 3.1559,
+      "step": 4159
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007748130681059742,
+      "loss": 3.3584,
+      "step": 4160
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007747039854999128,
+      "loss": 3.0762,
+      "step": 4161
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007745948841628391,
+      "loss": 3.2381,
+      "step": 4162
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007744857641021923,
+      "loss": 3.1051,
+      "step": 4163
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007743766253254128,
+      "loss": 3.0917,
+      "step": 4164
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007742674678399424,
+      "loss": 3.1141,
+      "step": 4165
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007741582916532242,
+      "loss": 3.0207,
+      "step": 4166
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007740490967727023,
+      "loss": 3.084,
+      "step": 4167
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007739398832058225,
+      "loss": 2.8186,
+      "step": 4168
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007738306509600317,
+      "loss": 3.2973,
+      "step": 4169
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007737214000427777,
+      "loss": 3.0738,
+      "step": 4170
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007736121304615102,
+      "loss": 2.9142,
+      "step": 4171
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007735028422236798,
+      "loss": 3.1805,
+      "step": 4172
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007733935353367385,
+      "loss": 3.1357,
+      "step": 4173
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007732842098081395,
+      "loss": 3.1318,
+      "step": 4174
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000773174865645337,
+      "loss": 3.096,
+      "step": 4175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007730655028557871,
+      "loss": 3.0997,
+      "step": 4176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007729561214469468,
+      "loss": 3.0051,
+      "step": 4177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007728467214262743,
+      "loss": 3.1019,
+      "step": 4178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007727373028012292,
+      "loss": 3.1148,
+      "step": 4179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007726278655792722,
+      "loss": 3.2891,
+      "step": 4180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007725184097678657,
+      "loss": 3.1173,
+      "step": 4181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007724089353744729,
+      "loss": 2.962,
+      "step": 4182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007722994424065584,
+      "loss": 3.2536,
+      "step": 4183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007721899308715881,
+      "loss": 3.1395,
+      "step": 4184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007720804007770293,
+      "loss": 2.9159,
+      "step": 4185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007719708521303503,
+      "loss": 2.9625,
+      "step": 4186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007718612849390209,
+      "loss": 3.0775,
+      "step": 4187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007717516992105121,
+      "loss": 3.146,
+      "step": 4188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000771642094952296,
+      "loss": 2.9029,
+      "step": 4189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007715324721718461,
+      "loss": 3.1076,
+      "step": 4190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007714228308766373,
+      "loss": 3.214,
+      "step": 4191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007713131710741454,
+      "loss": 2.9654,
+      "step": 4192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007712034927718481,
+      "loss": 3.2757,
+      "step": 4193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007710937959772234,
+      "loss": 3.0636,
+      "step": 4194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007709840806977515,
+      "loss": 2.8761,
+      "step": 4195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007708743469409132,
+      "loss": 3.1142,
+      "step": 4196
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000770764594714191,
+      "loss": 2.994,
+      "step": 4197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007706548240250685,
+      "loss": 3.2928,
+      "step": 4198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007705450348810304,
+      "loss": 3.0134,
+      "step": 4199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007704352272895629,
+      "loss": 2.9744,
+      "step": 4200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007703254012581535,
+      "loss": 3.0774,
+      "step": 4201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007702155567942905,
+      "loss": 3.1254,
+      "step": 4202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000770105693905464,
+      "loss": 3.0318,
+      "step": 4203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007699958125991648,
+      "loss": 3.1546,
+      "step": 4204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007698859128828859,
+      "loss": 2.9216,
+      "step": 4205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007697759947641205,
+      "loss": 3.0264,
+      "step": 4206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007696660582503634,
+      "loss": 3.069,
+      "step": 4207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007695561033491111,
+      "loss": 3.0796,
+      "step": 4208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007694461300678608,
+      "loss": 2.7729,
+      "step": 4209
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007693361384141112,
+      "loss": 3.2268,
+      "step": 4210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007692261283953621,
+      "loss": 3.0639,
+      "step": 4211
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007691161000191148,
+      "loss": 3.153,
+      "step": 4212
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007690060532928718,
+      "loss": 2.8153,
+      "step": 4213
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007688959882241364,
+      "loss": 3.133,
+      "step": 4214
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000768785904820414,
+      "loss": 2.935,
+      "step": 4215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007686758030892105,
+      "loss": 3.2021,
+      "step": 4216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007685656830380332,
+      "loss": 3.2022,
+      "step": 4217
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000768455544674391,
+      "loss": 2.8557,
+      "step": 4218
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007683453880057936,
+      "loss": 3.122,
+      "step": 4219
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007682352130397523,
+      "loss": 3.0644,
+      "step": 4220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007681250197837796,
+      "loss": 3.2374,
+      "step": 4221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000768014808245389,
+      "loss": 3.1187,
+      "step": 4222
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007679045784320954,
+      "loss": 3.199,
+      "step": 4223
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000767794330351415,
+      "loss": 3.0484,
+      "step": 4224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007676840640108653,
+      "loss": 3.0204,
+      "step": 4225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007675737794179646,
+      "loss": 3.0154,
+      "step": 4226
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007674634765802332,
+      "loss": 3.1207,
+      "step": 4227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007673531555051919,
+      "loss": 3.1968,
+      "step": 4228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007672428162003633,
+      "loss": 3.0761,
+      "step": 4229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007671324586732708,
+      "loss": 3.0501,
+      "step": 4230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007670220829314396,
+      "loss": 3.3652,
+      "step": 4231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007669116889823954,
+      "loss": 3.3865,
+      "step": 4232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007668012768336658,
+      "loss": 3.0913,
+      "step": 4233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007666908464927791,
+      "loss": 3.0135,
+      "step": 4234
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007665803979672656,
+      "loss": 3.0164,
+      "step": 4235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007664699312646559,
+      "loss": 3.4346,
+      "step": 4236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007663594463924825,
+      "loss": 3.0135,
+      "step": 4237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000766248943358279,
+      "loss": 3.2583,
+      "step": 4238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007661384221695802,
+      "loss": 3.0618,
+      "step": 4239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007660278828339218,
+      "loss": 3.1507,
+      "step": 4240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007659173253588414,
+      "loss": 3.2186,
+      "step": 4241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007658067497518773,
+      "loss": 2.9757,
+      "step": 4242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007656961560205692,
+      "loss": 3.0585,
+      "step": 4243
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007655855441724584,
+      "loss": 3.0306,
+      "step": 4244
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007654749142150867,
+      "loss": 3.0641,
+      "step": 4245
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007653642661559978,
+      "loss": 3.0343,
+      "step": 4246
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007652536000027361,
+      "loss": 3.0601,
+      "step": 4247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007651429157628477,
+      "loss": 3.1122,
+      "step": 4248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007650322134438798,
+      "loss": 3.145,
+      "step": 4249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007649214930533805,
+      "loss": 3.1497,
+      "step": 4250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007648107545988996,
+      "loss": 2.9668,
+      "step": 4251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007646999980879879,
+      "loss": 3.1296,
+      "step": 4252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007645892235281975,
+      "loss": 3.0883,
+      "step": 4253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007644784309270815,
+      "loss": 2.9613,
+      "step": 4254
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007643676202921946,
+      "loss": 2.9581,
+      "step": 4255
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007642567916310927,
+      "loss": 3.1567,
+      "step": 4256
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007641459449513324,
+      "loss": 3.2498,
+      "step": 4257
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000764035080260472,
+      "loss": 2.9462,
+      "step": 4258
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007639241975660714,
+      "loss": 2.9851,
+      "step": 4259
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007638132968756908,
+      "loss": 3.0166,
+      "step": 4260
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000763702378196892,
+      "loss": 2.9728,
+      "step": 4261
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007635914415372385,
+      "loss": 2.9433,
+      "step": 4262
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007634804869042944,
+      "loss": 3.1129,
+      "step": 4263
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007633695143056254,
+      "loss": 3.1446,
+      "step": 4264
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007632585237487981,
+      "loss": 3.2838,
+      "step": 4265
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007631475152413809,
+      "loss": 2.9243,
+      "step": 4266
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007630364887909425,
+      "loss": 3.0854,
+      "step": 4267
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000762925444405054,
+      "loss": 3.0351,
+      "step": 4268
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007628143820912865,
+      "loss": 3.082,
+      "step": 4269
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007627033018572134,
+      "loss": 2.9938,
+      "step": 4270
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007625922037104085,
+      "loss": 3.1147,
+      "step": 4271
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007624810876584472,
+      "loss": 2.8669,
+      "step": 4272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007623699537089062,
+      "loss": 3.0407,
+      "step": 4273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007622588018693633,
+      "loss": 3.1365,
+      "step": 4274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007621476321473973,
+      "loss": 3.151,
+      "step": 4275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007620364445505887,
+      "loss": 3.1434,
+      "step": 4276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000761925239086519,
+      "loss": 3.1036,
+      "step": 4277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007618140157627705,
+      "loss": 3.0121,
+      "step": 4278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0007617027745869275,
+      "loss": 3.187,
+      "step": 4279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007615915155665748,
+      "loss": 2.9352,
+      "step": 4280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000761480238709299,
+      "loss": 2.9782,
+      "step": 4281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007613689440226874,
+      "loss": 3.1393,
+      "step": 4282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007612576315143289,
+      "loss": 2.9987,
+      "step": 4283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007611463011918133,
+      "loss": 3.1549,
+      "step": 4284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007610349530627322,
+      "loss": 3.1907,
+      "step": 4285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007609235871346776,
+      "loss": 2.9963,
+      "step": 4286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007608122034152431,
+      "loss": 3.151,
+      "step": 4287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007607008019120239,
+      "loss": 3.0851,
+      "step": 4288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007605893826326156,
+      "loss": 3.1783,
+      "step": 4289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007604779455846158,
+      "loss": 3.1549,
+      "step": 4290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007603664907756229,
+      "loss": 3.1865,
+      "step": 4291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007602550182132364,
+      "loss": 3.0496,
+      "step": 4292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007601435279050572,
+      "loss": 3.1389,
+      "step": 4293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007600320198586876,
+      "loss": 3.0324,
+      "step": 4294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007599204940817308,
+      "loss": 3.1807,
+      "step": 4295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007598089505817913,
+      "loss": 2.9914,
+      "step": 4296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000759697389366475,
+      "loss": 3.0204,
+      "step": 4297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007595858104433884,
+      "loss": 3.2747,
+      "step": 4298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007594742138201401,
+      "loss": 3.1041,
+      "step": 4299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007593625995043392,
+      "loss": 2.9713,
+      "step": 4300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007592509675035963,
+      "loss": 3.0793,
+      "step": 4301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000759139317825523,
+      "loss": 2.9832,
+      "step": 4302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007590276504777326,
+      "loss": 3.1996,
+      "step": 4303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000758915965467839,
+      "loss": 3.0867,
+      "step": 4304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007588042628034578,
+      "loss": 2.9101,
+      "step": 4305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007586925424922054,
+      "loss": 2.9764,
+      "step": 4306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007585808045416998,
+      "loss": 3.0835,
+      "step": 4307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007584690489595596,
+      "loss": 2.9225,
+      "step": 4308
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007583572757534052,
+      "loss": 2.9449,
+      "step": 4309
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007582454849308579,
+      "loss": 3.1959,
+      "step": 4310
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007581336764995405,
+      "loss": 2.9394,
+      "step": 4311
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007580218504670766,
+      "loss": 2.892,
+      "step": 4312
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007579100068410914,
+      "loss": 2.9093,
+      "step": 4313
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007577981456292108,
+      "loss": 2.9341,
+      "step": 4314
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007576862668390623,
+      "loss": 3.2803,
+      "step": 4315
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007575743704782746,
+      "loss": 3.158,
+      "step": 4316
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007574624565544775,
+      "loss": 2.8359,
+      "step": 4317
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007573505250753014,
+      "loss": 2.9669,
+      "step": 4318
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007572385760483794,
+      "loss": 3.0026,
+      "step": 4319
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007571266094813443,
+      "loss": 3.0515,
+      "step": 4320
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007570146253818308,
+      "loss": 3.1768,
+      "step": 4321
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007569026237574748,
+      "loss": 2.7142,
+      "step": 4322
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000756790604615913,
+      "loss": 3.1066,
+      "step": 4323
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007566785679647836,
+      "loss": 3.2959,
+      "step": 4324
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007565665138117261,
+      "loss": 3.0241,
+      "step": 4325
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007564544421643811,
+      "loss": 3.2346,
+      "step": 4326
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007563423530303901,
+      "loss": 3.1251,
+      "step": 4327
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007562302464173962,
+      "loss": 3.1915,
+      "step": 4328
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007561181223330434,
+      "loss": 2.8316,
+      "step": 4329
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007560059807849771,
+      "loss": 3.0526,
+      "step": 4330
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007558938217808439,
+      "loss": 3.1519,
+      "step": 4331
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007557816453282911,
+      "loss": 3.0586,
+      "step": 4332
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000755669451434968,
+      "loss": 2.9538,
+      "step": 4333
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007555572401085247,
+      "loss": 2.9538,
+      "step": 4334
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000755445011356612,
+      "loss": 2.8797,
+      "step": 4335
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007553327651868828,
+      "loss": 3.078,
+      "step": 4336
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007552205016069906,
+      "loss": 2.9608,
+      "step": 4337
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007551082206245901,
+      "loss": 3.1123,
+      "step": 4338
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007549959222473374,
+      "loss": 3.2075,
+      "step": 4339
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007548836064828897,
+      "loss": 3.117,
+      "step": 4340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007547712733389053,
+      "loss": 3.0404,
+      "step": 4341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007546589228230441,
+      "loss": 2.9253,
+      "step": 4342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007545465549429664,
+      "loss": 3.0647,
+      "step": 4343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007544341697063344,
+      "loss": 3.0856,
+      "step": 4344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007543217671208112,
+      "loss": 3.1459,
+      "step": 4345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007542093471940609,
+      "loss": 3.1397,
+      "step": 4346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007540969099337492,
+      "loss": 2.9089,
+      "step": 4347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007539844553475426,
+      "loss": 3.0476,
+      "step": 4348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007538719834431093,
+      "loss": 2.8469,
+      "step": 4349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000753759494228118,
+      "loss": 2.957,
+      "step": 4350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000753646987710239,
+      "loss": 3.1926,
+      "step": 4351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007535344638971435,
+      "loss": 2.9391,
+      "step": 4352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007534219227965045,
+      "loss": 2.8324,
+      "step": 4353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007533093644159954,
+      "loss": 2.8443,
+      "step": 4354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007531967887632911,
+      "loss": 3.1245,
+      "step": 4355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007530841958460679,
+      "loss": 3.2319,
+      "step": 4356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007529715856720031,
+      "loss": 2.8489,
+      "step": 4357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007528589582487751,
+      "loss": 3.1175,
+      "step": 4358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007527463135840633,
+      "loss": 3.1321,
+      "step": 4359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007526336516855488,
+      "loss": 3.1201,
+      "step": 4360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007525209725609134,
+      "loss": 3.1572,
+      "step": 4361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007524082762178403,
+      "loss": 2.9642,
+      "step": 4362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007522955626640138,
+      "loss": 2.9035,
+      "step": 4363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007521828319071198,
+      "loss": 3.0015,
+      "step": 4364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007520700839548445,
+      "loss": 2.8889,
+      "step": 4365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007519573188148758,
+      "loss": 3.1196,
+      "step": 4366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000751844536494903,
+      "loss": 3.0177,
+      "step": 4367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000751731737002616,
+      "loss": 2.9587,
+      "step": 4368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007516189203457062,
+      "loss": 3.0119,
+      "step": 4369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007515060865318663,
+      "loss": 3.0003,
+      "step": 4370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007513932355687899,
+      "loss": 2.7931,
+      "step": 4371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007512803674641719,
+      "loss": 2.7956,
+      "step": 4372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007511674822257085,
+      "loss": 3.1775,
+      "step": 4373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007510545798610966,
+      "loss": 3.1266,
+      "step": 4374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007509416603780349,
+      "loss": 3.0714,
+      "step": 4375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007508287237842226,
+      "loss": 3.1434,
+      "step": 4376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007507157700873608,
+      "loss": 3.0161,
+      "step": 4377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000750602799295151,
+      "loss": 2.6317,
+      "step": 4378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007504898114152965,
+      "loss": 2.7929,
+      "step": 4379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007503768064555013,
+      "loss": 2.8929,
+      "step": 4380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007502637844234711,
+      "loss": 2.8724,
+      "step": 4381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000750150745326912,
+      "loss": 2.9686,
+      "step": 4382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007500376891735323,
+      "loss": 2.9003,
+      "step": 4383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007499246159710404,
+      "loss": 3.2539,
+      "step": 4384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007498115257271462,
+      "loss": 3.1793,
+      "step": 4385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007496984184495613,
+      "loss": 3.1531,
+      "step": 4386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000749585294145998,
+      "loss": 3.0291,
+      "step": 4387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007494721528241696,
+      "loss": 3.1184,
+      "step": 4388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000749358994491791,
+      "loss": 3.1427,
+      "step": 4389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007492458191565779,
+      "loss": 2.9829,
+      "step": 4390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007491326268262471,
+      "loss": 3.0059,
+      "step": 4391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000749019417508517,
+      "loss": 3.1618,
+      "step": 4392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007489061912111068,
+      "loss": 3.035,
+      "step": 4393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007487929479417372,
+      "loss": 3.1501,
+      "step": 4394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007486796877081296,
+      "loss": 3.2371,
+      "step": 4395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007485664105180067,
+      "loss": 2.9401,
+      "step": 4396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007484531163790928,
+      "loss": 2.9737,
+      "step": 4397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007483398052991128,
+      "loss": 2.9067,
+      "step": 4398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007482264772857928,
+      "loss": 3.2099,
+      "step": 4399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00074811313234686,
+      "loss": 3.0246,
+      "step": 4400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007479997704900437,
+      "loss": 2.9767,
+      "step": 4401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000747886391723073,
+      "loss": 2.9972,
+      "step": 4402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0007477729960536789,
+      "loss": 3.0376,
+      "step": 4403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007476595834895937,
+      "loss": 3.0088,
+      "step": 4404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007475461540385502,
+      "loss": 2.8098,
+      "step": 4405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007474327077082828,
+      "loss": 3.0901,
+      "step": 4406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007473192445065271,
+      "loss": 3.0426,
+      "step": 4407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007472057644410195,
+      "loss": 2.8167,
+      "step": 4408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007470922675194981,
+      "loss": 2.8829,
+      "step": 4409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007469787537497016,
+      "loss": 2.8056,
+      "step": 4410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00074686522313937,
+      "loss": 2.9744,
+      "step": 4411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007467516756962446,
+      "loss": 3.1005,
+      "step": 4412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000746638111428068,
+      "loss": 2.9186,
+      "step": 4413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007465245303425834,
+      "loss": 3.084,
+      "step": 4414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007464109324475355,
+      "loss": 2.9015,
+      "step": 4415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007462973177506702,
+      "loss": 3.0539,
+      "step": 4416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007461836862597345,
+      "loss": 3.0678,
+      "step": 4417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007460700379824763,
+      "loss": 3.1556,
+      "step": 4418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000745956372926645,
+      "loss": 3.0115,
+      "step": 4419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000745842691099991,
+      "loss": 3.0291,
+      "step": 4420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007457289925102655,
+      "loss": 2.9297,
+      "step": 4421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007456152771652217,
+      "loss": 3.0171,
+      "step": 4422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000745501545072613,
+      "loss": 2.9818,
+      "step": 4423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007453877962401946,
+      "loss": 3.0567,
+      "step": 4424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007452740306757226,
+      "loss": 3.0465,
+      "step": 4425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007451602483869539,
+      "loss": 2.8715,
+      "step": 4426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007450464493816472,
+      "loss": 3.0083,
+      "step": 4427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000744932633667562,
+      "loss": 2.9571,
+      "step": 4428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000744818801252459,
+      "loss": 2.9838,
+      "step": 4429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007447049521440996,
+      "loss": 3.1997,
+      "step": 4430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007445910863502472,
+      "loss": 3.21,
+      "step": 4431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007444772038786656,
+      "loss": 2.9983,
+      "step": 4432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007443633047371203,
+      "loss": 3.0381,
+      "step": 4433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007442493889333773,
+      "loss": 3.1472,
+      "step": 4434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007441354564752043,
+      "loss": 2.8424,
+      "step": 4435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007440215073703698,
+      "loss": 3.2764,
+      "step": 4436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007439075416266435,
+      "loss": 3.1519,
+      "step": 4437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007437935592517964,
+      "loss": 2.9321,
+      "step": 4438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007436795602536007,
+      "loss": 2.857,
+      "step": 4439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007435655446398293,
+      "loss": 3.1171,
+      "step": 4440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007434515124182566,
+      "loss": 3.2899,
+      "step": 4441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007433374635966579,
+      "loss": 2.9456,
+      "step": 4442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00074322339818281,
+      "loss": 3.0992,
+      "step": 4443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007431093161844902,
+      "loss": 2.8291,
+      "step": 4444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007429952176094776,
+      "loss": 3.0432,
+      "step": 4445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007428811024655519,
+      "loss": 3.139,
+      "step": 4446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007427669707604946,
+      "loss": 2.9159,
+      "step": 4447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007426528225020876,
+      "loss": 2.991,
+      "step": 4448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007425386576981142,
+      "loss": 3.1792,
+      "step": 4449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000742424476356359,
+      "loss": 3.0906,
+      "step": 4450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007423102784846076,
+      "loss": 3.0102,
+      "step": 4451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007421960640906465,
+      "loss": 3.0661,
+      "step": 4452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007420818331822639,
+      "loss": 3.1917,
+      "step": 4453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007419675857672484,
+      "loss": 3.0856,
+      "step": 4454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007418533218533903,
+      "loss": 3.0015,
+      "step": 4455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007417390414484807,
+      "loss": 2.8249,
+      "step": 4456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007416247445603121,
+      "loss": 2.7918,
+      "step": 4457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007415104311966778,
+      "loss": 2.815,
+      "step": 4458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007413961013653725,
+      "loss": 3.3299,
+      "step": 4459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007412817550741919,
+      "loss": 3.1817,
+      "step": 4460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007411673923309328,
+      "loss": 2.9935,
+      "step": 4461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007410530131433931,
+      "loss": 3.0912,
+      "step": 4462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007409386175193721,
+      "loss": 3.2294,
+      "step": 4463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007408242054666699,
+      "loss": 2.9882,
+      "step": 4464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007407097769930875,
+      "loss": 2.9437,
+      "step": 4465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007405953321064278,
+      "loss": 3.3182,
+      "step": 4466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007404808708144941,
+      "loss": 3.1143,
+      "step": 4467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007403663931250911,
+      "loss": 3.2196,
+      "step": 4468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007402518990460247,
+      "loss": 2.8819,
+      "step": 4469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007401373885851019,
+      "loss": 3.1583,
+      "step": 4470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007400228617501305,
+      "loss": 3.0796,
+      "step": 4471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007399083185489197,
+      "loss": 2.9884,
+      "step": 4472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007397937589892798,
+      "loss": 2.6317,
+      "step": 4473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007396791830790221,
+      "loss": 2.9765,
+      "step": 4474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007395645908259593,
+      "loss": 2.9506,
+      "step": 4475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007394499822379047,
+      "loss": 3.3637,
+      "step": 4476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007393353573226733,
+      "loss": 3.0842,
+      "step": 4477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007392207160880808,
+      "loss": 2.9294,
+      "step": 4478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007391060585419442,
+      "loss": 2.973,
+      "step": 4479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007389913846920815,
+      "loss": 2.9076,
+      "step": 4480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007388766945463119,
+      "loss": 3.0201,
+      "step": 4481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007387619881124558,
+      "loss": 3.127,
+      "step": 4482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007386472653983342,
+      "loss": 3.0395,
+      "step": 4483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00073853252641177,
+      "loss": 3.0358,
+      "step": 4484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007384177711605869,
+      "loss": 2.7679,
+      "step": 4485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007383029996526092,
+      "loss": 2.8851,
+      "step": 4486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007381882118956629,
+      "loss": 3.1337,
+      "step": 4487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007380734078975752,
+      "loss": 3.0541,
+      "step": 4488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007379585876661738,
+      "loss": 2.9451,
+      "step": 4489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007378437512092878,
+      "loss": 2.9835,
+      "step": 4490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007377288985347478,
+      "loss": 3.1004,
+      "step": 4491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000737614029650385,
+      "loss": 3.1589,
+      "step": 4492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007374991445640318,
+      "loss": 3.1773,
+      "step": 4493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007373842432835218,
+      "loss": 3.2587,
+      "step": 4494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007372693258166898,
+      "loss": 2.9399,
+      "step": 4495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007371543921713715,
+      "loss": 3.2431,
+      "step": 4496
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007370394423554038,
+      "loss": 3.0406,
+      "step": 4497
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007369244763766245,
+      "loss": 2.9985,
+      "step": 4498
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000736809494242873,
+      "loss": 3.1702,
+      "step": 4499
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007366944959619893,
+      "loss": 3.1185,
+      "step": 4500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007365794815418149,
+      "loss": 2.9941,
+      "step": 4501
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007364644509901918,
+      "loss": 3.0932,
+      "step": 4502
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007363494043149639,
+      "loss": 2.9534,
+      "step": 4503
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007362343415239757,
+      "loss": 2.9507,
+      "step": 4504
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007361192626250725,
+      "loss": 3.1893,
+      "step": 4505
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007360041676261015,
+      "loss": 2.955,
+      "step": 4506
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007358890565349105,
+      "loss": 2.9613,
+      "step": 4507
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007357739293593486,
+      "loss": 3.1707,
+      "step": 4508
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007356587861072657,
+      "loss": 3.1526,
+      "step": 4509
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000735543626786513,
+      "loss": 2.9448,
+      "step": 4510
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007354284514049428,
+      "loss": 3.2426,
+      "step": 4511
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007353132599704085,
+      "loss": 3.0678,
+      "step": 4512
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007351980524907645,
+      "loss": 3.0659,
+      "step": 4513
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007350828289738664,
+      "loss": 3.0564,
+      "step": 4514
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007349675894275708,
+      "loss": 2.7898,
+      "step": 4515
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007348523338597355,
+      "loss": 3.0767,
+      "step": 4516
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007347370622782193,
+      "loss": 2.805,
+      "step": 4517
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007346217746908823,
+      "loss": 2.967,
+      "step": 4518
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007345064711055852,
+      "loss": 3.1892,
+      "step": 4519
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007343911515301903,
+      "loss": 3.3526,
+      "step": 4520
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007342758159725608,
+      "loss": 2.9923,
+      "step": 4521
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007341604644405611,
+      "loss": 3.2306,
+      "step": 4522
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007340450969420563,
+      "loss": 2.9819,
+      "step": 4523
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.000733929713484913,
+      "loss": 2.9616,
+      "step": 4524
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007338143140769988,
+      "loss": 2.9824,
+      "step": 4525
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007336988987261822,
+      "loss": 3.0389,
+      "step": 4526
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0007335834674403332,
+      "loss": 3.2821,
+      "step": 4527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007334680202273222,
+      "loss": 3.1343,
+      "step": 4528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007333525570950216,
+      "loss": 2.9484,
+      "step": 4529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007332370780513042,
+      "loss": 3.0097,
+      "step": 4530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007331215831040439,
+      "loss": 3.0137,
+      "step": 4531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007330060722611161,
+      "loss": 3.0815,
+      "step": 4532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007328905455303968,
+      "loss": 3.0507,
+      "step": 4533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007327750029197635,
+      "loss": 3.0236,
+      "step": 4534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007326594444370945,
+      "loss": 2.9939,
+      "step": 4535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007325438700902694,
+      "loss": 3.2058,
+      "step": 4536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007324282798871687,
+      "loss": 3.0537,
+      "step": 4537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000732312673835674,
+      "loss": 3.0118,
+      "step": 4538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007321970519436683,
+      "loss": 3.0447,
+      "step": 4539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000732081414219035,
+      "loss": 3.0136,
+      "step": 4540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007319657606696593,
+      "loss": 2.9683,
+      "step": 4541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007318500913034272,
+      "loss": 3.1006,
+      "step": 4542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007317344061282255,
+      "loss": 3.058,
+      "step": 4543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007316187051519424,
+      "loss": 3.2111,
+      "step": 4544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007315029883824673,
+      "loss": 3.2343,
+      "step": 4545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007313872558276903,
+      "loss": 3.1542,
+      "step": 4546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007312715074955029,
+      "loss": 2.9899,
+      "step": 4547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007311557433937973,
+      "loss": 3.1304,
+      "step": 4548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007310399635304673,
+      "loss": 3.086,
+      "step": 4549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007309241679134073,
+      "loss": 3.1718,
+      "step": 4550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007308083565505131,
+      "loss": 3.1281,
+      "step": 4551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000730692529449681,
+      "loss": 2.902,
+      "step": 4552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007305766866188095,
+      "loss": 3.1253,
+      "step": 4553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007304608280657971,
+      "loss": 3.1525,
+      "step": 4554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007303449537985436,
+      "loss": 3.309,
+      "step": 4555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007302290638249502,
+      "loss": 2.9297,
+      "step": 4556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007301131581529193,
+      "loss": 2.9912,
+      "step": 4557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007299972367903533,
+      "loss": 3.0228,
+      "step": 4558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007298812997451571,
+      "loss": 2.9228,
+      "step": 4559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007297653470252358,
+      "loss": 3.0709,
+      "step": 4560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007296493786384957,
+      "loss": 3.0933,
+      "step": 4561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007295333945928444,
+      "loss": 2.9783,
+      "step": 4562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007294173948961903,
+      "loss": 3.0778,
+      "step": 4563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007293013795564429,
+      "loss": 3.1879,
+      "step": 4564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000729185348581513,
+      "loss": 2.9856,
+      "step": 4565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000729069301979312,
+      "loss": 3.1293,
+      "step": 4566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007289532397577532,
+      "loss": 3.2617,
+      "step": 4567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007288371619247502,
+      "loss": 2.9128,
+      "step": 4568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007287210684882176,
+      "loss": 2.9041,
+      "step": 4569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007286049594560718,
+      "loss": 3.1447,
+      "step": 4570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007284888348362296,
+      "loss": 2.8393,
+      "step": 4571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007283726946366089,
+      "loss": 2.904,
+      "step": 4572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007282565388651294,
+      "loss": 3.0051,
+      "step": 4573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007281403675297108,
+      "loss": 2.9169,
+      "step": 4574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007280241806382747,
+      "loss": 3.0312,
+      "step": 4575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007279079781987433,
+      "loss": 3.1146,
+      "step": 4576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00072779176021904,
+      "loss": 3.1138,
+      "step": 4577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007276755267070894,
+      "loss": 2.9815,
+      "step": 4578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007275592776708169,
+      "loss": 2.9756,
+      "step": 4579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007274430131181489,
+      "loss": 3.3328,
+      "step": 4580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007273267330570133,
+      "loss": 3.2104,
+      "step": 4581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007272104374953386,
+      "loss": 3.0173,
+      "step": 4582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007270941264410549,
+      "loss": 3.0008,
+      "step": 4583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007269777999020926,
+      "loss": 2.9597,
+      "step": 4584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007268614578863837,
+      "loss": 3.0652,
+      "step": 4585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007267451004018614,
+      "loss": 3.151,
+      "step": 4586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007266287274564591,
+      "loss": 3.0802,
+      "step": 4587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007265123390581122,
+      "loss": 2.8914,
+      "step": 4588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007263959352147568,
+      "loss": 3.0822,
+      "step": 4589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007262795159343298,
+      "loss": 3.1261,
+      "step": 4590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007261630812247697,
+      "loss": 2.9231,
+      "step": 4591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007260466310940155,
+      "loss": 2.9315,
+      "step": 4592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007259301655500075,
+      "loss": 3.03,
+      "step": 4593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007258136846006871,
+      "loss": 2.8248,
+      "step": 4594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007256971882539967,
+      "loss": 2.9919,
+      "step": 4595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007255806765178797,
+      "loss": 3.188,
+      "step": 4596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007254641494002806,
+      "loss": 2.7471,
+      "step": 4597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000725347606909145,
+      "loss": 3.1008,
+      "step": 4598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007252310490524195,
+      "loss": 3.0311,
+      "step": 4599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007251144758380516,
+      "loss": 3.0965,
+      "step": 4600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00072499788727399,
+      "loss": 3.1662,
+      "step": 4601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007248812833681845,
+      "loss": 3.1484,
+      "step": 4602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007247646641285859,
+      "loss": 3.1795,
+      "step": 4603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007246480295631458,
+      "loss": 3.0642,
+      "step": 4604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007245313796798174,
+      "loss": 2.9365,
+      "step": 4605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007244147144865545,
+      "loss": 3.1099,
+      "step": 4606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007242980339913119,
+      "loss": 3.1525,
+      "step": 4607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007241813382020456,
+      "loss": 3.0862,
+      "step": 4608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007240646271267128,
+      "loss": 3.0754,
+      "step": 4609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007239479007732717,
+      "loss": 3.1826,
+      "step": 4610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000723831159149681,
+      "loss": 3.1685,
+      "step": 4611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007237144022639011,
+      "loss": 3.0296,
+      "step": 4612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007235976301238933,
+      "loss": 2.7606,
+      "step": 4613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007234808427376197,
+      "loss": 3.0508,
+      "step": 4614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007233640401130436,
+      "loss": 2.9982,
+      "step": 4615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007232472222581294,
+      "loss": 3.1884,
+      "step": 4616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007231303891808425,
+      "loss": 2.9874,
+      "step": 4617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007230135408891491,
+      "loss": 3.27,
+      "step": 4618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007228966773910169,
+      "loss": 2.7435,
+      "step": 4619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000722779798694414,
+      "loss": 3.0507,
+      "step": 4620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007226629048073103,
+      "loss": 3.0255,
+      "step": 4621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007225459957376762,
+      "loss": 2.9919,
+      "step": 4622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007224290714934831,
+      "loss": 2.9445,
+      "step": 4623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000722312132082704,
+      "loss": 2.9952,
+      "step": 4624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007221951775133122,
+      "loss": 2.9694,
+      "step": 4625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007220782077932825,
+      "loss": 3.1842,
+      "step": 4626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007219612229305908,
+      "loss": 3.0858,
+      "step": 4627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007218442229332134,
+      "loss": 3.0074,
+      "step": 4628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007217272078091285,
+      "loss": 3.2368,
+      "step": 4629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007216101775663149,
+      "loss": 3.0248,
+      "step": 4630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007214931322127523,
+      "loss": 3.0901,
+      "step": 4631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007213760717564215,
+      "loss": 3.1721,
+      "step": 4632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007212589962053046,
+      "loss": 3.1249,
+      "step": 4633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007211419055673842,
+      "loss": 2.8878,
+      "step": 4634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007210247998506447,
+      "loss": 3.0994,
+      "step": 4635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007209076790630709,
+      "loss": 3.0671,
+      "step": 4636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007207905432126487,
+      "loss": 3.0554,
+      "step": 4637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007206733923073653,
+      "loss": 2.8931,
+      "step": 4638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007205562263552087,
+      "loss": 3.2606,
+      "step": 4639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007204390453641681,
+      "loss": 2.9892,
+      "step": 4640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007203218493422335,
+      "loss": 2.725,
+      "step": 4641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.000720204638297396,
+      "loss": 2.9883,
+      "step": 4642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007200874122376479,
+      "loss": 3.0917,
+      "step": 4643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007199701711709823,
+      "loss": 3.0266,
+      "step": 4644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007198529151053937,
+      "loss": 3.04,
+      "step": 4645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007197356440488769,
+      "loss": 2.7899,
+      "step": 4646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007196183580094284,
+      "loss": 3.0871,
+      "step": 4647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007195010569950456,
+      "loss": 3.0977,
+      "step": 4648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007193837410137265,
+      "loss": 3.027,
+      "step": 4649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007192664100734707,
+      "loss": 3.0192,
+      "step": 4650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0007191490641822784,
+      "loss": 3.0573,
+      "step": 4651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007190317033481513,
+      "loss": 2.9713,
+      "step": 4652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007189143275790912,
+      "loss": 2.9228,
+      "step": 4653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007187969368831019,
+      "loss": 2.8532,
+      "step": 4654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007186795312681877,
+      "loss": 2.7512,
+      "step": 4655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000718562110742354,
+      "loss": 3.0092,
+      "step": 4656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007184446753136074,
+      "loss": 2.9959,
+      "step": 4657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007183272249899553,
+      "loss": 2.9927,
+      "step": 4658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007182097597794062,
+      "loss": 2.9901,
+      "step": 4659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007180922796899695,
+      "loss": 3.101,
+      "step": 4660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007179747847296558,
+      "loss": 3.0604,
+      "step": 4661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007178572749064767,
+      "loss": 3.1311,
+      "step": 4662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007177397502284446,
+      "loss": 3.0905,
+      "step": 4663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007176222107035733,
+      "loss": 3.0039,
+      "step": 4664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007175046563398771,
+      "loss": 2.9698,
+      "step": 4665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007173870871453716,
+      "loss": 3.1953,
+      "step": 4666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007172695031280736,
+      "loss": 3.3176,
+      "step": 4667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007171519042960006,
+      "loss": 3.1682,
+      "step": 4668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000717034290657171,
+      "loss": 3.0536,
+      "step": 4669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007169166622196048,
+      "loss": 3.1598,
+      "step": 4670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007167990189913224,
+      "loss": 2.9649,
+      "step": 4671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007166813609803454,
+      "loss": 2.8763,
+      "step": 4672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007165636881946967,
+      "loss": 3.0141,
+      "step": 4673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007164460006423997,
+      "loss": 3.167,
+      "step": 4674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007163282983314793,
+      "loss": 3.2651,
+      "step": 4675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007162105812699609,
+      "loss": 2.9611,
+      "step": 4676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007160928494658714,
+      "loss": 2.8116,
+      "step": 4677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007159751029272383,
+      "loss": 3.1086,
+      "step": 4678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007158573416620903,
+      "loss": 3.1112,
+      "step": 4679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007157395656784574,
+      "loss": 2.9332,
+      "step": 4680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007156217749843699,
+      "loss": 3.1526,
+      "step": 4681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007155039695878596,
+      "loss": 2.8078,
+      "step": 4682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007153861494969593,
+      "loss": 3.2787,
+      "step": 4683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007152683147197028,
+      "loss": 3.103,
+      "step": 4684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007151504652641245,
+      "loss": 3.0382,
+      "step": 4685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007150326011382603,
+      "loss": 2.8538,
+      "step": 4686
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007149147223501469,
+      "loss": 3.2354,
+      "step": 4687
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007147968289078218,
+      "loss": 2.8991,
+      "step": 4688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007146789208193241,
+      "loss": 3.0218,
+      "step": 4689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007145609980926931,
+      "loss": 2.9186,
+      "step": 4690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007144430607359697,
+      "loss": 2.947,
+      "step": 4691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007143251087571956,
+      "loss": 2.9841,
+      "step": 4692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007142071421644134,
+      "loss": 2.9935,
+      "step": 4693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007140891609656668,
+      "loss": 3.38,
+      "step": 4694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007139711651690005,
+      "loss": 3.1568,
+      "step": 4695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007138531547824604,
+      "loss": 2.9053,
+      "step": 4696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007137351298140928,
+      "loss": 3.0499,
+      "step": 4697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007136170902719457,
+      "loss": 3.0059,
+      "step": 4698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007134990361640677,
+      "loss": 2.9856,
+      "step": 4699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007133809674985081,
+      "loss": 3.0619,
+      "step": 4700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007132628842833181,
+      "loss": 2.9322,
+      "step": 4701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007131447865265488,
+      "loss": 3.0515,
+      "step": 4702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007130266742362533,
+      "loss": 3.1155,
+      "step": 4703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000712908547420485,
+      "loss": 3.1871,
+      "step": 4704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007127904060872985,
+      "loss": 3.0247,
+      "step": 4705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007126722502447495,
+      "loss": 3.1005,
+      "step": 4706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007125540799008945,
+      "loss": 2.9014,
+      "step": 4707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007124358950637913,
+      "loss": 2.9887,
+      "step": 4708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000712317695741498,
+      "loss": 3.0476,
+      "step": 4709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007121994819420745,
+      "loss": 3.0862,
+      "step": 4710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007120812536735814,
+      "loss": 3.0287,
+      "step": 4711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007119630109440802,
+      "loss": 3.2308,
+      "step": 4712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007118447537616334,
+      "loss": 2.914,
+      "step": 4713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007117264821343044,
+      "loss": 3.1652,
+      "step": 4714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007116081960701577,
+      "loss": 3.188,
+      "step": 4715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007114898955772589,
+      "loss": 3.1053,
+      "step": 4716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007113715806636743,
+      "loss": 2.8853,
+      "step": 4717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007112532513374714,
+      "loss": 3.092,
+      "step": 4718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007111349076067186,
+      "loss": 2.9842,
+      "step": 4719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007110165494794854,
+      "loss": 3.0923,
+      "step": 4720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007108981769638422,
+      "loss": 3.2867,
+      "step": 4721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007107797900678603,
+      "loss": 2.9921,
+      "step": 4722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007106613887996121,
+      "loss": 3.0474,
+      "step": 4723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007105429731671706,
+      "loss": 3.0872,
+      "step": 4724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007104245431786106,
+      "loss": 3.0513,
+      "step": 4725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007103060988420071,
+      "loss": 3.3085,
+      "step": 4726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007101876401654365,
+      "loss": 3.082,
+      "step": 4727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007100691671569761,
+      "loss": 3.121,
+      "step": 4728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007099506798247039,
+      "loss": 3.0399,
+      "step": 4729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007098321781766992,
+      "loss": 2.9949,
+      "step": 4730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007097136622210422,
+      "loss": 3.0361,
+      "step": 4731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007095951319658139,
+      "loss": 2.8419,
+      "step": 4732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007094765874190966,
+      "loss": 2.8471,
+      "step": 4733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007093580285889735,
+      "loss": 2.8514,
+      "step": 4734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007092394554835285,
+      "loss": 2.9431,
+      "step": 4735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007091208681108465,
+      "loss": 3.1434,
+      "step": 4736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007090022664790138,
+      "loss": 3.0849,
+      "step": 4737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007088836505961172,
+      "loss": 2.9447,
+      "step": 4738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007087650204702447,
+      "loss": 2.9176,
+      "step": 4739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007086463761094854,
+      "loss": 2.907,
+      "step": 4740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000708527717521929,
+      "loss": 3.0211,
+      "step": 4741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007084090447156666,
+      "loss": 3.0277,
+      "step": 4742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007082903576987898,
+      "loss": 2.9839,
+      "step": 4743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007081716564793914,
+      "loss": 3.0476,
+      "step": 4744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007080529410655653,
+      "loss": 2.9644,
+      "step": 4745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007079342114654065,
+      "loss": 2.8524,
+      "step": 4746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007078154676870101,
+      "loss": 3.0907,
+      "step": 4747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007076967097384735,
+      "loss": 2.9326,
+      "step": 4748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007075779376278937,
+      "loss": 3.0763,
+      "step": 4749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007074591513633698,
+      "loss": 2.9643,
+      "step": 4750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007073403509530012,
+      "loss": 3.2976,
+      "step": 4751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007072215364048882,
+      "loss": 3.2585,
+      "step": 4752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007071027077271328,
+      "loss": 2.9218,
+      "step": 4753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007069838649278369,
+      "loss": 3.0334,
+      "step": 4754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007068650080151044,
+      "loss": 2.8636,
+      "step": 4755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007067461369970396,
+      "loss": 2.976,
+      "step": 4756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007066272518817478,
+      "loss": 3.1896,
+      "step": 4757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007065083526773353,
+      "loss": 2.9268,
+      "step": 4758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007063894393919094,
+      "loss": 3.0417,
+      "step": 4759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007062705120335784,
+      "loss": 3.1058,
+      "step": 4760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007061515706104513,
+      "loss": 3.0849,
+      "step": 4761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007060326151306385,
+      "loss": 3.0498,
+      "step": 4762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000705913645602251,
+      "loss": 2.959,
+      "step": 4763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007057946620334009,
+      "loss": 3.0778,
+      "step": 4764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007056756644322012,
+      "loss": 3.2253,
+      "step": 4765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007055566528067661,
+      "loss": 3.0151,
+      "step": 4766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007054376271652102,
+      "loss": 2.9468,
+      "step": 4767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007053185875156495,
+      "loss": 3.0486,
+      "step": 4768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.000705199533866201,
+      "loss": 2.7748,
+      "step": 4769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007050804662249825,
+      "loss": 3.1401,
+      "step": 4770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007049613846001129,
+      "loss": 3.0974,
+      "step": 4771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007048422889997114,
+      "loss": 3.1233,
+      "step": 4772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007047231794318991,
+      "loss": 2.9666,
+      "step": 4773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007046040559047977,
+      "loss": 3.0006,
+      "step": 4774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0007044849184265294,
+      "loss": 3.0457,
+      "step": 4775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000704365767005218,
+      "loss": 2.8149,
+      "step": 4776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007042466016489879,
+      "loss": 3.1334,
+      "step": 4777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007041274223659646,
+      "loss": 3.1007,
+      "step": 4778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007040082291642744,
+      "loss": 3.0027,
+      "step": 4779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007038890220520446,
+      "loss": 3.1895,
+      "step": 4780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007037698010374035,
+      "loss": 2.7915,
+      "step": 4781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007036505661284805,
+      "loss": 2.9736,
+      "step": 4782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007035313173334056,
+      "loss": 2.8672,
+      "step": 4783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007034120546603097,
+      "loss": 3.3315,
+      "step": 4784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007032927781173252,
+      "loss": 3.1469,
+      "step": 4785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007031734877125852,
+      "loss": 2.8231,
+      "step": 4786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007030541834542233,
+      "loss": 3.012,
+      "step": 4787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007029348653503747,
+      "loss": 2.98,
+      "step": 4788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000702815533409175,
+      "loss": 2.8958,
+      "step": 4789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007026961876387612,
+      "loss": 3.1854,
+      "step": 4790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007025768280472708,
+      "loss": 3.2371,
+      "step": 4791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007024574546428427,
+      "loss": 3.0892,
+      "step": 4792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007023380674336163,
+      "loss": 2.6971,
+      "step": 4793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007022186664277323,
+      "loss": 2.8955,
+      "step": 4794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007020992516333323,
+      "loss": 3.0963,
+      "step": 4795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007019798230585585,
+      "loss": 3.007,
+      "step": 4796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007018603807115545,
+      "loss": 2.9266,
+      "step": 4797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007017409246004645,
+      "loss": 2.9531,
+      "step": 4798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007016214547334337,
+      "loss": 2.9941,
+      "step": 4799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007015019711186083,
+      "loss": 2.9917,
+      "step": 4800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007013824737641355,
+      "loss": 3.1595,
+      "step": 4801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007012629626781634,
+      "loss": 3.0228,
+      "step": 4802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000701143437868841,
+      "loss": 3.0238,
+      "step": 4803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007010238993443181,
+      "loss": 2.963,
+      "step": 4804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007009043471127458,
+      "loss": 3.2366,
+      "step": 4805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007007847811822756,
+      "loss": 2.9821,
+      "step": 4806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007006652015610604,
+      "loss": 2.959,
+      "step": 4807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000700545608257254,
+      "loss": 3.0472,
+      "step": 4808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000700426001279011,
+      "loss": 2.9297,
+      "step": 4809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007003063806344866,
+      "loss": 2.8793,
+      "step": 4810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007001867463318377,
+      "loss": 3.0466,
+      "step": 4811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0007000670983792214,
+      "loss": 3.0051,
+      "step": 4812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006999474367847962,
+      "loss": 3.1552,
+      "step": 4813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006998277615567214,
+      "loss": 2.8374,
+      "step": 4814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000699708072703157,
+      "loss": 3.1881,
+      "step": 4815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006995883702322645,
+      "loss": 2.9463,
+      "step": 4816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006994686541522054,
+      "loss": 3.0049,
+      "step": 4817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006993489244711431,
+      "loss": 2.9978,
+      "step": 4818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006992291811972413,
+      "loss": 2.9906,
+      "step": 4819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000699109424338665,
+      "loss": 3.0369,
+      "step": 4820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006989896539035798,
+      "loss": 2.9605,
+      "step": 4821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006988698699001525,
+      "loss": 3.0067,
+      "step": 4822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006987500723365507,
+      "loss": 2.8779,
+      "step": 4823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000698630261220943,
+      "loss": 3.0341,
+      "step": 4824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006985104365614987,
+      "loss": 2.9077,
+      "step": 4825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006983905983663881,
+      "loss": 3.0456,
+      "step": 4826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000698270746643783,
+      "loss": 3.0621,
+      "step": 4827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006981508814018551,
+      "loss": 2.9642,
+      "step": 4828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006980310026487777,
+      "loss": 3.1202,
+      "step": 4829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000697911110392725,
+      "loss": 2.9904,
+      "step": 4830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006977912046418719,
+      "loss": 3.0205,
+      "step": 4831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006976712854043944,
+      "loss": 3.0583,
+      "step": 4832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006975513526884692,
+      "loss": 3.1921,
+      "step": 4833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006974314065022742,
+      "loss": 2.9271,
+      "step": 4834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006973114468539879,
+      "loss": 3.1352,
+      "step": 4835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00069719147375179,
+      "loss": 3.0494,
+      "step": 4836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006970714872038609,
+      "loss": 3.0286,
+      "step": 4837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006969514872183822,
+      "loss": 3.0384,
+      "step": 4838
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006968314738035362,
+      "loss": 3.0035,
+      "step": 4839
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000696711446967506,
+      "loss": 2.993,
+      "step": 4840
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006965914067184758,
+      "loss": 3.1295,
+      "step": 4841
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000696471353064631,
+      "loss": 3.1241,
+      "step": 4842
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006963512860141572,
+      "loss": 2.9468,
+      "step": 4843
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006962312055752415,
+      "loss": 2.8273,
+      "step": 4844
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006961111117560715,
+      "loss": 2.8577,
+      "step": 4845
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006959910045648363,
+      "loss": 3.0831,
+      "step": 4846
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006958708840097255,
+      "loss": 3.0688,
+      "step": 4847
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006957507500989294,
+      "loss": 2.8675,
+      "step": 4848
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006956306028406396,
+      "loss": 3.1459,
+      "step": 4849
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006955104422430486,
+      "loss": 3.001,
+      "step": 4850
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006953902683143495,
+      "loss": 2.9794,
+      "step": 4851
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006952700810627365,
+      "loss": 3.2063,
+      "step": 4852
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006951498804964049,
+      "loss": 2.9231,
+      "step": 4853
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006950296666235505,
+      "loss": 3.1396,
+      "step": 4854
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006949094394523705,
+      "loss": 2.6312,
+      "step": 4855
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006947891989910624,
+      "loss": 3.0512,
+      "step": 4856
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006946689452478252,
+      "loss": 3.0899,
+      "step": 4857
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006945486782308583,
+      "loss": 3.2849,
+      "step": 4858
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006944283979483624,
+      "loss": 3.2884,
+      "step": 4859
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006943081044085391,
+      "loss": 3.0492,
+      "step": 4860
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006941877976195905,
+      "loss": 2.9704,
+      "step": 4861
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00069406747758972,
+      "loss": 2.9084,
+      "step": 4862
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006939471443271317,
+      "loss": 3.0173,
+      "step": 4863
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006938267978400305,
+      "loss": 2.9353,
+      "step": 4864
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006937064381366228,
+      "loss": 2.9786,
+      "step": 4865
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006935860652251152,
+      "loss": 2.9741,
+      "step": 4866
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006934656791137151,
+      "loss": 2.7236,
+      "step": 4867
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006933452798106319,
+      "loss": 2.9904,
+      "step": 4868
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006932248673240747,
+      "loss": 2.9962,
+      "step": 4869
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006931044416622541,
+      "loss": 3.014,
+      "step": 4870
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006929840028333815,
+      "loss": 3.1058,
+      "step": 4871
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006928635508456692,
+      "loss": 3.1132,
+      "step": 4872
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00069274308570733,
+      "loss": 3.2188,
+      "step": 4873
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006926226074265785,
+      "loss": 3.0027,
+      "step": 4874
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000692502116011629,
+      "loss": 2.9932,
+      "step": 4875
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000692381611470698,
+      "loss": 3.184,
+      "step": 4876
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006922610938120019,
+      "loss": 2.8156,
+      "step": 4877
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006921405630437584,
+      "loss": 2.8376,
+      "step": 4878
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006920200191741861,
+      "loss": 2.8467,
+      "step": 4879
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006918994622115043,
+      "loss": 2.9399,
+      "step": 4880
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006917788921639334,
+      "loss": 2.9575,
+      "step": 4881
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006916583090396945,
+      "loss": 2.9864,
+      "step": 4882
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006915377128470098,
+      "loss": 2.8564,
+      "step": 4883
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006914171035941024,
+      "loss": 2.8599,
+      "step": 4884
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000691296481289196,
+      "loss": 2.9303,
+      "step": 4885
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006911758459405154,
+      "loss": 2.9274,
+      "step": 4886
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006910551975562863,
+      "loss": 2.9178,
+      "step": 4887
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006909345361447353,
+      "loss": 2.961,
+      "step": 4888
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006908138617140897,
+      "loss": 3.0569,
+      "step": 4889
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.000690693174272578,
+      "loss": 3.1063,
+      "step": 4890
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006905724738284293,
+      "loss": 3.2177,
+      "step": 4891
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006904517603898737,
+      "loss": 3.0245,
+      "step": 4892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006903310339651422,
+      "loss": 2.9787,
+      "step": 4893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006902102945624668,
+      "loss": 2.8804,
+      "step": 4894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00069008954219008,
+      "loss": 2.8161,
+      "step": 4895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006899687768562158,
+      "loss": 3.1963,
+      "step": 4896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006898479985691082,
+      "loss": 2.9784,
+      "step": 4897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006897272073369932,
+      "loss": 3.0654,
+      "step": 4898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0006896064031681066,
+      "loss": 3.0317,
+      "step": 4899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006894855860706859,
+      "loss": 3.0776,
+      "step": 4900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000689364756052969,
+      "loss": 3.1024,
+      "step": 4901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006892439131231949,
+      "loss": 3.175,
+      "step": 4902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006891230572896034,
+      "loss": 2.9898,
+      "step": 4903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006890021885604353,
+      "loss": 3.0308,
+      "step": 4904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006888813069439318,
+      "loss": 2.884,
+      "step": 4905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006887604124483358,
+      "loss": 2.6994,
+      "step": 4906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006886395050818905,
+      "loss": 3.1526,
+      "step": 4907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006885185848528399,
+      "loss": 2.8214,
+      "step": 4908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006883976517694294,
+      "loss": 3.1579,
+      "step": 4909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006882767058399048,
+      "loss": 3.0739,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000688155747072513,
+      "loss": 2.9627,
+      "step": 4911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006880347754755015,
+      "loss": 3.2632,
+      "step": 4912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006879137910571191,
+      "loss": 3.1181,
+      "step": 4913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006877927938256154,
+      "loss": 2.9642,
+      "step": 4914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006876717837892404,
+      "loss": 3.0319,
+      "step": 4915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006875507609562455,
+      "loss": 3.064,
+      "step": 4916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006874297253348827,
+      "loss": 3.1237,
+      "step": 4917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000687308676933405,
+      "loss": 3.0154,
+      "step": 4918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006871876157600663,
+      "loss": 3.0073,
+      "step": 4919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006870665418231212,
+      "loss": 2.8166,
+      "step": 4920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006869454551308253,
+      "loss": 2.9985,
+      "step": 4921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006868243556914351,
+      "loss": 3.121,
+      "step": 4922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006867032435132076,
+      "loss": 2.9544,
+      "step": 4923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006865821186044015,
+      "loss": 3.2412,
+      "step": 4924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006864609809732755,
+      "loss": 3.1167,
+      "step": 4925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006863398306280896,
+      "loss": 3.0265,
+      "step": 4926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006862186675771045,
+      "loss": 2.93,
+      "step": 4927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006860974918285819,
+      "loss": 3.1691,
+      "step": 4928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006859763033907843,
+      "loss": 2.988,
+      "step": 4929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006858551022719751,
+      "loss": 2.9958,
+      "step": 4930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006857338884804185,
+      "loss": 3.1225,
+      "step": 4931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006856126620243797,
+      "loss": 3.1398,
+      "step": 4932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006854914229121245,
+      "loss": 2.7836,
+      "step": 4933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006853701711519199,
+      "loss": 2.984,
+      "step": 4934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006852489067520334,
+      "loss": 3.1536,
+      "step": 4935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006851276297207338,
+      "loss": 2.9635,
+      "step": 4936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006850063400662903,
+      "loss": 3.0298,
+      "step": 4937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006848850377969733,
+      "loss": 3.1628,
+      "step": 4938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000684763722921054,
+      "loss": 3.09,
+      "step": 4939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006846423954468042,
+      "loss": 2.9807,
+      "step": 4940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006845210553824969,
+      "loss": 2.9171,
+      "step": 4941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006843997027364058,
+      "loss": 3.0418,
+      "step": 4942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006842783375168052,
+      "loss": 2.8543,
+      "step": 4943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000684156959731971,
+      "loss": 3.2656,
+      "step": 4944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006840355693901791,
+      "loss": 2.8736,
+      "step": 4945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006839141664997069,
+      "loss": 2.7372,
+      "step": 4946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006837927510688325,
+      "loss": 3.0626,
+      "step": 4947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006836713231058344,
+      "loss": 3.0891,
+      "step": 4948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006835498826189923,
+      "loss": 2.8714,
+      "step": 4949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000683428429616587,
+      "loss": 2.8989,
+      "step": 4950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006833069641068998,
+      "loss": 3.1023,
+      "step": 4951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006831854860982132,
+      "loss": 2.919,
+      "step": 4952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006830639955988099,
+      "loss": 2.7406,
+      "step": 4953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006829424926169742,
+      "loss": 3.2714,
+      "step": 4954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006828209771609908,
+      "loss": 2.7884,
+      "step": 4955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006826994492391454,
+      "loss": 2.9358,
+      "step": 4956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006825779088597244,
+      "loss": 3.0094,
+      "step": 4957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006824563560310154,
+      "loss": 3.2332,
+      "step": 4958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006823347907613064,
+      "loss": 2.9992,
+      "step": 4959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006822132130588867,
+      "loss": 3.0176,
+      "step": 4960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006820916229320462,
+      "loss": 2.9406,
+      "step": 4961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006819700203890753,
+      "loss": 2.9886,
+      "step": 4962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006818484054382661,
+      "loss": 3.1378,
+      "step": 4963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006817267780879106,
+      "loss": 2.8496,
+      "step": 4964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006816051383463025,
+      "loss": 3.0897,
+      "step": 4965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006814834862217356,
+      "loss": 3.0246,
+      "step": 4966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006813618217225053,
+      "loss": 3.1696,
+      "step": 4967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006812401448569071,
+      "loss": 3.0441,
+      "step": 4968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006811184556332378,
+      "loss": 3.1701,
+      "step": 4969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000680996754059795,
+      "loss": 3.0654,
+      "step": 4970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006808750401448768,
+      "loss": 3.1236,
+      "step": 4971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006807533138967828,
+      "loss": 2.8227,
+      "step": 4972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006806315753238125,
+      "loss": 3.0332,
+      "step": 4973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006805098244342673,
+      "loss": 2.9961,
+      "step": 4974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006803880612364488,
+      "loss": 2.8337,
+      "step": 4975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006802662857386594,
+      "loss": 2.9191,
+      "step": 4976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006801444979492026,
+      "loss": 3.1915,
+      "step": 4977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006800226978763828,
+      "loss": 3.138,
+      "step": 4978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006799008855285048,
+      "loss": 3.0722,
+      "step": 4979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006797790609138748,
+      "loss": 3.0414,
+      "step": 4980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006796572240407992,
+      "loss": 2.7315,
+      "step": 4981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000679535374917586,
+      "loss": 3.0152,
+      "step": 4982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006794135135525433,
+      "loss": 3.1599,
+      "step": 4983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006792916399539805,
+      "loss": 2.9746,
+      "step": 4984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006791697541302077,
+      "loss": 3.1557,
+      "step": 4985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006790478560895356,
+      "loss": 2.8861,
+      "step": 4986
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006789259458402764,
+      "loss": 2.9474,
+      "step": 4987
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006788040233907423,
+      "loss": 2.9869,
+      "step": 4988
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000678682088749247,
+      "loss": 2.896,
+      "step": 4989
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006785601419241046,
+      "loss": 3.1123,
+      "step": 4990
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006784381829236301,
+      "loss": 2.7416,
+      "step": 4991
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006783162117561397,
+      "loss": 3.2531,
+      "step": 4992
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00067819422842995,
+      "loss": 2.9469,
+      "step": 4993
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006780722329533785,
+      "loss": 3.1349,
+      "step": 4994
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006779502253347435,
+      "loss": 3.1124,
+      "step": 4995
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006778282055823646,
+      "loss": 2.9285,
+      "step": 4996
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006777061737045615,
+      "loss": 2.7609,
+      "step": 4997
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006775841297096554,
+      "loss": 3.0126,
+      "step": 4998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006774620736059678,
+      "loss": 3.0644,
+      "step": 4999
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006773400054018213,
+      "loss": 2.9558,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006772179251055393,
+      "loss": 2.8657,
+      "step": 5001
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006770958327254458,
+      "loss": 2.9863,
+      "step": 5002
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006769737282698661,
+      "loss": 2.8259,
+      "step": 5003
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006768516117471257,
+      "loss": 2.9234,
+      "step": 5004
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006767294831655516,
+      "loss": 2.7534,
+      "step": 5005
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006766073425334711,
+      "loss": 3.0967,
+      "step": 5006
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006764851898592126,
+      "loss": 3.1262,
+      "step": 5007
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006763630251511051,
+      "loss": 2.865,
+      "step": 5008
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006762408484174786,
+      "loss": 3.123,
+      "step": 5009
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006761186596666637,
+      "loss": 3.0071,
+      "step": 5010
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006759964589069923,
+      "loss": 3.1563,
+      "step": 5011
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006758742461467967,
+      "loss": 3.0005,
+      "step": 5012
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006757520213944101,
+      "loss": 3.0759,
+      "step": 5013
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006756297846581663,
+      "loss": 2.9521,
+      "step": 5014
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006755075359464004,
+      "loss": 3.1516,
+      "step": 5015
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006753852752674481,
+      "loss": 3.2737,
+      "step": 5016
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006752630026296458,
+      "loss": 3.013,
+      "step": 5017
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006751407180413307,
+      "loss": 3.0436,
+      "step": 5018
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006750184215108411,
+      "loss": 2.7141,
+      "step": 5019
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000674896113046516,
+      "loss": 2.8245,
+      "step": 5020
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006747737926566949,
+      "loss": 3.0892,
+      "step": 5021
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006746514603497185,
+      "loss": 3.129,
+      "step": 5022
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0006745291161339281,
+      "loss": 3.2025,
+      "step": 5023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000674406760017666,
+      "loss": 2.9562,
+      "step": 5024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006742843920092749,
+      "loss": 2.8989,
+      "step": 5025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006741620121170991,
+      "loss": 3.0049,
+      "step": 5026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006740396203494827,
+      "loss": 2.9545,
+      "step": 5027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006739172167147716,
+      "loss": 3.092,
+      "step": 5028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006737948012213118,
+      "loss": 2.9133,
+      "step": 5029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006736723738774502,
+      "loss": 3.2064,
+      "step": 5030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006735499346915348,
+      "loss": 2.9685,
+      "step": 5031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006734274836719144,
+      "loss": 2.8088,
+      "step": 5032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006733050208269381,
+      "loss": 3.0096,
+      "step": 5033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006731825461649565,
+      "loss": 3.1961,
+      "step": 5034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006730600596943206,
+      "loss": 2.842,
+      "step": 5035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006729375614233823,
+      "loss": 3.0715,
+      "step": 5036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006728150513604941,
+      "loss": 2.8695,
+      "step": 5037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006726925295140098,
+      "loss": 3.0572,
+      "step": 5038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006725699958922835,
+      "loss": 3.1158,
+      "step": 5039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006724474505036702,
+      "loss": 3.0843,
+      "step": 5040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000672324893356526,
+      "loss": 2.9319,
+      "step": 5041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006722023244592075,
+      "loss": 2.9481,
+      "step": 5042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006720797438200723,
+      "loss": 3.1048,
+      "step": 5043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006719571514474787,
+      "loss": 3.2696,
+      "step": 5044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006718345473497858,
+      "loss": 3.0161,
+      "step": 5045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006717119315353533,
+      "loss": 3.1523,
+      "step": 5046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006715893040125419,
+      "loss": 3.22,
+      "step": 5047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006714666647897134,
+      "loss": 3.0241,
+      "step": 5048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00067134401387523,
+      "loss": 2.8124,
+      "step": 5049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006712213512774548,
+      "loss": 2.91,
+      "step": 5050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006710986770047516,
+      "loss": 3.0619,
+      "step": 5051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000670975991065485,
+      "loss": 3.1396,
+      "step": 5052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006708532934680207,
+      "loss": 3.0249,
+      "step": 5053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006707305842207249,
+      "loss": 3.1543,
+      "step": 5054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006706078633319646,
+      "loss": 3.1555,
+      "step": 5055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006704851308101076,
+      "loss": 2.9569,
+      "step": 5056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006703623866635229,
+      "loss": 3.1967,
+      "step": 5057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006702396309005798,
+      "loss": 3.0272,
+      "step": 5058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006701168635296484,
+      "loss": 3.0992,
+      "step": 5059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006699940845590998,
+      "loss": 3.2342,
+      "step": 5060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006698712939973058,
+      "loss": 3.1936,
+      "step": 5061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006697484918526392,
+      "loss": 3.2094,
+      "step": 5062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006696256781334732,
+      "loss": 3.0703,
+      "step": 5063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006695028528481821,
+      "loss": 3.0714,
+      "step": 5064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006693800160051409,
+      "loss": 3.1843,
+      "step": 5065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006692571676127253,
+      "loss": 3.1026,
+      "step": 5066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006691343076793119,
+      "loss": 2.8064,
+      "step": 5067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000669011436213278,
+      "loss": 3.0348,
+      "step": 5068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006688885532230018,
+      "loss": 3.0585,
+      "step": 5069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006687656587168622,
+      "loss": 3.1801,
+      "step": 5070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006686427527032389,
+      "loss": 2.8738,
+      "step": 5071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006685198351905124,
+      "loss": 3.0067,
+      "step": 5072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006683969061870641,
+      "loss": 3.1308,
+      "step": 5073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006682739657012757,
+      "loss": 2.8436,
+      "step": 5074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006681510137415304,
+      "loss": 2.9894,
+      "step": 5075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006680280503162117,
+      "loss": 3.0466,
+      "step": 5076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006679050754337041,
+      "loss": 2.9413,
+      "step": 5077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006677820891023925,
+      "loss": 3.0886,
+      "step": 5078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000667659091330663,
+      "loss": 2.8419,
+      "step": 5079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006675360821269027,
+      "loss": 3.0185,
+      "step": 5080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006674130614994986,
+      "loss": 2.8568,
+      "step": 5081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006672900294568393,
+      "loss": 3.0934,
+      "step": 5082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006671669860073137,
+      "loss": 2.7878,
+      "step": 5083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006670439311593119,
+      "loss": 3.1466,
+      "step": 5084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006669208649212242,
+      "loss": 3.0237,
+      "step": 5085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006667977873014424,
+      "loss": 3.0675,
+      "step": 5086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006666746983083584,
+      "loss": 3.2132,
+      "step": 5087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006665515979503655,
+      "loss": 3.0168,
+      "step": 5088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006664284862358571,
+      "loss": 2.9581,
+      "step": 5089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006663053631732279,
+      "loss": 2.9121,
+      "step": 5090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000666182228770873,
+      "loss": 2.9057,
+      "step": 5091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006660590830371887,
+      "loss": 3.0054,
+      "step": 5092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006659359259805716,
+      "loss": 2.9443,
+      "step": 5093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006658127576094196,
+      "loss": 2.9238,
+      "step": 5094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006656895779321308,
+      "loss": 2.8879,
+      "step": 5095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006655663869571046,
+      "loss": 2.9094,
+      "step": 5096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006654431846927407,
+      "loss": 3.3531,
+      "step": 5097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006653199711474401,
+      "loss": 3.0237,
+      "step": 5098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006651967463296036,
+      "loss": 2.9706,
+      "step": 5099
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006650735102476341,
+      "loss": 2.9527,
+      "step": 5100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006649502629099345,
+      "loss": 3.0272,
+      "step": 5101
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006648270043249083,
+      "loss": 3.1701,
+      "step": 5102
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006647037345009602,
+      "loss": 3.2372,
+      "step": 5103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006645804534464954,
+      "loss": 3.1061,
+      "step": 5104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00066445716116992,
+      "loss": 2.7736,
+      "step": 5105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000664333857679641,
+      "loss": 3.1305,
+      "step": 5106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006642105429840657,
+      "loss": 2.8078,
+      "step": 5107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006640872170916026,
+      "loss": 2.8913,
+      "step": 5108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006639638800106606,
+      "loss": 2.9701,
+      "step": 5109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006638405317496502,
+      "loss": 3.0807,
+      "step": 5110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006637171723169816,
+      "loss": 3.1838,
+      "step": 5111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006635938017210661,
+      "loss": 3.0509,
+      "step": 5112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006634704199703161,
+      "loss": 2.9272,
+      "step": 5113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006633470270731443,
+      "loss": 3.0762,
+      "step": 5114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006632236230379645,
+      "loss": 3.1728,
+      "step": 5115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006631002078731914,
+      "loss": 2.8357,
+      "step": 5116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006629767815872399,
+      "loss": 3.1058,
+      "step": 5117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006628533441885261,
+      "loss": 2.8501,
+      "step": 5118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006627298956854667,
+      "loss": 3.1541,
+      "step": 5119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006626064360864791,
+      "loss": 2.9725,
+      "step": 5120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006624829653999817,
+      "loss": 3.114,
+      "step": 5121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006623594836343933,
+      "loss": 3.0559,
+      "step": 5122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006622359907981337,
+      "loss": 3.0325,
+      "step": 5123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006621124868996234,
+      "loss": 2.8229,
+      "step": 5124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006619889719472839,
+      "loss": 3.2713,
+      "step": 5125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006618654459495369,
+      "loss": 3.1257,
+      "step": 5126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006617419089148054,
+      "loss": 3.269,
+      "step": 5127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006616183608515126,
+      "loss": 3.0017,
+      "step": 5128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006614948017680831,
+      "loss": 2.8065,
+      "step": 5129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006613712316729417,
+      "loss": 2.9864,
+      "step": 5130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006612476505745142,
+      "loss": 2.9111,
+      "step": 5131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006611240584812273,
+      "loss": 3.102,
+      "step": 5132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006610004554015081,
+      "loss": 2.9168,
+      "step": 5133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006608768413437847,
+      "loss": 2.838,
+      "step": 5134
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006607532163164859,
+      "loss": 2.9232,
+      "step": 5135
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006606295803280411,
+      "loss": 3.0762,
+      "step": 5136
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006605059333868806,
+      "loss": 3.1823,
+      "step": 5137
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006603822755014355,
+      "loss": 3.0173,
+      "step": 5138
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006602586066801373,
+      "loss": 3.1334,
+      "step": 5139
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006601349269314187,
+      "loss": 2.9484,
+      "step": 5140
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006600112362637132,
+      "loss": 2.9483,
+      "step": 5141
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006598875346854544,
+      "loss": 3.0972,
+      "step": 5142
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006597638222050772,
+      "loss": 3.0967,
+      "step": 5143
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.000659640098831017,
+      "loss": 2.9938,
+      "step": 5144
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00065951636457171,
+      "loss": 2.8626,
+      "step": 5145
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006593926194355934,
+      "loss": 3.2433,
+      "step": 5146
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0006592688634311046,
+      "loss": 3.0728,
+      "step": 5147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006591450965666824,
+      "loss": 3.0672,
+      "step": 5148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006590213188507655,
+      "loss": 3.0948,
+      "step": 5149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006588975302917943,
+      "loss": 3.0559,
+      "step": 5150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006587737308982092,
+      "loss": 3.026,
+      "step": 5151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006586499206784517,
+      "loss": 2.9493,
+      "step": 5152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006585260996409639,
+      "loss": 3.0994,
+      "step": 5153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006584022677941887,
+      "loss": 3.169,
+      "step": 5154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006582784251465697,
+      "loss": 2.8886,
+      "step": 5155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006581545717065514,
+      "loss": 3.0883,
+      "step": 5156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006580307074825788,
+      "loss": 3.1371,
+      "step": 5157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006579068324830976,
+      "loss": 2.9088,
+      "step": 5158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006577829467165546,
+      "loss": 3.0289,
+      "step": 5159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006576590501913968,
+      "loss": 2.8969,
+      "step": 5160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006575351429160723,
+      "loss": 3.0675,
+      "step": 5161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006574112248990302,
+      "loss": 3.1654,
+      "step": 5162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006572872961487196,
+      "loss": 3.2002,
+      "step": 5163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006571633566735909,
+      "loss": 3.1901,
+      "step": 5164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000657039406482095,
+      "loss": 3.0698,
+      "step": 5165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006569154455826838,
+      "loss": 3.0058,
+      "step": 5166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006567914739838093,
+      "loss": 2.93,
+      "step": 5167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000656667491693925,
+      "loss": 2.8549,
+      "step": 5168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006565434987214847,
+      "loss": 3.1915,
+      "step": 5169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000656419495074943,
+      "loss": 2.9316,
+      "step": 5170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000656295480762755,
+      "loss": 3.1687,
+      "step": 5171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006561714557933771,
+      "loss": 2.9662,
+      "step": 5172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000656047420175266,
+      "loss": 3.0516,
+      "step": 5173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006559233739168791,
+      "loss": 2.9183,
+      "step": 5174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006557993170266748,
+      "loss": 3.0094,
+      "step": 5175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006556752495131118,
+      "loss": 3.1329,
+      "step": 5176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006555511713846501,
+      "loss": 3.0733,
+      "step": 5177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006554270826497499,
+      "loss": 3.0339,
+      "step": 5178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006553029833168724,
+      "loss": 2.9448,
+      "step": 5179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006551788733944794,
+      "loss": 3.0034,
+      "step": 5180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006550547528910338,
+      "loss": 2.9879,
+      "step": 5181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006549306218149984,
+      "loss": 2.8878,
+      "step": 5182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006548064801748375,
+      "loss": 2.8927,
+      "step": 5183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006546823279790159,
+      "loss": 2.9481,
+      "step": 5184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006545581652359989,
+      "loss": 2.9199,
+      "step": 5185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006544339919542527,
+      "loss": 3.027,
+      "step": 5186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006543098081422444,
+      "loss": 3.0682,
+      "step": 5187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006541856138084413,
+      "loss": 2.9733,
+      "step": 5188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000654061408961312,
+      "loss": 3.1302,
+      "step": 5189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006539371936093256,
+      "loss": 3.0898,
+      "step": 5190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006538129677609515,
+      "loss": 2.8272,
+      "step": 5191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006536887314246605,
+      "loss": 3.1655,
+      "step": 5192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006535644846089237,
+      "loss": 3.0235,
+      "step": 5193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006534402273222131,
+      "loss": 3.0272,
+      "step": 5194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006533159595730013,
+      "loss": 3.1953,
+      "step": 5195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006531916813697615,
+      "loss": 2.9616,
+      "step": 5196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006530673927209678,
+      "loss": 3.0184,
+      "step": 5197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006529430936350951,
+      "loss": 2.8092,
+      "step": 5198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006528187841206188,
+      "loss": 3.1263,
+      "step": 5199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000652694464186015,
+      "loss": 2.9206,
+      "step": 5200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006525701338397608,
+      "loss": 3.2256,
+      "step": 5201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006524457930903337,
+      "loss": 3.1387,
+      "step": 5202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000652321441946212,
+      "loss": 3.1668,
+      "step": 5203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006521970804158748,
+      "loss": 3.0196,
+      "step": 5204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006520727085078016,
+      "loss": 3.0605,
+      "step": 5205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006519483262304731,
+      "loss": 2.8106,
+      "step": 5206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006518239335923705,
+      "loss": 3.2214,
+      "step": 5207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006516995306019754,
+      "loss": 3.0908,
+      "step": 5208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006515751172677706,
+      "loss": 2.8715,
+      "step": 5209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006514506935982393,
+      "loss": 3.0003,
+      "step": 5210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006513262596018654,
+      "loss": 2.8011,
+      "step": 5211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006512018152871335,
+      "loss": 3.0107,
+      "step": 5212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000651077360662529,
+      "loss": 2.8221,
+      "step": 5213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006509528957365383,
+      "loss": 2.8469,
+      "step": 5214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000650828420517648,
+      "loss": 3.1748,
+      "step": 5215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006507039350143456,
+      "loss": 2.9208,
+      "step": 5216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006505794392351191,
+      "loss": 3.1506,
+      "step": 5217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006504549331884575,
+      "loss": 3.0028,
+      "step": 5218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006503304168828506,
+      "loss": 2.9562,
+      "step": 5219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006502058903267884,
+      "loss": 3.166,
+      "step": 5220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000650081353528762,
+      "loss": 2.7158,
+      "step": 5221
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000649956806497263,
+      "loss": 3.1058,
+      "step": 5222
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000649832249240784,
+      "loss": 3.0707,
+      "step": 5223
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000649707681767818,
+      "loss": 3.1911,
+      "step": 5224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006495831040868587,
+      "loss": 3.0263,
+      "step": 5225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006494585162064007,
+      "loss": 2.9764,
+      "step": 5226
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006493339181349389,
+      "loss": 2.9354,
+      "step": 5227
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006492093098809695,
+      "loss": 3.0259,
+      "step": 5228
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006490846914529887,
+      "loss": 2.8544,
+      "step": 5229
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000648960062859494,
+      "loss": 3.2072,
+      "step": 5230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006488354241089836,
+      "loss": 3.0086,
+      "step": 5231
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006487107752099556,
+      "loss": 3.0211,
+      "step": 5232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006485861161709096,
+      "loss": 3.0918,
+      "step": 5233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006484614470003457,
+      "loss": 3.0004,
+      "step": 5234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006483367677067644,
+      "loss": 3.0838,
+      "step": 5235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006482120782986674,
+      "loss": 2.8101,
+      "step": 5236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006480873787845563,
+      "loss": 2.9823,
+      "step": 5237
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006479626691729345,
+      "loss": 3.0231,
+      "step": 5238
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000647837949472305,
+      "loss": 3.0342,
+      "step": 5239
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006477132196911721,
+      "loss": 2.9332,
+      "step": 5240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006475884798380407,
+      "loss": 2.8344,
+      "step": 5241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006474637299214165,
+      "loss": 3.2134,
+      "step": 5242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006473389699498053,
+      "loss": 2.9511,
+      "step": 5243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006472141999317143,
+      "loss": 3.0357,
+      "step": 5244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000647089419875651,
+      "loss": 3.1906,
+      "step": 5245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006469646297901235,
+      "loss": 3.181,
+      "step": 5246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006468398296836412,
+      "loss": 2.8257,
+      "step": 5247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006467150195647133,
+      "loss": 2.9191,
+      "step": 5248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006465901994418505,
+      "loss": 3.1273,
+      "step": 5249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006464653693235635,
+      "loss": 3.1577,
+      "step": 5250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.000646340529218364,
+      "loss": 3.0915,
+      "step": 5251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006462156791347646,
+      "loss": 2.9114,
+      "step": 5252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006460908190812782,
+      "loss": 2.9619,
+      "step": 5253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006459659490664186,
+      "loss": 3.1064,
+      "step": 5254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006458410690987001,
+      "loss": 2.8006,
+      "step": 5255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006457161791866379,
+      "loss": 3.1915,
+      "step": 5256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006455912793387477,
+      "loss": 3.1158,
+      "step": 5257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006454663695635459,
+      "loss": 2.8638,
+      "step": 5258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006453414498695496,
+      "loss": 2.8632,
+      "step": 5259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006452165202652768,
+      "loss": 2.8609,
+      "step": 5260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006450915807592457,
+      "loss": 3.1095,
+      "step": 5261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006449666313599757,
+      "loss": 3.1781,
+      "step": 5262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006448416720759864,
+      "loss": 2.8379,
+      "step": 5263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006447167029157985,
+      "loss": 2.9033,
+      "step": 5264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006445917238879331,
+      "loss": 2.8514,
+      "step": 5265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006444667350009119,
+      "loss": 2.8854,
+      "step": 5266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006443417362632575,
+      "loss": 2.6534,
+      "step": 5267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006442167276834932,
+      "loss": 2.9915,
+      "step": 5268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006440917092701427,
+      "loss": 2.9442,
+      "step": 5269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006439666810317306,
+      "loss": 3.0004,
+      "step": 5270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.0006438416429767822,
+      "loss": 2.9773,
+      "step": 5271
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000643716595113823,
+      "loss": 3.3628,
+      "step": 5272
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00064359153745138,
+      "loss": 2.9774,
+      "step": 5273
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006434664699979802,
+      "loss": 3.118,
+      "step": 5274
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006433413927621513,
+      "loss": 2.9622,
+      "step": 5275
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000643216305752422,
+      "loss": 3.0768,
+      "step": 5276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006430912089773216,
+      "loss": 2.9685,
+      "step": 5277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006429661024453799,
+      "loss": 3.0326,
+      "step": 5278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006428409861651274,
+      "loss": 2.6677,
+      "step": 5279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006427158601450953,
+      "loss": 2.9952,
+      "step": 5280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006425907243938153,
+      "loss": 2.9023,
+      "step": 5281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006424655789198201,
+      "loss": 3.0785,
+      "step": 5282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000642340423731643,
+      "loss": 2.8881,
+      "step": 5283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006422152588378178,
+      "loss": 2.9064,
+      "step": 5284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000642090084246879,
+      "loss": 2.951,
+      "step": 5285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006419648999673616,
+      "loss": 2.8681,
+      "step": 5286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006418397060078017,
+      "loss": 2.919,
+      "step": 5287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006417145023767354,
+      "loss": 2.981,
+      "step": 5288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006415892890827002,
+      "loss": 2.9821,
+      "step": 5289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000641464066134234,
+      "loss": 3.1475,
+      "step": 5290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006413388335398749,
+      "loss": 2.945,
+      "step": 5291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006412135913081623,
+      "loss": 2.9855,
+      "step": 5292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006410883394476361,
+      "loss": 3.0119,
+      "step": 5293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006409630779668364,
+      "loss": 3.0996,
+      "step": 5294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006408378068743046,
+      "loss": 2.9968,
+      "step": 5295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006407125261785822,
+      "loss": 2.9797,
+      "step": 5296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006405872358882117,
+      "loss": 3.2878,
+      "step": 5297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006404619360117363,
+      "loss": 3.236,
+      "step": 5298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006403366265576995,
+      "loss": 3.26,
+      "step": 5299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000640211307534646,
+      "loss": 2.7535,
+      "step": 5300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006400859789511205,
+      "loss": 2.991,
+      "step": 5301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006399606408156687,
+      "loss": 3.1216,
+      "step": 5302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006398352931368372,
+      "loss": 2.9393,
+      "step": 5303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006397099359231726,
+      "loss": 2.9225,
+      "step": 5304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006395845691832228,
+      "loss": 3.0489,
+      "step": 5305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000639459192925536,
+      "loss": 2.9374,
+      "step": 5306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006393338071586613,
+      "loss": 3.1243,
+      "step": 5307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000639208411891148,
+      "loss": 2.9026,
+      "step": 5308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006390830071315465,
+      "loss": 3.0245,
+      "step": 5309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006389575928884074,
+      "loss": 3.1118,
+      "step": 5310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006388321691702824,
+      "loss": 3.1539,
+      "step": 5311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006387067359857238,
+      "loss": 3.1427,
+      "step": 5312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006385812933432843,
+      "loss": 2.7473,
+      "step": 5313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006384558412515174,
+      "loss": 2.9235,
+      "step": 5314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000638330379718977,
+      "loss": 3.2692,
+      "step": 5315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.000638204908754218,
+      "loss": 3.1479,
+      "step": 5316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006380794283657958,
+      "loss": 3.1246,
+      "step": 5317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006379539385622663,
+      "loss": 2.943,
+      "step": 5318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006378284393521863,
+      "loss": 3.1115,
+      "step": 5319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006377029307441132,
+      "loss": 2.9662,
+      "step": 5320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006375774127466046,
+      "loss": 3.023,
+      "step": 5321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006374518853682194,
+      "loss": 3.0743,
+      "step": 5322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006373263486175168,
+      "loss": 2.9203,
+      "step": 5323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006372008025030566,
+      "loss": 2.8859,
+      "step": 5324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006370752470333993,
+      "loss": 3.1366,
+      "step": 5325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006369496822171061,
+      "loss": 2.92,
+      "step": 5326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006368241080627386,
+      "loss": 3.1344,
+      "step": 5327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006366985245788595,
+      "loss": 3.392,
+      "step": 5328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006365729317740316,
+      "loss": 3.1177,
+      "step": 5329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006364473296568189,
+      "loss": 2.9518,
+      "step": 5330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006363217182357855,
+      "loss": 3.225,
+      "step": 5331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006361960975194965,
+      "loss": 2.936,
+      "step": 5332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006360704675165172,
+      "loss": 3.0718,
+      "step": 5333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006359448282354141,
+      "loss": 3.1209,
+      "step": 5334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006358191796847539,
+      "loss": 2.8752,
+      "step": 5335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006356935218731044,
+      "loss": 3.1368,
+      "step": 5336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006355678548090333,
+      "loss": 3.0609,
+      "step": 5337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006354421785011098,
+      "loss": 3.0532,
+      "step": 5338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006353164929579029,
+      "loss": 2.9859,
+      "step": 5339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006351907981879828,
+      "loss": 3.0036,
+      "step": 5340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006350650941999201,
+      "loss": 3.1231,
+      "step": 5341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006349393810022861,
+      "loss": 2.9176,
+      "step": 5342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006348136586036526,
+      "loss": 3.0734,
+      "step": 5343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006346879270125924,
+      "loss": 2.9077,
+      "step": 5344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006345621862376785,
+      "loss": 2.9104,
+      "step": 5345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006344364362874845,
+      "loss": 3.1471,
+      "step": 5346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006343106771705849,
+      "loss": 3.1399,
+      "step": 5347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006341849088955549,
+      "loss": 2.8304,
+      "step": 5348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006340591314709699,
+      "loss": 3.0267,
+      "step": 5349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006339333449054066,
+      "loss": 2.8218,
+      "step": 5350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006338075492074415,
+      "loss": 2.836,
+      "step": 5351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006336817443856522,
+      "loss": 3.2144,
+      "step": 5352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006335559304486171,
+      "loss": 3.0564,
+      "step": 5353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006334301074049145,
+      "loss": 3.2333,
+      "step": 5354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006333042752631243,
+      "loss": 3.1842,
+      "step": 5355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006331784340318263,
+      "loss": 2.9615,
+      "step": 5356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006330525837196008,
+      "loss": 2.8932,
+      "step": 5357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006329267243350296,
+      "loss": 2.8277,
+      "step": 5358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006328008558866944,
+      "loss": 3.2494,
+      "step": 5359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006326749783831776,
+      "loss": 3.0039,
+      "step": 5360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006325490918330621,
+      "loss": 3.2611,
+      "step": 5361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006324231962449322,
+      "loss": 2.9823,
+      "step": 5362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006322972916273718,
+      "loss": 2.9377,
+      "step": 5363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006321713779889657,
+      "loss": 2.8193,
+      "step": 5364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006320454553382999,
+      "loss": 2.9432,
+      "step": 5365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006319195236839603,
+      "loss": 3.0156,
+      "step": 5366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006317935830345338,
+      "loss": 3.101,
+      "step": 5367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006316676333986079,
+      "loss": 3.3301,
+      "step": 5368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006315416747847704,
+      "loss": 2.9819,
+      "step": 5369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006314157072016101,
+      "loss": 3.3378,
+      "step": 5370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006312897306577163,
+      "loss": 2.7422,
+      "step": 5371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006311637451616787,
+      "loss": 2.7312,
+      "step": 5372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006310377507220878,
+      "loss": 3.03,
+      "step": 5373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006309117473475347,
+      "loss": 3.0562,
+      "step": 5374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006307857350466112,
+      "loss": 3.1641,
+      "step": 5375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006306597138279094,
+      "loss": 2.9562,
+      "step": 5376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006305336837000224,
+      "loss": 3.0118,
+      "step": 5377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006304076446715434,
+      "loss": 3.0239,
+      "step": 5378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006302815967510668,
+      "loss": 3.0131,
+      "step": 5379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006301555399471873,
+      "loss": 3.0453,
+      "step": 5380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006300294742685002,
+      "loss": 2.8824,
+      "step": 5381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006299033997236015,
+      "loss": 3.1962,
+      "step": 5382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006297773163210875,
+      "loss": 3.0052,
+      "step": 5383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006296512240695555,
+      "loss": 2.8861,
+      "step": 5384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006295251229776034,
+      "loss": 2.9055,
+      "step": 5385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006293990130538293,
+      "loss": 2.9301,
+      "step": 5386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006292728943068321,
+      "loss": 2.8178,
+      "step": 5387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006291467667452116,
+      "loss": 3.0033,
+      "step": 5388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006290206303775681,
+      "loss": 3.2655,
+      "step": 5389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006288944852125019,
+      "loss": 2.9011,
+      "step": 5390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006287683312586145,
+      "loss": 2.9566,
+      "step": 5391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006286421685245082,
+      "loss": 2.9936,
+      "step": 5392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006285159970187851,
+      "loss": 3.0964,
+      "step": 5393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006283898167500486,
+      "loss": 3.0546,
+      "step": 5394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0006282636277269021,
+      "loss": 3.1196,
+      "step": 5395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006281374299579505,
+      "loss": 2.9863,
+      "step": 5396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006280112234517986,
+      "loss": 3.0654,
+      "step": 5397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006278850082170515,
+      "loss": 3.1281,
+      "step": 5398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006277587842623159,
+      "loss": 3.0966,
+      "step": 5399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006276325515961981,
+      "loss": 3.2521,
+      "step": 5400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006275063102273057,
+      "loss": 2.8248,
+      "step": 5401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006273800601642464,
+      "loss": 3.0735,
+      "step": 5402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006272538014156288,
+      "loss": 3.0634,
+      "step": 5403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000627127533990062,
+      "loss": 2.9423,
+      "step": 5404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006270012578961559,
+      "loss": 2.821,
+      "step": 5405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006268749731425206,
+      "loss": 2.9485,
+      "step": 5406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006267486797377668,
+      "loss": 2.9337,
+      "step": 5407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006266223776905062,
+      "loss": 3.0363,
+      "step": 5408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006264960670093507,
+      "loss": 2.843,
+      "step": 5409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006263697477029131,
+      "loss": 2.9907,
+      "step": 5410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006262434197798065,
+      "loss": 3.0599,
+      "step": 5411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006261170832486448,
+      "loss": 3.1012,
+      "step": 5412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006259907381180423,
+      "loss": 2.899,
+      "step": 5413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006258643843966142,
+      "loss": 3.1764,
+      "step": 5414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006257380220929758,
+      "loss": 3.0766,
+      "step": 5415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006256116512157435,
+      "loss": 3.1631,
+      "step": 5416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006254852717735339,
+      "loss": 3.0579,
+      "step": 5417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006253588837749644,
+      "loss": 2.978,
+      "step": 5418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006252324872286528,
+      "loss": 2.9907,
+      "step": 5419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006251060821432177,
+      "loss": 2.959,
+      "step": 5420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006249796685272782,
+      "loss": 3.2134,
+      "step": 5421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006248532463894541,
+      "loss": 3.0749,
+      "step": 5422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006247268157383653,
+      "loss": 2.8437,
+      "step": 5423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006246003765826328,
+      "loss": 2.933,
+      "step": 5424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006244739289308778,
+      "loss": 3.0197,
+      "step": 5425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006243474727917228,
+      "loss": 2.9111,
+      "step": 5426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006242210081737898,
+      "loss": 3.0706,
+      "step": 5427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006240945350857024,
+      "loss": 3.1933,
+      "step": 5428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006239680535360839,
+      "loss": 2.842,
+      "step": 5429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006238415635335588,
+      "loss": 3.0381,
+      "step": 5430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006237150650867519,
+      "loss": 3.0923,
+      "step": 5431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006235885582042888,
+      "loss": 2.882,
+      "step": 5432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006234620428947952,
+      "loss": 2.9211,
+      "step": 5433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006233355191668981,
+      "loss": 2.9835,
+      "step": 5434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006232089870292245,
+      "loss": 3.3002,
+      "step": 5435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000623082446490402,
+      "loss": 3.0758,
+      "step": 5436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000622955897559059,
+      "loss": 2.9946,
+      "step": 5437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006228293402438246,
+      "loss": 3.0128,
+      "step": 5438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006227027745533279,
+      "loss": 3.0519,
+      "step": 5439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006225762004961991,
+      "loss": 2.9855,
+      "step": 5440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006224496180810689,
+      "loss": 3.0425,
+      "step": 5441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006223230273165683,
+      "loss": 3.0735,
+      "step": 5442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006221964282113291,
+      "loss": 2.9052,
+      "step": 5443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006220698207739837,
+      "loss": 3.1226,
+      "step": 5444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006219432050131648,
+      "loss": 3.2255,
+      "step": 5445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006218165809375061,
+      "loss": 2.929,
+      "step": 5446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006216899485556411,
+      "loss": 2.9145,
+      "step": 5447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006215633078762051,
+      "loss": 2.7563,
+      "step": 5448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006214366589078327,
+      "loss": 3.0174,
+      "step": 5449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006213100016591597,
+      "loss": 3.0741,
+      "step": 5450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006211833361388226,
+      "loss": 3.0019,
+      "step": 5451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000621056662355458,
+      "loss": 2.8904,
+      "step": 5452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006209299803177034,
+      "loss": 3.2229,
+      "step": 5453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006208032900341969,
+      "loss": 3.0553,
+      "step": 5454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006206765915135766,
+      "loss": 3.0933,
+      "step": 5455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006205498847644819,
+      "loss": 3.0632,
+      "step": 5456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006204231697955526,
+      "loss": 3.1037,
+      "step": 5457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006202964466154286,
+      "loss": 2.7908,
+      "step": 5458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006201697152327508,
+      "loss": 2.9009,
+      "step": 5459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006200429756561605,
+      "loss": 3.2148,
+      "step": 5460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006199162278942997,
+      "loss": 2.9506,
+      "step": 5461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006197894719558106,
+      "loss": 2.8827,
+      "step": 5462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006196627078493363,
+      "loss": 2.9285,
+      "step": 5463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006195359355835205,
+      "loss": 3.0849,
+      "step": 5464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006194091551670072,
+      "loss": 3.0388,
+      "step": 5465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006192823666084412,
+      "loss": 3.0086,
+      "step": 5466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006191555699164675,
+      "loss": 3.1578,
+      "step": 5467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000619028765099732,
+      "loss": 3.2675,
+      "step": 5468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006189019521668811,
+      "loss": 2.9164,
+      "step": 5469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006187751311265614,
+      "loss": 3.0598,
+      "step": 5470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006186483019874208,
+      "loss": 3.0538,
+      "step": 5471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000618521464758107,
+      "loss": 3.034,
+      "step": 5472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006183946194472686,
+      "loss": 3.0042,
+      "step": 5473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006182677660635547,
+      "loss": 3.1865,
+      "step": 5474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006181409046156149,
+      "loss": 3.2428,
+      "step": 5475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006180140351120994,
+      "loss": 2.9131,
+      "step": 5476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006178871575616589,
+      "loss": 3.1416,
+      "step": 5477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006177602719729449,
+      "loss": 3.0649,
+      "step": 5478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006176333783546092,
+      "loss": 2.9822,
+      "step": 5479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000617506476715304,
+      "loss": 3.2173,
+      "step": 5480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006173795670636824,
+      "loss": 2.8539,
+      "step": 5481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006172526494083977,
+      "loss": 3.018,
+      "step": 5482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006171257237581042,
+      "loss": 2.9364,
+      "step": 5483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006169987901214562,
+      "loss": 3.2304,
+      "step": 5484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006168718485071089,
+      "loss": 2.8149,
+      "step": 5485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006167448989237182,
+      "loss": 2.9721,
+      "step": 5486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00061661794137994,
+      "loss": 3.0116,
+      "step": 5487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006164909758844312,
+      "loss": 2.7918,
+      "step": 5488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006163640024458491,
+      "loss": 3.049,
+      "step": 5489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006162370210728514,
+      "loss": 2.8477,
+      "step": 5490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006161100317740966,
+      "loss": 3.1941,
+      "step": 5491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006159830345582436,
+      "loss": 3.1075,
+      "step": 5492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006158560294339517,
+      "loss": 3.0835,
+      "step": 5493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006157290164098811,
+      "loss": 3.0151,
+      "step": 5494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006156019954946923,
+      "loss": 2.8188,
+      "step": 5495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006154749666970463,
+      "loss": 3.151,
+      "step": 5496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006153479300256046,
+      "loss": 3.2142,
+      "step": 5497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006152208854890296,
+      "loss": 3.0483,
+      "step": 5498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006150938330959838,
+      "loss": 3.1297,
+      "step": 5499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006149667728551306,
+      "loss": 2.9596,
+      "step": 5500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006148397047751334,
+      "loss": 3.0145,
+      "step": 5501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006147126288646569,
+      "loss": 2.816,
+      "step": 5502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006145855451323656,
+      "loss": 2.9839,
+      "step": 5503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006144584535869252,
+      "loss": 3.0938,
+      "step": 5504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006143313542370011,
+      "loss": 3.0991,
+      "step": 5505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006142042470912601,
+      "loss": 2.9826,
+      "step": 5506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006140771321583689,
+      "loss": 3.1478,
+      "step": 5507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006139500094469952,
+      "loss": 2.9834,
+      "step": 5508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000613822878965807,
+      "loss": 3.0575,
+      "step": 5509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006136957407234727,
+      "loss": 2.8377,
+      "step": 5510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006135685947286615,
+      "loss": 3.0144,
+      "step": 5511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006134414409900427,
+      "loss": 2.9206,
+      "step": 5512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006133142795162869,
+      "loss": 3.0163,
+      "step": 5513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006131871103160643,
+      "loss": 3.1401,
+      "step": 5514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006130599333980464,
+      "loss": 3.2236,
+      "step": 5515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006129327487709048,
+      "loss": 2.6802,
+      "step": 5516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006128055564433117,
+      "loss": 2.939,
+      "step": 5517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006126783564239399,
+      "loss": 2.9226,
+      "step": 5518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.0006125511487214627,
+      "loss": 2.8258,
+      "step": 5519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006124239333445538,
+      "loss": 2.9005,
+      "step": 5520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006122967103018877,
+      "loss": 3.005,
+      "step": 5521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006121694796021391,
+      "loss": 2.8907,
+      "step": 5522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006120422412539833,
+      "loss": 2.6832,
+      "step": 5523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006119149952660965,
+      "loss": 3.2988,
+      "step": 5524
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006117877416471549,
+      "loss": 3.1041,
+      "step": 5525
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006116604804058355,
+      "loss": 2.9687,
+      "step": 5526
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006115332115508158,
+      "loss": 3.0487,
+      "step": 5527
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006114059350907736,
+      "loss": 2.9311,
+      "step": 5528
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006112786510343875,
+      "loss": 3.2424,
+      "step": 5529
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006111513593903365,
+      "loss": 3.2249,
+      "step": 5530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006110240601673,
+      "loss": 3.2396,
+      "step": 5531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006108967533739585,
+      "loss": 2.8748,
+      "step": 5532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006107694390189922,
+      "loss": 3.0082,
+      "step": 5533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006106421171110821,
+      "loss": 3.0698,
+      "step": 5534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00061051478765891,
+      "loss": 3.1164,
+      "step": 5535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006103874506711579,
+      "loss": 3.0676,
+      "step": 5536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006102601061565086,
+      "loss": 2.9027,
+      "step": 5537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006101327541236449,
+      "loss": 3.1941,
+      "step": 5538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006100053945812508,
+      "loss": 2.9495,
+      "step": 5539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006098780275380101,
+      "loss": 2.9779,
+      "step": 5540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006097506530026078,
+      "loss": 3.2523,
+      "step": 5541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006096232709837288,
+      "loss": 3.0265,
+      "step": 5542
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000609495881490059,
+      "loss": 2.7802,
+      "step": 5543
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006093684845302847,
+      "loss": 3.0541,
+      "step": 5544
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006092410801130922,
+      "loss": 2.8979,
+      "step": 5545
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006091136682471692,
+      "loss": 3.0572,
+      "step": 5546
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000608986248941203,
+      "loss": 3.0084,
+      "step": 5547
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006088588222038822,
+      "loss": 2.9043,
+      "step": 5548
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006087313880438954,
+      "loss": 2.9576,
+      "step": 5549
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006086039464699317,
+      "loss": 3.0567,
+      "step": 5550
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000608476497490681,
+      "loss": 3.0806,
+      "step": 5551
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006083490411148335,
+      "loss": 2.8387,
+      "step": 5552
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00060822157735108,
+      "loss": 2.9992,
+      "step": 5553
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006080941062081117,
+      "loss": 3.0306,
+      "step": 5554
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006079666276946205,
+      "loss": 2.907,
+      "step": 5555
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006078391418192987,
+      "loss": 3.1208,
+      "step": 5556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006077116485908388,
+      "loss": 2.9656,
+      "step": 5557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006075841480179344,
+      "loss": 2.8949,
+      "step": 5558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000607456640109279,
+      "loss": 2.9869,
+      "step": 5559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006073291248735671,
+      "loss": 2.9001,
+      "step": 5560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006072016023194932,
+      "loss": 2.9251,
+      "step": 5561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006070740724557529,
+      "loss": 2.8433,
+      "step": 5562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006069465352910418,
+      "loss": 2.9951,
+      "step": 5563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006068189908340563,
+      "loss": 2.964,
+      "step": 5564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006066914390934929,
+      "loss": 3.0891,
+      "step": 5565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006065638800780489,
+      "loss": 2.9505,
+      "step": 5566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006064363137964225,
+      "loss": 2.8856,
+      "step": 5567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006063087402573115,
+      "loss": 3.2042,
+      "step": 5568
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006061811594694148,
+      "loss": 3.0108,
+      "step": 5569
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006060535714414316,
+      "loss": 2.8263,
+      "step": 5570
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006059259761820617,
+      "loss": 2.9642,
+      "step": 5571
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006057983737000052,
+      "loss": 3.0737,
+      "step": 5572
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006056707640039632,
+      "loss": 3.0886,
+      "step": 5573
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006055431471026363,
+      "loss": 3.1261,
+      "step": 5574
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006054155230047266,
+      "loss": 3.0971,
+      "step": 5575
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006052878917189365,
+      "loss": 2.9535,
+      "step": 5576
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006051602532539682,
+      "loss": 3.0516,
+      "step": 5577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006050326076185254,
+      "loss": 3.0041,
+      "step": 5578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006049049548213114,
+      "loss": 2.8585,
+      "step": 5579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006047772948710303,
+      "loss": 2.934,
+      "step": 5580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000604649627776387,
+      "loss": 2.8997,
+      "step": 5581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006045219535460864,
+      "loss": 3.139,
+      "step": 5582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006043942721888343,
+      "loss": 3.3459,
+      "step": 5583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006042665837133367,
+      "loss": 3.15,
+      "step": 5584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006041388881283002,
+      "loss": 2.885,
+      "step": 5585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006040111854424319,
+      "loss": 3.0232,
+      "step": 5586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006038834756644395,
+      "loss": 3.2194,
+      "step": 5587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006037557588030308,
+      "loss": 3.1211,
+      "step": 5588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006036280348669144,
+      "loss": 3.1783,
+      "step": 5589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006035003038647994,
+      "loss": 2.9795,
+      "step": 5590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000603372565805395,
+      "loss": 3.0426,
+      "step": 5591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006032448206974115,
+      "loss": 3.0417,
+      "step": 5592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006031170685495593,
+      "loss": 2.9865,
+      "step": 5593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006029893093705492,
+      "loss": 2.941,
+      "step": 5594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006028615431690927,
+      "loss": 3.0738,
+      "step": 5595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006027337699539016,
+      "loss": 2.9951,
+      "step": 5596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006026059897336884,
+      "loss": 2.8251,
+      "step": 5597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006024782025171658,
+      "loss": 2.9531,
+      "step": 5598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006023504083130471,
+      "loss": 3.1155,
+      "step": 5599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006022226071300462,
+      "loss": 3.2562,
+      "step": 5600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006020947989768775,
+      "loss": 2.8663,
+      "step": 5601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006019669838622556,
+      "loss": 2.8347,
+      "step": 5602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006018391617948956,
+      "loss": 2.9787,
+      "step": 5603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006017113327835134,
+      "loss": 3.1127,
+      "step": 5604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000601583496836825,
+      "loss": 3.1623,
+      "step": 5605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006014556539635471,
+      "loss": 2.9029,
+      "step": 5606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006013278041723971,
+      "loss": 2.9151,
+      "step": 5607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006011999474720922,
+      "loss": 3.1527,
+      "step": 5608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006010720838713506,
+      "loss": 3.1667,
+      "step": 5609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006009442133788908,
+      "loss": 2.898,
+      "step": 5610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006008163360034319,
+      "loss": 2.8516,
+      "step": 5611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006006884517536932,
+      "loss": 3.003,
+      "step": 5612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006005605606383947,
+      "loss": 3.1129,
+      "step": 5613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000600432662666257,
+      "loss": 3.0096,
+      "step": 5614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006003047578460005,
+      "loss": 2.9111,
+      "step": 5615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006001768461863472,
+      "loss": 3.1032,
+      "step": 5616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0006000489276960185,
+      "loss": 2.8852,
+      "step": 5617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005999210023837366,
+      "loss": 2.8202,
+      "step": 5618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005997930702582244,
+      "loss": 3.1369,
+      "step": 5619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005996651313282051,
+      "loss": 3.1353,
+      "step": 5620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005995371856024022,
+      "loss": 2.9427,
+      "step": 5621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005994092330895401,
+      "loss": 3.0376,
+      "step": 5622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005992812737983432,
+      "loss": 3.2021,
+      "step": 5623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005991533077375366,
+      "loss": 3.1826,
+      "step": 5624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005990253349158459,
+      "loss": 3.2355,
+      "step": 5625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000598897355341997,
+      "loss": 3.0936,
+      "step": 5626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005987693690247162,
+      "loss": 3.0205,
+      "step": 5627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005986413759727307,
+      "loss": 2.819,
+      "step": 5628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005985133761947675,
+      "loss": 2.9576,
+      "step": 5629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005983853696995547,
+      "loss": 3.0195,
+      "step": 5630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005982573564958204,
+      "loss": 2.8184,
+      "step": 5631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005981293365922935,
+      "loss": 2.9915,
+      "step": 5632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005980013099977031,
+      "loss": 3.231,
+      "step": 5633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005978732767207788,
+      "loss": 2.977,
+      "step": 5634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005977452367702508,
+      "loss": 3.146,
+      "step": 5635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005976171901548495,
+      "loss": 3.036,
+      "step": 5636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000597489136883306,
+      "loss": 3.2188,
+      "step": 5637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005973610769643517,
+      "loss": 3.1805,
+      "step": 5638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005972330104067187,
+      "loss": 3.1371,
+      "step": 5639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005971049372191392,
+      "loss": 3.0149,
+      "step": 5640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.000596976857410346,
+      "loss": 3.0395,
+      "step": 5641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005968487709890725,
+      "loss": 3.0279,
+      "step": 5642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0005967206779640522,
+      "loss": 2.9563,
+      "step": 5643
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005965925783440194,
+      "loss": 3.1685,
+      "step": 5644
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000596464472137709,
+      "loss": 3.0804,
+      "step": 5645
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005963363593538557,
+      "loss": 3.0816,
+      "step": 5646
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005962082400011952,
+      "loss": 3.1125,
+      "step": 5647
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005960801140884633,
+      "loss": 3.2009,
+      "step": 5648
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005959519816243965,
+      "loss": 3.0791,
+      "step": 5649
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005958238426177318,
+      "loss": 3.0652,
+      "step": 5650
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005956956970772063,
+      "loss": 2.9989,
+      "step": 5651
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005955675450115579,
+      "loss": 2.9597,
+      "step": 5652
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005954393864295248,
+      "loss": 2.9488,
+      "step": 5653
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005953112213398455,
+      "loss": 3.1995,
+      "step": 5654
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005951830497512591,
+      "loss": 2.9595,
+      "step": 5655
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005950548716725054,
+      "loss": 3.1188,
+      "step": 5656
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000594926687112324,
+      "loss": 2.9309,
+      "step": 5657
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005947984960794556,
+      "loss": 3.1067,
+      "step": 5658
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005946702985826408,
+      "loss": 2.9729,
+      "step": 5659
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005945420946306211,
+      "loss": 3.1051,
+      "step": 5660
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005944138842321383,
+      "loss": 3.2853,
+      "step": 5661
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005942856673959344,
+      "loss": 3.0074,
+      "step": 5662
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000594157444130752,
+      "loss": 3.0439,
+      "step": 5663
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005940292144453341,
+      "loss": 3.1186,
+      "step": 5664
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005939009783484246,
+      "loss": 2.9899,
+      "step": 5665
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000593772735848767,
+      "loss": 3.1404,
+      "step": 5666
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005936444869551057,
+      "loss": 3.1838,
+      "step": 5667
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005935162316761858,
+      "loss": 3.0309,
+      "step": 5668
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005933879700207523,
+      "loss": 2.926,
+      "step": 5669
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005932597019975509,
+      "loss": 2.9488,
+      "step": 5670
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005931314276153277,
+      "loss": 2.9621,
+      "step": 5671
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005930031468828292,
+      "loss": 2.8648,
+      "step": 5672
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005928748598088024,
+      "loss": 3.1971,
+      "step": 5673
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005927465664019948,
+      "loss": 2.8595,
+      "step": 5674
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005926182666711542,
+      "loss": 2.8707,
+      "step": 5675
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005924899606250287,
+      "loss": 2.9238,
+      "step": 5676
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005923616482723672,
+      "loss": 2.895,
+      "step": 5677
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005922333296219189,
+      "loss": 2.7999,
+      "step": 5678
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005921050046824331,
+      "loss": 2.9466,
+      "step": 5679
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005919766734626599,
+      "loss": 3.0639,
+      "step": 5680
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005918483359713496,
+      "loss": 2.665,
+      "step": 5681
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005917199922172532,
+      "loss": 2.9833,
+      "step": 5682
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005915916422091221,
+      "loss": 2.8754,
+      "step": 5683
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005914632859557078,
+      "loss": 3.0534,
+      "step": 5684
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005913349234657623,
+      "loss": 3.0699,
+      "step": 5685
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005912065547480385,
+      "loss": 2.98,
+      "step": 5686
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005910781798112889,
+      "loss": 3.2958,
+      "step": 5687
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005909497986642673,
+      "loss": 3.1507,
+      "step": 5688
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005908214113157273,
+      "loss": 3.0555,
+      "step": 5689
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005906930177744232,
+      "loss": 2.9585,
+      "step": 5690
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005905646180491097,
+      "loss": 3.0818,
+      "step": 5691
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005904362121485421,
+      "loss": 3.1826,
+      "step": 5692
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005903078000814754,
+      "loss": 2.8884,
+      "step": 5693
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005901793818566659,
+      "loss": 3.1522,
+      "step": 5694
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005900509574828698,
+      "loss": 2.9482,
+      "step": 5695
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000589922526968844,
+      "loss": 2.8726,
+      "step": 5696
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005897940903233454,
+      "loss": 3.0236,
+      "step": 5697
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000589665647555132,
+      "loss": 3.134,
+      "step": 5698
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005895371986729617,
+      "loss": 3.1573,
+      "step": 5699
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005894087436855926,
+      "loss": 3.0201,
+      "step": 5700
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005892802826017841,
+      "loss": 3.0151,
+      "step": 5701
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005891518154302949,
+      "loss": 3.1356,
+      "step": 5702
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005890233421798851,
+      "loss": 3.1451,
+      "step": 5703
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005888948628593146,
+      "loss": 2.9016,
+      "step": 5704
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000588766377477344,
+      "loss": 3.0686,
+      "step": 5705
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005886378860427341,
+      "loss": 2.9288,
+      "step": 5706
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005885093885642465,
+      "loss": 2.8221,
+      "step": 5707
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005883808850506428,
+      "loss": 3.0802,
+      "step": 5708
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005882523755106851,
+      "loss": 3.0378,
+      "step": 5709
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000588123859953136,
+      "loss": 2.8427,
+      "step": 5710
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005879953383867584,
+      "loss": 3.0204,
+      "step": 5711
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000587866810820316,
+      "loss": 3.0576,
+      "step": 5712
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005877382772625724,
+      "loss": 3.0757,
+      "step": 5713
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005876097377222917,
+      "loss": 3.0826,
+      "step": 5714
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005874811922082387,
+      "loss": 2.8139,
+      "step": 5715
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005873526407291782,
+      "loss": 2.9518,
+      "step": 5716
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000587224083293876,
+      "loss": 3.1595,
+      "step": 5717
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005870955199110977,
+      "loss": 2.9346,
+      "step": 5718
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005869669505896095,
+      "loss": 2.9861,
+      "step": 5719
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005868383753381782,
+      "loss": 3.0115,
+      "step": 5720
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005867097941655707,
+      "loss": 2.9998,
+      "step": 5721
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005865812070805546,
+      "loss": 3.0436,
+      "step": 5722
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005864526140918978,
+      "loss": 3.1524,
+      "step": 5723
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005863240152083682,
+      "loss": 3.0892,
+      "step": 5724
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005861954104387349,
+      "loss": 2.9352,
+      "step": 5725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005860667997917668,
+      "loss": 3.2473,
+      "step": 5726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005859381832762333,
+      "loss": 2.9228,
+      "step": 5727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005858095609009044,
+      "loss": 3.0489,
+      "step": 5728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005856809326745501,
+      "loss": 3.1031,
+      "step": 5729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005855522986059415,
+      "loss": 3.0703,
+      "step": 5730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005854236587038492,
+      "loss": 3.1681,
+      "step": 5731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005852950129770451,
+      "loss": 3.0455,
+      "step": 5732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005851663614343008,
+      "loss": 2.867,
+      "step": 5733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005850377040843886,
+      "loss": 2.7243,
+      "step": 5734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005849090409360812,
+      "loss": 3.0102,
+      "step": 5735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005847803719981515,
+      "loss": 2.8896,
+      "step": 5736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005846516972793731,
+      "loss": 3.1058,
+      "step": 5737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005845230167885197,
+      "loss": 2.913,
+      "step": 5738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005843943305343657,
+      "loss": 2.7495,
+      "step": 5739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005842656385256856,
+      "loss": 3.1335,
+      "step": 5740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005841369407712544,
+      "loss": 2.9107,
+      "step": 5741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005840082372798476,
+      "loss": 3.0187,
+      "step": 5742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005838795280602409,
+      "loss": 2.6693,
+      "step": 5743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005837508131212106,
+      "loss": 3.1715,
+      "step": 5744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005836220924715331,
+      "loss": 3.2641,
+      "step": 5745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005834933661199857,
+      "loss": 2.9492,
+      "step": 5746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005833646340753455,
+      "loss": 2.8198,
+      "step": 5747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005832358963463902,
+      "loss": 2.9974,
+      "step": 5748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005831071529418979,
+      "loss": 2.946,
+      "step": 5749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005829784038706474,
+      "loss": 3.1128,
+      "step": 5750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005828496491414175,
+      "loss": 3.0593,
+      "step": 5751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005827208887629874,
+      "loss": 3.0976,
+      "step": 5752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005825921227441367,
+      "loss": 3.1045,
+      "step": 5753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005824633510936457,
+      "loss": 2.7405,
+      "step": 5754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005823345738202947,
+      "loss": 3.1938,
+      "step": 5755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005822057909328645,
+      "loss": 3.2054,
+      "step": 5756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005820770024401363,
+      "loss": 2.9751,
+      "step": 5757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005819482083508919,
+      "loss": 3.166,
+      "step": 5758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000581819408673913,
+      "loss": 3.1323,
+      "step": 5759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005816906034179822,
+      "loss": 3.1678,
+      "step": 5760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005815617925918822,
+      "loss": 2.6353,
+      "step": 5761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005814329762043959,
+      "loss": 2.9411,
+      "step": 5762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000581304154264307,
+      "loss": 2.8864,
+      "step": 5763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005811753267803991,
+      "loss": 3.0077,
+      "step": 5764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.000581046493761457,
+      "loss": 2.9787,
+      "step": 5765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005809176552162648,
+      "loss": 3.3542,
+      "step": 5766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0005807888111536079,
+      "loss": 3.2285,
+      "step": 5767
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005806599615822714,
+      "loss": 2.8678,
+      "step": 5768
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005805311065110409,
+      "loss": 2.9569,
+      "step": 5769
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005804022459487031,
+      "loss": 3.0679,
+      "step": 5770
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005802733799040438,
+      "loss": 2.9318,
+      "step": 5771
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005801445083858506,
+      "loss": 2.8444,
+      "step": 5772
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005800156314029102,
+      "loss": 2.9184,
+      "step": 5773
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005798867489640108,
+      "loss": 2.9688,
+      "step": 5774
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005797578610779397,
+      "loss": 2.9492,
+      "step": 5775
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005796289677534857,
+      "loss": 2.7282,
+      "step": 5776
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005795000689994376,
+      "loss": 2.9403,
+      "step": 5777
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005793711648245843,
+      "loss": 2.8772,
+      "step": 5778
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005792422552377152,
+      "loss": 3.1409,
+      "step": 5779
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005791133402476206,
+      "loss": 3.1533,
+      "step": 5780
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005789844198630901,
+      "loss": 2.8426,
+      "step": 5781
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005788554940929148,
+      "loss": 2.9254,
+      "step": 5782
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005787265629458855,
+      "loss": 3.0667,
+      "step": 5783
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005785976264307936,
+      "loss": 3.1068,
+      "step": 5784
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005784686845564304,
+      "loss": 3.3962,
+      "step": 5785
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005783397373315885,
+      "loss": 3.0896,
+      "step": 5786
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005782107847650597,
+      "loss": 3.135,
+      "step": 5787
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005780818268656373,
+      "loss": 3.0414,
+      "step": 5788
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005779528636421143,
+      "loss": 3.1082,
+      "step": 5789
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005778238951032841,
+      "loss": 2.8468,
+      "step": 5790
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005776949212579408,
+      "loss": 2.9626,
+      "step": 5791
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005775659421148785,
+      "loss": 3.2197,
+      "step": 5792
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005774369576828916,
+      "loss": 3.0841,
+      "step": 5793
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005773079679707753,
+      "loss": 3.0651,
+      "step": 5794
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005771789729873248,
+      "loss": 2.8203,
+      "step": 5795
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000577049972741336,
+      "loss": 3.0795,
+      "step": 5796
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005769209672416046,
+      "loss": 2.9124,
+      "step": 5797
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005767919564969271,
+      "loss": 3.1523,
+      "step": 5798
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005766629405161006,
+      "loss": 3.039,
+      "step": 5799
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005765339193079216,
+      "loss": 3.2148,
+      "step": 5800
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005764048928811879,
+      "loss": 3.1114,
+      "step": 5801
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005762758612446973,
+      "loss": 2.9539,
+      "step": 5802
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000576146824407248,
+      "loss": 3.1393,
+      "step": 5803
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005760177823776383,
+      "loss": 2.8729,
+      "step": 5804
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005758887351646674,
+      "loss": 2.9917,
+      "step": 5805
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005757596827771345,
+      "loss": 3.069,
+      "step": 5806
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005756306252238389,
+      "loss": 3.037,
+      "step": 5807
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005755015625135809,
+      "loss": 3.1029,
+      "step": 5808
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005753724946551602,
+      "loss": 2.8313,
+      "step": 5809
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005752434216573783,
+      "loss": 3.1375,
+      "step": 5810
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005751143435290356,
+      "loss": 3.0128,
+      "step": 5811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005749852602789336,
+      "loss": 3.0913,
+      "step": 5812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000574856171915874,
+      "loss": 2.8699,
+      "step": 5813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005747270784486588,
+      "loss": 3.0861,
+      "step": 5814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005745979798860904,
+      "loss": 2.9609,
+      "step": 5815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005744688762369716,
+      "loss": 2.8802,
+      "step": 5816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005743397675101053,
+      "loss": 3.0032,
+      "step": 5817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005742106537142952,
+      "loss": 2.9878,
+      "step": 5818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005740815348583449,
+      "loss": 2.8407,
+      "step": 5819
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005739524109510585,
+      "loss": 3.1329,
+      "step": 5820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005738232820012407,
+      "loss": 2.9637,
+      "step": 5821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000573694148017696,
+      "loss": 3.0492,
+      "step": 5822
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005735650090092297,
+      "loss": 2.9781,
+      "step": 5823
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005734358649846471,
+      "loss": 3.2257,
+      "step": 5824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005733067159527545,
+      "loss": 3.0061,
+      "step": 5825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005731775619223576,
+      "loss": 2.9054,
+      "step": 5826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005730484029022632,
+      "loss": 3.1401,
+      "step": 5827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005729192389012781,
+      "loss": 2.9005,
+      "step": 5828
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005727900699282095,
+      "loss": 2.8331,
+      "step": 5829
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005726608959918648,
+      "loss": 2.9731,
+      "step": 5830
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000572531717101052,
+      "loss": 2.9127,
+      "step": 5831
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005724025332645794,
+      "loss": 2.6893,
+      "step": 5832
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005722733444912554,
+      "loss": 3.0647,
+      "step": 5833
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005721441507898888,
+      "loss": 3.0075,
+      "step": 5834
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000572014952169289,
+      "loss": 3.2686,
+      "step": 5835
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005718857486382656,
+      "loss": 3.1361,
+      "step": 5836
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005717565402056284,
+      "loss": 2.7286,
+      "step": 5837
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005716273268801876,
+      "loss": 2.8446,
+      "step": 5838
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005714981086707536,
+      "loss": 2.7529,
+      "step": 5839
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005713688855861378,
+      "loss": 2.8581,
+      "step": 5840
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000571239657635151,
+      "loss": 2.9858,
+      "step": 5841
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005711104248266048,
+      "loss": 3.03,
+      "step": 5842
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005709811871693113,
+      "loss": 3.0959,
+      "step": 5843
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005708519446720826,
+      "loss": 3.1685,
+      "step": 5844
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005707226973437311,
+      "loss": 2.8536,
+      "step": 5845
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00057059344519307,
+      "loss": 2.8395,
+      "step": 5846
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005704641882289121,
+      "loss": 3.1978,
+      "step": 5847
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005703349264600713,
+      "loss": 3.1134,
+      "step": 5848
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005702056598953613,
+      "loss": 2.8879,
+      "step": 5849
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005700763885435965,
+      "loss": 3.1268,
+      "step": 5850
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000569947112413591,
+      "loss": 2.667,
+      "step": 5851
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00056981783151416,
+      "loss": 3.1777,
+      "step": 5852
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005696885458541186,
+      "loss": 3.0338,
+      "step": 5853
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000569559255442282,
+      "loss": 3.2131,
+      "step": 5854
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005694299602874665,
+      "loss": 3.0228,
+      "step": 5855
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000569300660398488,
+      "loss": 3.0244,
+      "step": 5856
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005691713557841629,
+      "loss": 2.8206,
+      "step": 5857
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005690420464533082,
+      "loss": 3.0264,
+      "step": 5858
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005689127324147406,
+      "loss": 2.971,
+      "step": 5859
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000568783413677278,
+      "loss": 2.9788,
+      "step": 5860
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000568654090249738,
+      "loss": 3.0775,
+      "step": 5861
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005685247621409383,
+      "loss": 2.8773,
+      "step": 5862
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005683954293596978,
+      "loss": 2.9709,
+      "step": 5863
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005682660919148351,
+      "loss": 2.9459,
+      "step": 5864
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005681367498151691,
+      "loss": 3.0543,
+      "step": 5865
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005680074030695191,
+      "loss": 3.2391,
+      "step": 5866
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005678780516867049,
+      "loss": 2.9217,
+      "step": 5867
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005677486956755463,
+      "loss": 2.9231,
+      "step": 5868
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005676193350448637,
+      "loss": 3.1388,
+      "step": 5869
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005674899698034778,
+      "loss": 3.0505,
+      "step": 5870
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005673605999602094,
+      "loss": 3.2172,
+      "step": 5871
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005672312255238797,
+      "loss": 3.149,
+      "step": 5872
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005671018465033105,
+      "loss": 3.0712,
+      "step": 5873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005669724629073233,
+      "loss": 2.926,
+      "step": 5874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005668430747447405,
+      "loss": 3.1282,
+      "step": 5875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005667136820243845,
+      "loss": 2.9136,
+      "step": 5876
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000566584284755078,
+      "loss": 3.1239,
+      "step": 5877
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005664548829456443,
+      "loss": 3.0134,
+      "step": 5878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005663254766049067,
+      "loss": 2.8724,
+      "step": 5879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005661960657416891,
+      "loss": 3.0553,
+      "step": 5880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005660666503648153,
+      "loss": 2.9773,
+      "step": 5881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005659372304831098,
+      "loss": 2.9566,
+      "step": 5882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000565807806105397,
+      "loss": 3.1237,
+      "step": 5883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005656783772405021,
+      "loss": 3.0474,
+      "step": 5884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005655489438972503,
+      "loss": 2.8393,
+      "step": 5885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005654195060844671,
+      "loss": 2.9752,
+      "step": 5886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005652900638109785,
+      "loss": 2.897,
+      "step": 5887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005651606170856105,
+      "loss": 2.9582,
+      "step": 5888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005650311659171898,
+      "loss": 3.1493,
+      "step": 5889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.000564901710314543,
+      "loss": 3.0261,
+      "step": 5890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0005647722502864971,
+      "loss": 2.7855,
+      "step": 5891
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005646427858418797,
+      "loss": 3.0724,
+      "step": 5892
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005645133169895184,
+      "loss": 3.1357,
+      "step": 5893
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005643838437382412,
+      "loss": 2.6257,
+      "step": 5894
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005642543660968762,
+      "loss": 3.0608,
+      "step": 5895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005641248840742523,
+      "loss": 2.9719,
+      "step": 5896
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005639953976791982,
+      "loss": 2.7742,
+      "step": 5897
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005638659069205433,
+      "loss": 2.853,
+      "step": 5898
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005637364118071165,
+      "loss": 3.1623,
+      "step": 5899
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005636069123477482,
+      "loss": 2.9146,
+      "step": 5900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005634774085512682,
+      "loss": 3.018,
+      "step": 5901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000563347900426507,
+      "loss": 3.0844,
+      "step": 5902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005632183879822949,
+      "loss": 3.0582,
+      "step": 5903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005630888712274634,
+      "loss": 2.8493,
+      "step": 5904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005629593501708434,
+      "loss": 2.7752,
+      "step": 5905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005628298248212666,
+      "loss": 3.0103,
+      "step": 5906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005627002951875645,
+      "loss": 3.0357,
+      "step": 5907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005625707612785697,
+      "loss": 3.0284,
+      "step": 5908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005624412231031143,
+      "loss": 3.04,
+      "step": 5909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005623116806700311,
+      "loss": 2.8818,
+      "step": 5910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005621821339881533,
+      "loss": 2.9891,
+      "step": 5911
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005620525830663139,
+      "loss": 3.1469,
+      "step": 5912
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005619230279133466,
+      "loss": 2.9995,
+      "step": 5913
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005617934685380852,
+      "loss": 2.9937,
+      "step": 5914
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000561663904949364,
+      "loss": 2.868,
+      "step": 5915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005615343371560173,
+      "loss": 3.1863,
+      "step": 5916
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005614047651668801,
+      "loss": 3.0308,
+      "step": 5917
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000561275188990787,
+      "loss": 2.9367,
+      "step": 5918
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005611456086365736,
+      "loss": 3.0469,
+      "step": 5919
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005610160241130755,
+      "loss": 2.9899,
+      "step": 5920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005608864354291283,
+      "loss": 2.8116,
+      "step": 5921
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005607568425935684,
+      "loss": 3.1515,
+      "step": 5922
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005606272456152321,
+      "loss": 2.97,
+      "step": 5923
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005604976445029564,
+      "loss": 3.0389,
+      "step": 5924
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000560368039265578,
+      "loss": 3.0375,
+      "step": 5925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005602384299119343,
+      "loss": 2.7945,
+      "step": 5926
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005601088164508628,
+      "loss": 3.0357,
+      "step": 5927
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005599791988912014,
+      "loss": 2.9927,
+      "step": 5928
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005598495772417883,
+      "loss": 2.9889,
+      "step": 5929
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005597199515114617,
+      "loss": 2.9479,
+      "step": 5930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005595903217090605,
+      "loss": 2.9876,
+      "step": 5931
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005594606878434236,
+      "loss": 2.8766,
+      "step": 5932
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005593310499233901,
+      "loss": 2.6715,
+      "step": 5933
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005592014079577998,
+      "loss": 2.7092,
+      "step": 5934
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000559071761955492,
+      "loss": 3.3391,
+      "step": 5935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005589421119253072,
+      "loss": 3.0016,
+      "step": 5936
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005588124578760856,
+      "loss": 3.2178,
+      "step": 5937
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005586827998166678,
+      "loss": 3.0535,
+      "step": 5938
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005585531377558947,
+      "loss": 2.8737,
+      "step": 5939
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005584234717026077,
+      "loss": 2.9523,
+      "step": 5940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005582938016656477,
+      "loss": 2.9556,
+      "step": 5941
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005581641276538569,
+      "loss": 3.0005,
+      "step": 5942
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000558034449676077,
+      "loss": 3.0555,
+      "step": 5943
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005579047677411502,
+      "loss": 2.9392,
+      "step": 5944
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005577750818579191,
+      "loss": 2.7005,
+      "step": 5945
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005576453920352268,
+      "loss": 2.9999,
+      "step": 5946
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005575156982819159,
+      "loss": 3.0036,
+      "step": 5947
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005573860006068299,
+      "loss": 3.1414,
+      "step": 5948
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005572562990188123,
+      "loss": 2.9732,
+      "step": 5949
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005571265935267072,
+      "loss": 3.0541,
+      "step": 5950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005569968841393584,
+      "loss": 2.967,
+      "step": 5951
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005568671708656105,
+      "loss": 2.8897,
+      "step": 5952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005567374537143081,
+      "loss": 2.8096,
+      "step": 5953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000556607732694296,
+      "loss": 2.851,
+      "step": 5954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005564780078144196,
+      "loss": 2.8664,
+      "step": 5955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005563482790835242,
+      "loss": 3.2696,
+      "step": 5956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005562185465104555,
+      "loss": 3.0073,
+      "step": 5957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005560888101040595,
+      "loss": 2.8516,
+      "step": 5958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005559590698731826,
+      "loss": 2.9911,
+      "step": 5959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005558293258266707,
+      "loss": 3.2009,
+      "step": 5960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005556995779733713,
+      "loss": 2.9892,
+      "step": 5961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005555698263221311,
+      "loss": 2.7766,
+      "step": 5962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005554400708817972,
+      "loss": 3.0352,
+      "step": 5963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005553103116612174,
+      "loss": 2.977,
+      "step": 5964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005551805486692393,
+      "loss": 2.9935,
+      "step": 5965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005550507819147108,
+      "loss": 2.851,
+      "step": 5966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005549210114064807,
+      "loss": 2.7943,
+      "step": 5967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000554791237153397,
+      "loss": 3.0097,
+      "step": 5968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005546614591643089,
+      "loss": 2.8113,
+      "step": 5969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005545316774480652,
+      "loss": 3.0326,
+      "step": 5970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005544018920135154,
+      "loss": 3.057,
+      "step": 5971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000554272102869509,
+      "loss": 2.8331,
+      "step": 5972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005541423100248959,
+      "loss": 2.9319,
+      "step": 5973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000554012513488526,
+      "loss": 2.9261,
+      "step": 5974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005538827132692498,
+      "loss": 2.8481,
+      "step": 5975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005537529093759177,
+      "loss": 2.8911,
+      "step": 5976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005536231018173809,
+      "loss": 3.0222,
+      "step": 5977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005534932906024899,
+      "loss": 3.0657,
+      "step": 5978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005533634757400967,
+      "loss": 2.8634,
+      "step": 5979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005532336572390524,
+      "loss": 2.9555,
+      "step": 5980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005531038351082089,
+      "loss": 2.8999,
+      "step": 5981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005529740093564183,
+      "loss": 3.1831,
+      "step": 5982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000552844179992533,
+      "loss": 3.0836,
+      "step": 5983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005527143470254055,
+      "loss": 2.9414,
+      "step": 5984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005525845104638887,
+      "loss": 2.8946,
+      "step": 5985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005524546703168356,
+      "loss": 2.7784,
+      "step": 5986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005523248265930995,
+      "loss": 3.1306,
+      "step": 5987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000552194979301534,
+      "loss": 3.0081,
+      "step": 5988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005520651284509927,
+      "loss": 2.9802,
+      "step": 5989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005519352740503297,
+      "loss": 2.9926,
+      "step": 5990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005518054161083994,
+      "loss": 3.0775,
+      "step": 5991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005516755546340563,
+      "loss": 2.9844,
+      "step": 5992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005515456896361551,
+      "loss": 3.1002,
+      "step": 5993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005514158211235509,
+      "loss": 3.1947,
+      "step": 5994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005512859491050986,
+      "loss": 3.0396,
+      "step": 5995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005511560735896541,
+      "loss": 2.958,
+      "step": 5996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000551026194586073,
+      "loss": 3.0902,
+      "step": 5997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000550896312103211,
+      "loss": 2.9372,
+      "step": 5998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005507664261499245,
+      "loss": 3.0249,
+      "step": 5999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005506365367350701,
+      "loss": 3.0328,
+      "step": 6000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000550506643867504,
+      "loss": 2.938,
+      "step": 6001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005503767475560834,
+      "loss": 3.0515,
+      "step": 6002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005502468478096656,
+      "loss": 2.8952,
+      "step": 6003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005501169446371075,
+      "loss": 2.9534,
+      "step": 6004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.000549987038047267,
+      "loss": 2.9808,
+      "step": 6005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005498571280490019,
+      "loss": 3.1083,
+      "step": 6006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005497272146511704,
+      "loss": 3.0483,
+      "step": 6007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005495972978626305,
+      "loss": 3.0248,
+      "step": 6008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005494673776922409,
+      "loss": 2.9257,
+      "step": 6009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005493374541488603,
+      "loss": 3.0904,
+      "step": 6010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005492075272413478,
+      "loss": 2.934,
+      "step": 6011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005490775969785624,
+      "loss": 2.8267,
+      "step": 6012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005489476633693638,
+      "loss": 2.9871,
+      "step": 6013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005488177264226117,
+      "loss": 3.0636,
+      "step": 6014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0005486877861471657,
+      "loss": 3.083,
+      "step": 6015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000548557842551886,
+      "loss": 3.0461,
+      "step": 6016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005484278956456333,
+      "loss": 3.0017,
+      "step": 6017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005482979454372678,
+      "loss": 3.0884,
+      "step": 6018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005481679919356504,
+      "loss": 3.1136,
+      "step": 6019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005480380351496421,
+      "loss": 3.1733,
+      "step": 6020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005479080750881044,
+      "loss": 2.7359,
+      "step": 6021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005477781117598985,
+      "loss": 3.0209,
+      "step": 6022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005476481451738862,
+      "loss": 3.0704,
+      "step": 6023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005475181753389296,
+      "loss": 2.8783,
+      "step": 6024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005473882022638905,
+      "loss": 3.2626,
+      "step": 6025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005472582259576315,
+      "loss": 3.1202,
+      "step": 6026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005471282464290151,
+      "loss": 3.0008,
+      "step": 6027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005469982636869042,
+      "loss": 2.9931,
+      "step": 6028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005468682777401616,
+      "loss": 3.093,
+      "step": 6029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005467382885976509,
+      "loss": 2.9364,
+      "step": 6030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005466082962682353,
+      "loss": 2.901,
+      "step": 6031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005464783007607786,
+      "loss": 3.0558,
+      "step": 6032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005463483020841445,
+      "loss": 2.9282,
+      "step": 6033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005462183002471972,
+      "loss": 3.0244,
+      "step": 6034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005460882952588013,
+      "loss": 2.7647,
+      "step": 6035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005459582871278209,
+      "loss": 3.0467,
+      "step": 6036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005458282758631211,
+      "loss": 2.6642,
+      "step": 6037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005456982614735667,
+      "loss": 3.0634,
+      "step": 6038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005455682439680229,
+      "loss": 3.1768,
+      "step": 6039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005454382233553553,
+      "loss": 3.0884,
+      "step": 6040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005453081996444292,
+      "loss": 2.9948,
+      "step": 6041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005451781728441104,
+      "loss": 3.0565,
+      "step": 6042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005450481429632654,
+      "loss": 2.8353,
+      "step": 6043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005449181100107599,
+      "loss": 3.023,
+      "step": 6044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005447880739954606,
+      "loss": 3.2158,
+      "step": 6045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005446580349262343,
+      "loss": 3.0718,
+      "step": 6046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005445279928119474,
+      "loss": 2.9871,
+      "step": 6047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005443979476614675,
+      "loss": 2.7593,
+      "step": 6048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005442678994836615,
+      "loss": 2.896,
+      "step": 6049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000544137848287397,
+      "loss": 3.0285,
+      "step": 6050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005440077940815418,
+      "loss": 2.9111,
+      "step": 6051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005438777368749637,
+      "loss": 2.7234,
+      "step": 6052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000543747676676531,
+      "loss": 3.1724,
+      "step": 6053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005436176134951116,
+      "loss": 3.0747,
+      "step": 6054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005434875473395744,
+      "loss": 2.9849,
+      "step": 6055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005433574782187879,
+      "loss": 3.0316,
+      "step": 6056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005432274061416212,
+      "loss": 2.9207,
+      "step": 6057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000543097331116943,
+      "loss": 3.1069,
+      "step": 6058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005429672531536231,
+      "loss": 3.1375,
+      "step": 6059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005428371722605307,
+      "loss": 3.0606,
+      "step": 6060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005427070884465359,
+      "loss": 2.9197,
+      "step": 6061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005425770017205082,
+      "loss": 2.9388,
+      "step": 6062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005424469120913178,
+      "loss": 2.9333,
+      "step": 6063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005423168195678351,
+      "loss": 2.7916,
+      "step": 6064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005421867241589307,
+      "loss": 3.0956,
+      "step": 6065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005420566258734752,
+      "loss": 2.9948,
+      "step": 6066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005419265247203393,
+      "loss": 3.0976,
+      "step": 6067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005417964207083945,
+      "loss": 3.0008,
+      "step": 6068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005416663138465119,
+      "loss": 3.2832,
+      "step": 6069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000541536204143563,
+      "loss": 3.0116,
+      "step": 6070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005414060916084194,
+      "loss": 2.923,
+      "step": 6071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005412759762499531,
+      "loss": 2.9317,
+      "step": 6072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000541145858077036,
+      "loss": 2.9275,
+      "step": 6073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005410157370985406,
+      "loss": 3.0273,
+      "step": 6074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005408856133233392,
+      "loss": 3.1435,
+      "step": 6075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005407554867603046,
+      "loss": 3.114,
+      "step": 6076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005406253574183092,
+      "loss": 3.1829,
+      "step": 6077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005404952253062265,
+      "loss": 3.1267,
+      "step": 6078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005403650904329296,
+      "loss": 3.0419,
+      "step": 6079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005402349528072916,
+      "loss": 3.0918,
+      "step": 6080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005401048124381864,
+      "loss": 3.1143,
+      "step": 6081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005399746693344878,
+      "loss": 2.8425,
+      "step": 6082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005398445235050696,
+      "loss": 2.8963,
+      "step": 6083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000539714374958806,
+      "loss": 2.7588,
+      "step": 6084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005395842237045714,
+      "loss": 2.8873,
+      "step": 6085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005394540697512401,
+      "loss": 2.8747,
+      "step": 6086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000539323913107687,
+      "loss": 2.9117,
+      "step": 6087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005391937537827869,
+      "loss": 3.0185,
+      "step": 6088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005390635917854148,
+      "loss": 3.1453,
+      "step": 6089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005389334271244463,
+      "loss": 2.8427,
+      "step": 6090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005388032598087563,
+      "loss": 2.9396,
+      "step": 6091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005386730898472208,
+      "loss": 2.9686,
+      "step": 6092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005385429172487155,
+      "loss": 2.9722,
+      "step": 6093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005384127420221163,
+      "loss": 3.045,
+      "step": 6094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005382825641762994,
+      "loss": 2.8726,
+      "step": 6095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005381523837201411,
+      "loss": 2.7897,
+      "step": 6096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005380222006625179,
+      "loss": 2.8975,
+      "step": 6097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005378920150123066,
+      "loss": 3.1133,
+      "step": 6098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005377618267783839,
+      "loss": 3.0239,
+      "step": 6099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005376316359696268,
+      "loss": 3.047,
+      "step": 6100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005375014425949129,
+      "loss": 2.8661,
+      "step": 6101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005373712466631189,
+      "loss": 2.8551,
+      "step": 6102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000537241048183123,
+      "loss": 3.0257,
+      "step": 6103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005371108471638028,
+      "loss": 3.1016,
+      "step": 6104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005369806436140359,
+      "loss": 2.9038,
+      "step": 6105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005368504375427007,
+      "loss": 2.9808,
+      "step": 6106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005367202289586753,
+      "loss": 3.0506,
+      "step": 6107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005365900178708383,
+      "loss": 3.0251,
+      "step": 6108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005364598042880681,
+      "loss": 3.0204,
+      "step": 6109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005363295882192434,
+      "loss": 3.0631,
+      "step": 6110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005361993696732433,
+      "loss": 3.0695,
+      "step": 6111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005360691486589469,
+      "loss": 2.8251,
+      "step": 6112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005359389251852336,
+      "loss": 2.9582,
+      "step": 6113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005358086992609825,
+      "loss": 3.1406,
+      "step": 6114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005356784708950735,
+      "loss": 2.7772,
+      "step": 6115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005355482400963863,
+      "loss": 2.9459,
+      "step": 6116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005354180068738006,
+      "loss": 2.9,
+      "step": 6117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005352877712361967,
+      "loss": 3.1585,
+      "step": 6118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005351575331924551,
+      "loss": 2.7018,
+      "step": 6119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005350272927514558,
+      "loss": 3.1719,
+      "step": 6120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005348970499220798,
+      "loss": 3.0231,
+      "step": 6121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005347668047132075,
+      "loss": 3.0826,
+      "step": 6122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00053463655713372,
+      "loss": 3.1227,
+      "step": 6123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005345063071924985,
+      "loss": 3.0203,
+      "step": 6124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000534376054898424,
+      "loss": 2.945,
+      "step": 6125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000534245800260378,
+      "loss": 2.9504,
+      "step": 6126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005341155432872422,
+      "loss": 2.929,
+      "step": 6127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005339852839878981,
+      "loss": 3.1025,
+      "step": 6128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005338550223712278,
+      "loss": 3.0349,
+      "step": 6129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005337247584461132,
+      "loss": 3.0872,
+      "step": 6130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005335944922214364,
+      "loss": 3.0723,
+      "step": 6131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005334642237060798,
+      "loss": 2.9685,
+      "step": 6132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005333339529089261,
+      "loss": 2.9492,
+      "step": 6133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.000533203679838858,
+      "loss": 3.0966,
+      "step": 6134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005330734045047582,
+      "loss": 3.1448,
+      "step": 6135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005329431269155095,
+      "loss": 3.1975,
+      "step": 6136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005328128470799953,
+      "loss": 3.1583,
+      "step": 6137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005326825650070987,
+      "loss": 2.9912,
+      "step": 6138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0005325522807057032,
+      "loss": 2.8914,
+      "step": 6139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005324219941846923,
+      "loss": 3.2525,
+      "step": 6140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005322917054529499,
+      "loss": 2.7987,
+      "step": 6141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005321614145193599,
+      "loss": 3.0774,
+      "step": 6142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005320311213928062,
+      "loss": 2.9803,
+      "step": 6143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000531900826082173,
+      "loss": 3.2477,
+      "step": 6144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005317705285963447,
+      "loss": 2.8772,
+      "step": 6145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005316402289442059,
+      "loss": 2.7431,
+      "step": 6146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005315099271346411,
+      "loss": 2.8936,
+      "step": 6147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000531379623176535,
+      "loss": 3.0685,
+      "step": 6148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005312493170787725,
+      "loss": 3.1796,
+      "step": 6149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000531119008850239,
+      "loss": 2.8239,
+      "step": 6150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005309886984998194,
+      "loss": 3.043,
+      "step": 6151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005308583860363994,
+      "loss": 2.8464,
+      "step": 6152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005307280714688641,
+      "loss": 3.2358,
+      "step": 6153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005305977548060993,
+      "loss": 3.0905,
+      "step": 6154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005304674360569908,
+      "loss": 3.0924,
+      "step": 6155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005303371152304245,
+      "loss": 2.9917,
+      "step": 6156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005302067923352867,
+      "loss": 3.02,
+      "step": 6157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005300764673804634,
+      "loss": 3.1305,
+      "step": 6158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000529946140374841,
+      "loss": 2.8854,
+      "step": 6159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005298158113273061,
+      "loss": 2.8817,
+      "step": 6160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005296854802467454,
+      "loss": 3.2072,
+      "step": 6161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005295551471420452,
+      "loss": 3.1316,
+      "step": 6162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005294248120220928,
+      "loss": 2.9297,
+      "step": 6163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005292944748957753,
+      "loss": 2.99,
+      "step": 6164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005291641357719799,
+      "loss": 3.0233,
+      "step": 6165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005290337946595937,
+      "loss": 2.931,
+      "step": 6166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005289034515675041,
+      "loss": 3.0877,
+      "step": 6167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005287731065045992,
+      "loss": 2.9662,
+      "step": 6168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005286427594797663,
+      "loss": 2.965,
+      "step": 6169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005285124105018932,
+      "loss": 2.9357,
+      "step": 6170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000528382059579868,
+      "loss": 3.0345,
+      "step": 6171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005282517067225792,
+      "loss": 2.743,
+      "step": 6172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005281213519389146,
+      "loss": 2.9177,
+      "step": 6173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005279909952377627,
+      "loss": 3.1476,
+      "step": 6174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005278606366280122,
+      "loss": 2.7666,
+      "step": 6175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005277302761185517,
+      "loss": 2.8244,
+      "step": 6176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005275999137182698,
+      "loss": 2.753,
+      "step": 6177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005274695494360553,
+      "loss": 3.013,
+      "step": 6178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005273391832807977,
+      "loss": 2.9557,
+      "step": 6179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005272088152613859,
+      "loss": 2.8904,
+      "step": 6180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005270784453867092,
+      "loss": 2.9182,
+      "step": 6181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005269480736656571,
+      "loss": 2.8572,
+      "step": 6182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005268177001071192,
+      "loss": 2.8309,
+      "step": 6183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005266873247199849,
+      "loss": 3.0925,
+      "step": 6184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005265569475131443,
+      "loss": 2.8545,
+      "step": 6185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000526426568495487,
+      "loss": 3.0508,
+      "step": 6186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005262961876759033,
+      "loss": 3.0325,
+      "step": 6187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005261658050632835,
+      "loss": 2.8544,
+      "step": 6188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005260354206665176,
+      "loss": 3.1223,
+      "step": 6189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000525905034494496,
+      "loss": 2.9885,
+      "step": 6190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005257746465561096,
+      "loss": 3.209,
+      "step": 6191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005256442568602485,
+      "loss": 2.8165,
+      "step": 6192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005255138654158039,
+      "loss": 2.8772,
+      "step": 6193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005253834722316666,
+      "loss": 3.0525,
+      "step": 6194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005252530773167275,
+      "loss": 3.1178,
+      "step": 6195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005251226806798779,
+      "loss": 2.9487,
+      "step": 6196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005249922823300089,
+      "loss": 2.9966,
+      "step": 6197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000524861882276012,
+      "loss": 2.9706,
+      "step": 6198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005247314805267786,
+      "loss": 3.1152,
+      "step": 6199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005246010770911999,
+      "loss": 2.9609,
+      "step": 6200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005244706719781684,
+      "loss": 3.0155,
+      "step": 6201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005243402651965755,
+      "loss": 3.0371,
+      "step": 6202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005242098567553132,
+      "loss": 2.822,
+      "step": 6203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005240794466632733,
+      "loss": 3.2303,
+      "step": 6204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005239490349293484,
+      "loss": 2.8358,
+      "step": 6205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005238186215624305,
+      "loss": 3.1591,
+      "step": 6206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000523688206571412,
+      "loss": 2.9121,
+      "step": 6207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005235577899651854,
+      "loss": 2.8242,
+      "step": 6208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005234273717526436,
+      "loss": 3.0756,
+      "step": 6209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005232969519426788,
+      "loss": 3.2281,
+      "step": 6210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005231665305441843,
+      "loss": 3.0643,
+      "step": 6211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005230361075660527,
+      "loss": 2.9318,
+      "step": 6212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005229056830171773,
+      "loss": 2.9713,
+      "step": 6213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000522775256906451,
+      "loss": 2.8571,
+      "step": 6214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005226448292427674,
+      "loss": 2.9693,
+      "step": 6215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005225144000350193,
+      "loss": 2.9214,
+      "step": 6216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005223839692921008,
+      "loss": 3.0781,
+      "step": 6217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000522253537022905,
+      "loss": 3.0709,
+      "step": 6218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005221231032363259,
+      "loss": 2.9857,
+      "step": 6219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005219926679412572,
+      "loss": 3.0562,
+      "step": 6220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005218622311465928,
+      "loss": 2.8962,
+      "step": 6221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005217317928612264,
+      "loss": 3.0012,
+      "step": 6222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005216013530940523,
+      "loss": 2.8892,
+      "step": 6223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005214709118539646,
+      "loss": 3.2173,
+      "step": 6224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005213404691498581,
+      "loss": 2.9517,
+      "step": 6225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005212100249906266,
+      "loss": 2.9574,
+      "step": 6226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005210795793851645,
+      "loss": 2.8843,
+      "step": 6227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005209491323423669,
+      "loss": 2.916,
+      "step": 6228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005208186838711282,
+      "loss": 3.0592,
+      "step": 6229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005206882339803431,
+      "loss": 3.1264,
+      "step": 6230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005205577826789067,
+      "loss": 3.1232,
+      "step": 6231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005204273299757137,
+      "loss": 3.159,
+      "step": 6232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005202968758796594,
+      "loss": 2.9056,
+      "step": 6233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000520166420399639,
+      "loss": 2.7392,
+      "step": 6234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005200359635445475,
+      "loss": 3.0669,
+      "step": 6235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005199055053232805,
+      "loss": 2.8566,
+      "step": 6236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005197750457447334,
+      "loss": 2.948,
+      "step": 6237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005196445848178015,
+      "loss": 3.174,
+      "step": 6238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005195141225513807,
+      "loss": 2.7698,
+      "step": 6239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005193836589543668,
+      "loss": 2.7378,
+      "step": 6240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005192531940356554,
+      "loss": 3.0888,
+      "step": 6241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005191227278041424,
+      "loss": 3.2048,
+      "step": 6242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005189922602687238,
+      "loss": 3.209,
+      "step": 6243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005188617914382959,
+      "loss": 3.1902,
+      "step": 6244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005187313213217546,
+      "loss": 2.8551,
+      "step": 6245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005186008499279962,
+      "loss": 2.8824,
+      "step": 6246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005184703772659171,
+      "loss": 2.7532,
+      "step": 6247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000518339903344414,
+      "loss": 3.0786,
+      "step": 6248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005182094281723829,
+      "loss": 3.2235,
+      "step": 6249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005180789517587208,
+      "loss": 3.0815,
+      "step": 6250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005179484741123241,
+      "loss": 3.0621,
+      "step": 6251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005178179952420899,
+      "loss": 3.0375,
+      "step": 6252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005176875151569147,
+      "loss": 2.9483,
+      "step": 6253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005175570338656955,
+      "loss": 2.9195,
+      "step": 6254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005174265513773296,
+      "loss": 2.8981,
+      "step": 6255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000517296067700714,
+      "loss": 3.156,
+      "step": 6256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005171655828447457,
+      "loss": 3.03,
+      "step": 6257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005170350968183221,
+      "loss": 2.7686,
+      "step": 6258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005169046096303406,
+      "loss": 3.0066,
+      "step": 6259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005167741212896983,
+      "loss": 3.0337,
+      "step": 6260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005166436318052932,
+      "loss": 2.7721,
+      "step": 6261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005165131411860224,
+      "loss": 2.89,
+      "step": 6262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005163826494407839,
+      "loss": 2.9965,
+      "step": 6263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.0005162521565784753,
+      "loss": 2.9825,
+      "step": 6264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005161216626079943,
+      "loss": 3.0961,
+      "step": 6265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005159911675382391,
+      "loss": 3.1141,
+      "step": 6266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005158606713781072,
+      "loss": 3.1464,
+      "step": 6267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005157301741364969,
+      "loss": 3.0841,
+      "step": 6268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005155996758223064,
+      "loss": 3.2172,
+      "step": 6269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005154691764444337,
+      "loss": 3.0346,
+      "step": 6270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005153386760117772,
+      "loss": 2.9725,
+      "step": 6271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005152081745332352,
+      "loss": 3.0436,
+      "step": 6272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000515077672017706,
+      "loss": 2.8542,
+      "step": 6273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005149471684740881,
+      "loss": 2.9661,
+      "step": 6274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005148166639112799,
+      "loss": 2.6923,
+      "step": 6275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005146861583381802,
+      "loss": 2.8621,
+      "step": 6276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005145556517636878,
+      "loss": 2.9905,
+      "step": 6277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005144251441967011,
+      "loss": 3.0823,
+      "step": 6278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005142946356461192,
+      "loss": 3.1161,
+      "step": 6279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005141641261208409,
+      "loss": 3.0825,
+      "step": 6280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005140336156297649,
+      "loss": 3.1711,
+      "step": 6281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005139031041817906,
+      "loss": 2.9852,
+      "step": 6282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005137725917858168,
+      "loss": 3.0172,
+      "step": 6283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005136420784507428,
+      "loss": 2.8331,
+      "step": 6284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005135115641854676,
+      "loss": 3.0306,
+      "step": 6285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005133810489988907,
+      "loss": 3.3184,
+      "step": 6286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005132505328999114,
+      "loss": 2.8877,
+      "step": 6287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005131200158974289,
+      "loss": 2.9505,
+      "step": 6288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005129894980003429,
+      "loss": 2.9888,
+      "step": 6289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005128589792175528,
+      "loss": 2.785,
+      "step": 6290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005127284595579581,
+      "loss": 2.8228,
+      "step": 6291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005125979390304585,
+      "loss": 3.0511,
+      "step": 6292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005124674176439538,
+      "loss": 2.8053,
+      "step": 6293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005123368954073435,
+      "loss": 3.0673,
+      "step": 6294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005122063723295277,
+      "loss": 2.8891,
+      "step": 6295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005120758484194061,
+      "loss": 2.5906,
+      "step": 6296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005119453236858786,
+      "loss": 2.9773,
+      "step": 6297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005118147981378453,
+      "loss": 2.7587,
+      "step": 6298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000511684271784206,
+      "loss": 2.9764,
+      "step": 6299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005115537446338613,
+      "loss": 3.1225,
+      "step": 6300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005114232166957107,
+      "loss": 2.973,
+      "step": 6301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000511292687978655,
+      "loss": 2.9392,
+      "step": 6302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005111621584915941,
+      "loss": 3.0184,
+      "step": 6303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005110316282434283,
+      "loss": 3.1568,
+      "step": 6304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005109010972430581,
+      "loss": 2.9546,
+      "step": 6305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005107705654993838,
+      "loss": 2.9406,
+      "step": 6306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005106400330213058,
+      "loss": 2.9157,
+      "step": 6307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005105094998177249,
+      "loss": 2.7653,
+      "step": 6308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005103789658975413,
+      "loss": 3.0914,
+      "step": 6309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005102484312696559,
+      "loss": 2.7304,
+      "step": 6310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005101178959429692,
+      "loss": 3.0132,
+      "step": 6311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000509987359926382,
+      "loss": 2.8308,
+      "step": 6312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005098568232287949,
+      "loss": 3.0276,
+      "step": 6313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005097262858591087,
+      "loss": 2.8667,
+      "step": 6314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005095957478262244,
+      "loss": 2.9649,
+      "step": 6315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000509465209139043,
+      "loss": 2.8835,
+      "step": 6316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000509334669806465,
+      "loss": 3.1201,
+      "step": 6317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005092041298373918,
+      "loss": 2.9646,
+      "step": 6318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005090735892407242,
+      "loss": 2.8492,
+      "step": 6319
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005089430480253634,
+      "loss": 3.0744,
+      "step": 6320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005088125062002103,
+      "loss": 2.9781,
+      "step": 6321
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005086819637741661,
+      "loss": 2.9159,
+      "step": 6322
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005085514207561323,
+      "loss": 3.1378,
+      "step": 6323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005084208771550097,
+      "loss": 3.0481,
+      "step": 6324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005082903329796997,
+      "loss": 2.8666,
+      "step": 6325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005081597882391038,
+      "loss": 2.9305,
+      "step": 6326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005080292429421231,
+      "loss": 2.9942,
+      "step": 6327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005078986970976591,
+      "loss": 2.7875,
+      "step": 6328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005077681507146132,
+      "loss": 3.0241,
+      "step": 6329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000507637603801887,
+      "loss": 3.106,
+      "step": 6330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005075070563683817,
+      "loss": 3.2597,
+      "step": 6331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005073765084229989,
+      "loss": 2.9893,
+      "step": 6332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005072459599746405,
+      "loss": 3.2132,
+      "step": 6333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005071154110322077,
+      "loss": 2.9129,
+      "step": 6334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005069848616046024,
+      "loss": 2.9678,
+      "step": 6335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005068543117007259,
+      "loss": 2.9925,
+      "step": 6336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005067237613294804,
+      "loss": 2.9305,
+      "step": 6337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005065932104997672,
+      "loss": 3.0679,
+      "step": 6338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005064626592204884,
+      "loss": 3.0873,
+      "step": 6339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005063321075005454,
+      "loss": 2.8079,
+      "step": 6340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005062015553488404,
+      "loss": 2.9983,
+      "step": 6341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005060710027742752,
+      "loss": 3.1613,
+      "step": 6342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005059404497857515,
+      "loss": 3.1199,
+      "step": 6343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005058098963921712,
+      "loss": 3.1235,
+      "step": 6344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005056793426024364,
+      "loss": 2.6191,
+      "step": 6345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005055487884254489,
+      "loss": 2.9919,
+      "step": 6346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000505418233870111,
+      "loss": 2.9563,
+      "step": 6347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005052876789453242,
+      "loss": 2.9848,
+      "step": 6348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005051571236599911,
+      "loss": 2.9394,
+      "step": 6349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005050265680230134,
+      "loss": 2.9876,
+      "step": 6350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005048960120432931,
+      "loss": 3.0106,
+      "step": 6351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005047654557297326,
+      "loss": 3.1343,
+      "step": 6352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005046348990912341,
+      "loss": 3.0642,
+      "step": 6353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005045043421366994,
+      "loss": 2.889,
+      "step": 6354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000504373784875031,
+      "loss": 3.0387,
+      "step": 6355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005042432273151309,
+      "loss": 2.8712,
+      "step": 6356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005041126694659015,
+      "loss": 2.759,
+      "step": 6357
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005039821113362447,
+      "loss": 2.9692,
+      "step": 6358
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005038515529350631,
+      "loss": 2.9455,
+      "step": 6359
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005037209942712588,
+      "loss": 3.3654,
+      "step": 6360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005035904353537342,
+      "loss": 3.3067,
+      "step": 6361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005034598761913916,
+      "loss": 2.9911,
+      "step": 6362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005033293167931335,
+      "loss": 2.9796,
+      "step": 6363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005031987571678617,
+      "loss": 2.9079,
+      "step": 6364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005030681973244791,
+      "loss": 2.8183,
+      "step": 6365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005029376372718877,
+      "loss": 3.0893,
+      "step": 6366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005028070770189902,
+      "loss": 3.0308,
+      "step": 6367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005026765165746888,
+      "loss": 2.9329,
+      "step": 6368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005025459559478861,
+      "loss": 2.8566,
+      "step": 6369
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005024153951474844,
+      "loss": 3.0291,
+      "step": 6370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000502284834182386,
+      "loss": 2.7963,
+      "step": 6371
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005021542730614937,
+      "loss": 2.8537,
+      "step": 6372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005020237117937097,
+      "loss": 2.915,
+      "step": 6373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005018931503879365,
+      "loss": 3.097,
+      "step": 6374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005017625888530765,
+      "loss": 2.9959,
+      "step": 6375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005016320271980326,
+      "loss": 2.9,
+      "step": 6376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000501501465431707,
+      "loss": 3.1725,
+      "step": 6377
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005013709035630021,
+      "loss": 3.0554,
+      "step": 6378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005012403416008206,
+      "loss": 2.8135,
+      "step": 6379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005011097795540649,
+      "loss": 2.8977,
+      "step": 6380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005009792174316376,
+      "loss": 2.8991,
+      "step": 6381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005008486552424412,
+      "loss": 3.1318,
+      "step": 6382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005007180929953783,
+      "loss": 2.8775,
+      "step": 6383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005005875306993514,
+      "loss": 2.9737,
+      "step": 6384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000500456968363263,
+      "loss": 2.7183,
+      "step": 6385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005003264059960156,
+      "loss": 3.08,
+      "step": 6386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.000500195843606512,
+      "loss": 2.9772,
+      "step": 6387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0005000652812036545,
+      "loss": 2.9469,
+      "step": 6388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004999347187963456,
+      "loss": 3.078,
+      "step": 6389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004998041563934881,
+      "loss": 2.9286,
+      "step": 6390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004996735940039844,
+      "loss": 2.8927,
+      "step": 6391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000499543031636737,
+      "loss": 3.1416,
+      "step": 6392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004994124693006488,
+      "loss": 2.9231,
+      "step": 6393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004992819070046218,
+      "loss": 2.9952,
+      "step": 6394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000499151344757559,
+      "loss": 3.0946,
+      "step": 6395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004990207825683625,
+      "loss": 3.116,
+      "step": 6396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004988902204459351,
+      "loss": 2.9089,
+      "step": 6397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004987596583991795,
+      "loss": 2.9924,
+      "step": 6398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000498629096436998,
+      "loss": 2.7397,
+      "step": 6399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004984985345682931,
+      "loss": 2.8668,
+      "step": 6400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004983679728019674,
+      "loss": 2.9421,
+      "step": 6401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004982374111469234,
+      "loss": 2.9197,
+      "step": 6402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004981068496120636,
+      "loss": 3.1597,
+      "step": 6403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004979762882062903,
+      "loss": 2.9388,
+      "step": 6404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004978457269385064,
+      "loss": 2.9556,
+      "step": 6405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004977151658176141,
+      "loss": 2.8263,
+      "step": 6406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004975846048525158,
+      "loss": 2.9506,
+      "step": 6407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004974540440521141,
+      "loss": 2.9356,
+      "step": 6408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004973234834253112,
+      "loss": 2.9628,
+      "step": 6409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004971929229810099,
+      "loss": 3.3681,
+      "step": 6410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004970623627281125,
+      "loss": 3.0165,
+      "step": 6411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004969318026755211,
+      "loss": 2.7433,
+      "step": 6412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004968012428321384,
+      "loss": 3.0552,
+      "step": 6413
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004966706832068668,
+      "loss": 3.1091,
+      "step": 6414
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004965401238086084,
+      "loss": 2.8605,
+      "step": 6415
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004964095646462659,
+      "loss": 2.8058,
+      "step": 6416
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004962790057287412,
+      "loss": 3.0563,
+      "step": 6417
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000496148447064937,
+      "loss": 2.9243,
+      "step": 6418
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004960178886637554,
+      "loss": 2.9864,
+      "step": 6419
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004958873305340987,
+      "loss": 2.8969,
+      "step": 6420
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004957567726848691,
+      "loss": 2.7806,
+      "step": 6421
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004956262151249691,
+      "loss": 3.0421,
+      "step": 6422
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004954956578633005,
+      "loss": 3.0429,
+      "step": 6423
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000495365100908766,
+      "loss": 2.8796,
+      "step": 6424
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004952345442702673,
+      "loss": 2.8701,
+      "step": 6425
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004951039879567069,
+      "loss": 2.9284,
+      "step": 6426
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004949734319769867,
+      "loss": 2.8023,
+      "step": 6427
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000494842876340009,
+      "loss": 2.7991,
+      "step": 6428
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004947123210546757,
+      "loss": 2.8868,
+      "step": 6429
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004945817661298891,
+      "loss": 3.0178,
+      "step": 6430
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000494451211574551,
+      "loss": 3.0133,
+      "step": 6431
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004943206573975637,
+      "loss": 3.0527,
+      "step": 6432
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004941901036078288,
+      "loss": 3.0005,
+      "step": 6433
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004940595502142485,
+      "loss": 2.9373,
+      "step": 6434
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004939289972257248,
+      "loss": 2.7974,
+      "step": 6435
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004937984446511596,
+      "loss": 3.0997,
+      "step": 6436
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004936678924994545,
+      "loss": 3.2186,
+      "step": 6437
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004935373407795118,
+      "loss": 2.9078,
+      "step": 6438
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004934067895002329,
+      "loss": 2.9375,
+      "step": 6439
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004932762386705197,
+      "loss": 3.2149,
+      "step": 6440
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004931456882992742,
+      "loss": 3.2171,
+      "step": 6441
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004930151383953976,
+      "loss": 3.148,
+      "step": 6442
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004928845889677925,
+      "loss": 3.0859,
+      "step": 6443
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004927540400253595,
+      "loss": 3.0969,
+      "step": 6444
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004926234915770012,
+      "loss": 3.0978,
+      "step": 6445
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004924929436316183,
+      "loss": 2.9192,
+      "step": 6446
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004923623961981133,
+      "loss": 2.904,
+      "step": 6447
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004922318492853869,
+      "loss": 2.9217,
+      "step": 6448
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000492101302902341,
+      "loss": 3.0988,
+      "step": 6449
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004919707570578769,
+      "loss": 2.936,
+      "step": 6450
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004918402117608964,
+      "loss": 3.0392,
+      "step": 6451
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004917096670203003,
+      "loss": 2.938,
+      "step": 6452
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004915791228449905,
+      "loss": 2.8139,
+      "step": 6453
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004914485792438678,
+      "loss": 2.9569,
+      "step": 6454
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000491318036225834,
+      "loss": 3.0845,
+      "step": 6455
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004911874937997898,
+      "loss": 3.0652,
+      "step": 6456
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004910569519746367,
+      "loss": 2.9579,
+      "step": 6457
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004909264107592757,
+      "loss": 2.9222,
+      "step": 6458
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004907958701626082,
+      "loss": 2.7642,
+      "step": 6459
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004906653301935349,
+      "loss": 2.9564,
+      "step": 6460
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004905347908609572,
+      "loss": 3.2458,
+      "step": 6461
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004904042521737755,
+      "loss": 2.9616,
+      "step": 6462
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004902737141408914,
+      "loss": 2.8851,
+      "step": 6463
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004901431767712052,
+      "loss": 2.9093,
+      "step": 6464
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004900126400736181,
+      "loss": 3.0123,
+      "step": 6465
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004898821040570308,
+      "loss": 3.1247,
+      "step": 6466
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004897515687303443,
+      "loss": 2.8733,
+      "step": 6467
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004896210341024587,
+      "loss": 2.9544,
+      "step": 6468
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004894905001822752,
+      "loss": 2.9141,
+      "step": 6469
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004893599669786941,
+      "loss": 2.9383,
+      "step": 6470
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004892294345006165,
+      "loss": 2.9945,
+      "step": 6471
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004890989027569421,
+      "loss": 2.9319,
+      "step": 6472
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004889683717565718,
+      "loss": 2.8627,
+      "step": 6473
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004888378415084061,
+      "loss": 3.0256,
+      "step": 6474
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004887073120213451,
+      "loss": 2.9689,
+      "step": 6475
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004885767833042894,
+      "loss": 3.1378,
+      "step": 6476
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004884462553661388,
+      "loss": 3.0155,
+      "step": 6477
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000488315728215794,
+      "loss": 3.1741,
+      "step": 6478
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004881852018621548,
+      "loss": 3.0715,
+      "step": 6479
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048805467631412154,
+      "loss": 2.7235,
+      "step": 6480
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000487924151580594,
+      "loss": 3.1656,
+      "step": 6481
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004877936276704725,
+      "loss": 2.8986,
+      "step": 6482
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004876631045926565,
+      "loss": 3.0027,
+      "step": 6483
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004875325823560464,
+      "loss": 3.0946,
+      "step": 6484
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048740206096954154,
+      "loss": 2.7899,
+      "step": 6485
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000487271540442042,
+      "loss": 2.765,
+      "step": 6486
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048714102078244734,
+      "loss": 3.3038,
+      "step": 6487
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004870105019996572,
+      "loss": 2.7941,
+      "step": 6488
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000486879984102571,
+      "loss": 3.0222,
+      "step": 6489
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004867494671000887,
+      "loss": 2.9337,
+      "step": 6490
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004866189510011092,
+      "loss": 2.8544,
+      "step": 6491
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048648843581453245,
+      "loss": 2.9976,
+      "step": 6492
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048635792154925724,
+      "loss": 2.7895,
+      "step": 6493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004862274082141832,
+      "loss": 3.1708,
+      "step": 6494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004860968958182095,
+      "loss": 2.9564,
+      "step": 6495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004859663843702351,
+      "loss": 2.8575,
+      "step": 6496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048583587387915914,
+      "loss": 2.867,
+      "step": 6497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004857053643538809,
+      "loss": 3.0454,
+      "step": 6498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048557485580329884,
+      "loss": 3.0122,
+      "step": 6499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004854443482363123,
+      "loss": 3.0191,
+      "step": 6500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048531384166181974,
+      "loss": 2.7858,
+      "step": 6501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004851833360887201,
+      "loss": 2.9491,
+      "step": 6502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048505283152591207,
+      "loss": 3.1065,
+      "step": 6503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048492232798229415,
+      "loss": 2.926,
+      "step": 6504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.000484791825466765,
+      "loss": 2.9057,
+      "step": 6505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004846613239882228,
+      "loss": 2.8903,
+      "step": 6506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048453082355556646,
+      "loss": 2.9911,
+      "step": 6507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004844003241776937,
+      "loss": 2.9404,
+      "step": 6508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004842698258635032,
+      "loss": 2.8485,
+      "step": 6509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048413932862189285,
+      "loss": 2.8807,
+      "step": 6510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0004840088324617612,
+      "loss": 2.9334,
+      "step": 6511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.00048387833739200575,
+      "loss": 3.0808,
+      "step": 6512
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004837478434215249,
+      "loss": 3.0584,
+      "step": 6513
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004836173505592162,
+      "loss": 3.0337,
+      "step": 6514
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048348685881397776,
+      "loss": 3.1378,
+      "step": 6515
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000483356368194707,
+      "loss": 2.8835,
+      "step": 6516
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048322587871030173,
+      "loss": 2.7802,
+      "step": 6517
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048309539036965946,
+      "loss": 2.6127,
+      "step": 6518
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048296490318167794,
+      "loss": 3.0173,
+      "step": 6519
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048283441715525426,
+      "loss": 3.2071,
+      "step": 6520
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048270393229928613,
+      "loss": 2.9583,
+      "step": 6521
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048257344862267035,
+      "loss": 2.9965,
+      "step": 6522
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048244296613430453,
+      "loss": 2.7741,
+      "step": 6523
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004823124848430854,
+      "loss": 3.1138,
+      "step": 6524
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004821820047579103,
+      "loss": 2.9235,
+      "step": 6525
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048205152588767585,
+      "loss": 2.969,
+      "step": 6526
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048192104824127936,
+      "loss": 2.8706,
+      "step": 6527
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004817905718276171,
+      "loss": 3.0786,
+      "step": 6528
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048166009665558617,
+      "loss": 3.2233,
+      "step": 6529
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004815296227340828,
+      "loss": 3.0885,
+      "step": 6530
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004813991500720039,
+      "loss": 2.9636,
+      "step": 6531
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048126867867824547,
+      "loss": 2.9069,
+      "step": 6532
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004811382085617042,
+      "loss": 2.7706,
+      "step": 6533
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004810077397312761,
+      "loss": 2.9131,
+      "step": 6534
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004808772721958577,
+      "loss": 3.1297,
+      "step": 6535
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048074680596434463,
+      "loss": 2.8201,
+      "step": 6536
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004806163410456333,
+      "loss": 2.9044,
+      "step": 6537
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004804858774486193,
+      "loss": 3.1913,
+      "step": 6538
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048035541518219853,
+      "loss": 3.0888,
+      "step": 6539
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004802249542552668,
+      "loss": 3.0014,
+      "step": 6540
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00048009449467671954,
+      "loss": 3.1217,
+      "step": 6541
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004799640364554527,
+      "loss": 3.1667,
+      "step": 6542
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004798335796003611,
+      "loss": 2.9399,
+      "step": 6543
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004797031241203408,
+      "loss": 2.8304,
+      "step": 6544
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004795726700242864,
+      "loss": 3.0718,
+      "step": 6545
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004794422173210935,
+      "loss": 3.133,
+      "step": 6546
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047931176601965705,
+      "loss": 2.8812,
+      "step": 6547
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000479181316128872,
+      "loss": 3.1108,
+      "step": 6548
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047905086765763315,
+      "loss": 2.8791,
+      "step": 6549
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004789204206148356,
+      "loss": 2.906,
+      "step": 6550
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047878997500937356,
+      "loss": 3.0196,
+      "step": 6551
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004786595308501421,
+      "loss": 2.843,
+      "step": 6552
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047852908814603523,
+      "loss": 2.8994,
+      "step": 6553
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047839864690594775,
+      "loss": 2.9566,
+      "step": 6554
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004782682071387737,
+      "loss": 2.6922,
+      "step": 6555
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047813776885340743,
+      "loss": 2.8889,
+      "step": 6556
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004780073320587428,
+      "loss": 3.0473,
+      "step": 6557
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047787689676367413,
+      "loss": 3.1949,
+      "step": 6558
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004777464629770949,
+      "loss": 3.0689,
+      "step": 6559
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004776160307078993,
+      "loss": 3.0659,
+      "step": 6560
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004774855999649806,
+      "loss": 3.2142,
+      "step": 6561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004773551707572328,
+      "loss": 3.0096,
+      "step": 6562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047722474309354904,
+      "loss": 3.0663,
+      "step": 6563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004770943169828228,
+      "loss": 3.0899,
+      "step": 6564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004769638924339473,
+      "loss": 3.0707,
+      "step": 6565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004768334694558159,
+      "loss": 2.9837,
+      "step": 6566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004767030480573211,
+      "loss": 2.9333,
+      "step": 6567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047657262824735657,
+      "loss": 3.0034,
+      "step": 6568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004764422100348146,
+      "loss": 3.158,
+      "step": 6569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000476311793428588,
+      "loss": 2.9878,
+      "step": 6570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004761813784375697,
+      "loss": 2.9485,
+      "step": 6571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047605096507065165,
+      "loss": 3.1123,
+      "step": 6572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004759205533367268,
+      "loss": 2.9853,
+      "step": 6573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047579014324468697,
+      "loss": 3.21,
+      "step": 6574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047565973480342467,
+      "loss": 3.051,
+      "step": 6575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047552932802183163,
+      "loss": 3.0913,
+      "step": 6576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047539892290880003,
+      "loss": 3.0347,
+      "step": 6577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004752685194732216,
+      "loss": 2.9273,
+      "step": 6578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004751381177239882,
+      "loss": 3.0118,
+      "step": 6579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047500771766999114,
+      "loss": 3.0838,
+      "step": 6580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004748773193201223,
+      "loss": 2.907,
+      "step": 6581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004747469226832725,
+      "loss": 2.9517,
+      "step": 6582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047461652776833356,
+      "loss": 3.0031,
+      "step": 6583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047448613458419615,
+      "loss": 3.3344,
+      "step": 6584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047435574313975154,
+      "loss": 3.0195,
+      "step": 6585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004742253534438905,
+      "loss": 3.3328,
+      "step": 6586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047409496550550404,
+      "loss": 3.0232,
+      "step": 6587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004739645793334824,
+      "loss": 2.9224,
+      "step": 6588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047383419493671663,
+      "loss": 3.0394,
+      "step": 6589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004737038123240966,
+      "loss": 2.8717,
+      "step": 6590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004735734315045131,
+      "loss": 2.9087,
+      "step": 6591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047344305248685583,
+      "loss": 3.1541,
+      "step": 6592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047331267528001516,
+      "loss": 3.088,
+      "step": 6593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004731822998928808,
+      "loss": 2.9476,
+      "step": 6594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004730519263343429,
+      "loss": 2.8268,
+      "step": 6595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047292155461329074,
+      "loss": 2.9782,
+      "step": 6596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047279118473861413,
+      "loss": 3.1421,
+      "step": 6597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047266081671920227,
+      "loss": 2.9495,
+      "step": 6598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004725304505639448,
+      "loss": 2.8887,
+      "step": 6599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004724000862817304,
+      "loss": 2.9309,
+      "step": 6600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004722697238814485,
+      "loss": 2.8406,
+      "step": 6601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047213936337198794,
+      "loss": 3.0213,
+      "step": 6602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004720090047622373,
+      "loss": 3.1095,
+      "step": 6603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004718786480610856,
+      "loss": 3.0997,
+      "step": 6604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047174829327742093,
+      "loss": 2.8793,
+      "step": 6605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047161794042013203,
+      "loss": 3.0099,
+      "step": 6606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000471487589498107,
+      "loss": 2.9482,
+      "step": 6607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000471357240520234,
+      "loss": 3.0433,
+      "step": 6608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004712268934954009,
+      "loss": 2.9437,
+      "step": 6609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000471096548432496,
+      "loss": 2.8652,
+      "step": 6610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004709662053404065,
+      "loss": 2.7556,
+      "step": 6611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004708358642280204,
+      "loss": 3.0939,
+      "step": 6612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004707055251042247,
+      "loss": 3.0988,
+      "step": 6613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004705751879779072,
+      "loss": 3.1623,
+      "step": 6614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004704448528579549,
+      "loss": 2.8108,
+      "step": 6615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047031451975325486,
+      "loss": 2.7775,
+      "step": 6616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00047018418867269385,
+      "loss": 3.0499,
+      "step": 6617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000470053859625159,
+      "loss": 2.8284,
+      "step": 6618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046992353261953655,
+      "loss": 3.1425,
+      "step": 6619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004697932076647134,
+      "loss": 3.0181,
+      "step": 6620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046966288476957545,
+      "loss": 2.7434,
+      "step": 6621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046953256394300925,
+      "loss": 3.0133,
+      "step": 6622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046940224519390083,
+      "loss": 2.9246,
+      "step": 6623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004692719285311361,
+      "loss": 3.0823,
+      "step": 6624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004691416139636007,
+      "loss": 3.1637,
+      "step": 6625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046901130150018064,
+      "loss": 3.1477,
+      "step": 6626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.000468880991149761,
+      "loss": 2.8766,
+      "step": 6627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046875068292122747,
+      "loss": 2.7904,
+      "step": 6628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046862037682346506,
+      "loss": 2.8661,
+      "step": 6629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046849007286535894,
+      "loss": 2.8423,
+      "step": 6630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004683597710557941,
+      "loss": 2.7701,
+      "step": 6631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004682294714036552,
+      "loss": 2.7874,
+      "step": 6632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004680991739178269,
+      "loss": 2.8298,
+      "step": 6633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.0004679688786071939,
+      "loss": 2.9971,
+      "step": 6634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046783858548064036,
+      "loss": 3.1186,
+      "step": 6635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 0.00046770829454705016,
+      "loss": 2.9108,
+      "step": 6636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046757800581530783,
+      "loss": 3.1227,
+      "step": 6637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046744771929429693,
+      "loss": 2.9663,
+      "step": 6638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004673174349929016,
+      "loss": 2.8932,
+      "step": 6639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004671871529200049,
+      "loss": 2.8278,
+      "step": 6640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046705687308449075,
+      "loss": 2.9993,
+      "step": 6641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000466926595495242,
+      "loss": 3.0156,
+      "step": 6642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004667963201611421,
+      "loss": 3.0172,
+      "step": 6643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046666604709107385,
+      "loss": 3.0721,
+      "step": 6644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004665357762939202,
+      "loss": 3.1383,
+      "step": 6645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004664055077785637,
+      "loss": 3.0032,
+      "step": 6646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004662752415538871,
+      "loss": 3.1749,
+      "step": 6647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046614497762877236,
+      "loss": 3.0586,
+      "step": 6648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004660147160121021,
+      "loss": 3.1639,
+      "step": 6649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004658844567127579,
+      "loss": 3.0656,
+      "step": 6650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004657541997396221,
+      "loss": 2.931,
+      "step": 6651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046562394510157607,
+      "loss": 2.882,
+      "step": 6652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046549369280750167,
+      "loss": 2.9507,
+      "step": 6653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004653634428662799,
+      "loss": 3.0379,
+      "step": 6654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004652331952867926,
+      "loss": 3.1274,
+      "step": 6655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046510295007792026,
+      "loss": 3.0825,
+      "step": 6656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046497270724854425,
+      "loss": 3.1258,
+      "step": 6657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046484246680754496,
+      "loss": 2.8883,
+      "step": 6658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046471222876380336,
+      "loss": 3.0565,
+      "step": 6659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046458199312619943,
+      "loss": 2.8781,
+      "step": 6660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004644517599036139,
+      "loss": 3.1273,
+      "step": 6661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004643215291049265,
+      "loss": 2.8513,
+      "step": 6662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046419130073901754,
+      "loss": 3.0197,
+      "step": 6663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004640610748147664,
+      "loss": 2.968,
+      "step": 6664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004639308513410531,
+      "loss": 3.2234,
+      "step": 6665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004638006303267566,
+      "loss": 2.8458,
+      "step": 6666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004636704117807567,
+      "loss": 2.9943,
+      "step": 6667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046354019571193214,
+      "loss": 2.7391,
+      "step": 6668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004634099821291618,
+      "loss": 2.8563,
+      "step": 6669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004632797710413248,
+      "loss": 3.031,
+      "step": 6670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004631495624572993,
+      "loss": 2.776,
+      "step": 6671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046301935638596426,
+      "loss": 3.2012,
+      "step": 6672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004628891528361974,
+      "loss": 2.6728,
+      "step": 6673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046275895181687707,
+      "loss": 2.7177,
+      "step": 6674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046262875333688113,
+      "loss": 2.9054,
+      "step": 6675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004624985574050874,
+      "loss": 3.0464,
+      "step": 6676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004623683640303732,
+      "loss": 3.0211,
+      "step": 6677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004622381732216163,
+      "loss": 2.8392,
+      "step": 6678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004621079849876935,
+      "loss": 3.1509,
+      "step": 6679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046197779933748224,
+      "loss": 2.9854,
+      "step": 6680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000461847616279859,
+      "loss": 2.8893,
+      "step": 6681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004617174358237007,
+      "loss": 3.044,
+      "step": 6682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046158725797788375,
+      "loss": 3.2124,
+      "step": 6683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004614570827512846,
+      "loss": 3.0091,
+      "step": 6684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004613269101527791,
+      "loss": 2.8367,
+      "step": 6685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004611967401912437,
+      "loss": 3.1273,
+      "step": 6686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046106657287555374,
+      "loss": 3.0612,
+      "step": 6687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004609364082145852,
+      "loss": 3.0221,
+      "step": 6688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046080624621721314,
+      "loss": 2.9385,
+      "step": 6689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046067608689231304,
+      "loss": 2.7769,
+      "step": 6690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046054593024875994,
+      "loss": 2.7811,
+      "step": 6691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046041577629542875,
+      "loss": 3.0992,
+      "step": 6692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046028562504119393,
+      "loss": 2.9973,
+      "step": 6693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046015547649493044,
+      "loss": 2.9476,
+      "step": 6694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00046002533066551214,
+      "loss": 2.9905,
+      "step": 6695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004598951875618136,
+      "loss": 2.9777,
+      "step": 6696
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004597650471927084,
+      "loss": 2.9244,
+      "step": 6697
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045963490956707053,
+      "loss": 2.7926,
+      "step": 6698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045950477469377344,
+      "loss": 2.8743,
+      "step": 6699
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004593746425816909,
+      "loss": 2.8976,
+      "step": 6700
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045924451323969573,
+      "loss": 2.9397,
+      "step": 6701
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004591143866766609,
+      "loss": 3.2711,
+      "step": 6702
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004589842629014596,
+      "loss": 2.723,
+      "step": 6703
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045885414192296407,
+      "loss": 2.8776,
+      "step": 6704
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045872402375004705,
+      "loss": 3.084,
+      "step": 6705
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004585939083915807,
+      "loss": 2.9841,
+      "step": 6706
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004584637958564372,
+      "loss": 3.2779,
+      "step": 6707
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045833368615348814,
+      "loss": 2.7967,
+      "step": 6708
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045820357929160564,
+      "loss": 3.2153,
+      "step": 6709
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004580734752796607,
+      "loss": 2.9833,
+      "step": 6710
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004579433741265251,
+      "loss": 2.9375,
+      "step": 6711
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004578132758410694,
+      "loss": 3.1134,
+      "step": 6712
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000457683180432165,
+      "loss": 2.9657,
+      "step": 6713
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004575530879086822,
+      "loss": 2.8026,
+      "step": 6714
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045742299827949203,
+      "loss": 2.8617,
+      "step": 6715
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004572929115534642,
+      "loss": 2.7862,
+      "step": 6716
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004571628277394693,
+      "loss": 2.7798,
+      "step": 6717
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004570327468463769,
+      "loss": 2.9197,
+      "step": 6718
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004569026688830571,
+      "loss": 2.9859,
+      "step": 6719
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045677259385837896,
+      "loss": 2.738,
+      "step": 6720
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045664252178121215,
+      "loss": 3.1052,
+      "step": 6721
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045651245266042554,
+      "loss": 2.9444,
+      "step": 6722
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004563823865048884,
+      "loss": 2.9591,
+      "step": 6723
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000456252323323469,
+      "loss": 2.8903,
+      "step": 6724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004561222631250363,
+      "loss": 2.9096,
+      "step": 6725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045599220591845813,
+      "loss": 2.8105,
+      "step": 6726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045586215171260306,
+      "loss": 2.7209,
+      "step": 6727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045573210051633856,
+      "loss": 3.0443,
+      "step": 6728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045560205233853264,
+      "loss": 2.8181,
+      "step": 6729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004554720071880525,
+      "loss": 2.7889,
+      "step": 6730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004553419650737659,
+      "loss": 3.0823,
+      "step": 6731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004552119260045394,
+      "loss": 2.9287,
+      "step": 6732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004550818899892402,
+      "loss": 3.0489,
+      "step": 6733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045495185703673475,
+      "loss": 2.8285,
+      "step": 6734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004548218271558896,
+      "loss": 2.892,
+      "step": 6735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000454691800355571,
+      "loss": 3.1009,
+      "step": 6736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045456177664464484,
+      "loss": 2.8419,
+      "step": 6737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004544317560319772,
+      "loss": 2.8461,
+      "step": 6738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045430173852643337,
+      "loss": 2.8986,
+      "step": 6739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004541717241368791,
+      "loss": 2.9401,
+      "step": 6740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004540417128721792,
+      "loss": 3.0353,
+      "step": 6741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004539117047411989,
+      "loss": 3.0189,
+      "step": 6742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045378169975280285,
+      "loss": 3.1296,
+      "step": 6743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004536516979158557,
+      "loss": 3.0877,
+      "step": 6744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045352169923922156,
+      "loss": 3.1695,
+      "step": 6745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004533917037317649,
+      "loss": 3.1015,
+      "step": 6746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004532617114023492,
+      "loss": 3.0999,
+      "step": 6747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045313172225983853,
+      "loss": 2.7918,
+      "step": 6748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004530017363130959,
+      "loss": 3.212,
+      "step": 6749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.000452871753570985,
+      "loss": 3.132,
+      "step": 6750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004527417740423686,
+      "loss": 2.8311,
+      "step": 6751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004526117977361096,
+      "loss": 3.1209,
+      "step": 6752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045248182466107046,
+      "loss": 3.1411,
+      "step": 6753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045235185482611384,
+      "loss": 2.957,
+      "step": 6754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045222188824010147,
+      "loss": 3.0171,
+      "step": 6755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004520919249118958,
+      "loss": 3.0661,
+      "step": 6756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004519619648503579,
+      "loss": 3.0518,
+      "step": 6757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004518320080643497,
+      "loss": 2.9704,
+      "step": 6758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.0004517020545627322,
+      "loss": 3.0517,
+      "step": 6759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 0.00045157210435436686,
+      "loss": 2.8972,
+      "step": 6760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004514421574481139,
+      "loss": 2.7847,
+      "step": 6761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045131221385283447,
+      "loss": 2.8058,
+      "step": 6762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004511822735773884,
+      "loss": 3.0571,
+      "step": 6763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045105233663063625,
+      "loss": 3.2285,
+      "step": 6764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045092240302143754,
+      "loss": 2.8755,
+      "step": 6765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045079247275865226,
+      "loss": 3.2418,
+      "step": 6766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045066254585113984,
+      "loss": 2.9649,
+      "step": 6767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045053262230775916,
+      "loss": 2.985,
+      "step": 6768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004504027021373697,
+      "loss": 2.9676,
+      "step": 6769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045027278534882974,
+      "loss": 2.8997,
+      "step": 6770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004501428719509982,
+      "loss": 2.9958,
+      "step": 6771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00045001296195273303,
+      "loss": 3.1127,
+      "step": 6772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044988305536289264,
+      "loss": 3.0272,
+      "step": 6773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004497531521903345,
+      "loss": 2.8291,
+      "step": 6774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044962325244391664,
+      "loss": 2.9697,
+      "step": 6775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000449493356132496,
+      "loss": 2.9315,
+      "step": 6776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044936346326493015,
+      "loss": 2.9247,
+      "step": 6777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044923357385007553,
+      "loss": 2.9248,
+      "step": 6778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004491036878967892,
+      "loss": 2.8301,
+      "step": 6779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044897380541392715,
+      "loss": 2.8153,
+      "step": 6780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044884392641034594,
+      "loss": 3.0635,
+      "step": 6781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004487140508949013,
+      "loss": 2.9709,
+      "step": 6782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044858417887644927,
+      "loss": 3.0999,
+      "step": 6783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044845431036384485,
+      "loss": 2.7449,
+      "step": 6784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004483244453659438,
+      "loss": 2.8345,
+      "step": 6785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044819458389160055,
+      "loss": 3.1485,
+      "step": 6786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044806472594967047,
+      "loss": 2.931,
+      "step": 6787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044793487154900743,
+      "loss": 3.0811,
+      "step": 6788
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004478050206984662,
+      "loss": 2.7254,
+      "step": 6789
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004476751734069005,
+      "loss": 2.8969,
+      "step": 6790
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004475453296831645,
+      "loss": 2.8056,
+      "step": 6791
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044741548953611124,
+      "loss": 2.7513,
+      "step": 6792
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044728565297459454,
+      "loss": 2.8586,
+      "step": 6793
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000447155820007467,
+      "loss": 3.1399,
+      "step": 6794
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044702599064358185,
+      "loss": 3.0447,
+      "step": 6795
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004468961648917912,
+      "loss": 2.9809,
+      "step": 6796
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044676634276094774,
+      "loss": 2.9331,
+      "step": 6797
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004466365242599033,
+      "loss": 3.0198,
+      "step": 6798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044650670939751,
+      "loss": 3.0943,
+      "step": 6799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044637689818261933,
+      "loss": 2.8066,
+      "step": 6800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044624709062408233,
+      "loss": 2.8631,
+      "step": 6801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004461172867307503,
+      "loss": 2.9999,
+      "step": 6802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044598748651147413,
+      "loss": 3.0583,
+      "step": 6803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004458576899751043,
+      "loss": 2.7082,
+      "step": 6804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000445727897130491,
+      "loss": 2.8731,
+      "step": 6805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004455981079864848,
+      "loss": 2.782,
+      "step": 6806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004454683225519349,
+      "loss": 3.0148,
+      "step": 6807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004453385408356913,
+      "loss": 2.9293,
+      "step": 6808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044520876284660305,
+      "loss": 3.0207,
+      "step": 6809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044507898859351945,
+      "loss": 2.9309,
+      "step": 6810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044494921808528923,
+      "loss": 2.8778,
+      "step": 6811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004448194513307609,
+      "loss": 3.1456,
+      "step": 6812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044468968833878263,
+      "loss": 2.9548,
+      "step": 6813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044455992911820286,
+      "loss": 3.2346,
+      "step": 6814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004444301736778689,
+      "loss": 2.9583,
+      "step": 6815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004443004220266287,
+      "loss": 3.1697,
+      "step": 6816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044417067417332917,
+      "loss": 3.0577,
+      "step": 6817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044404093012681756,
+      "loss": 2.7874,
+      "step": 6818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044391118989594035,
+      "loss": 3.1994,
+      "step": 6819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004437814534895445,
+      "loss": 2.8365,
+      "step": 6820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044365172091647575,
+      "loss": 2.9602,
+      "step": 6821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004435219921855805,
+      "loss": 2.8338,
+      "step": 6822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004433922673057039,
+      "loss": 2.7717,
+      "step": 6823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044326254628569207,
+      "loss": 3.0704,
+      "step": 6824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004431328291343895,
+      "loss": 3.139,
+      "step": 6825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044300311586064164,
+      "loss": 2.9392,
+      "step": 6826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004428734064732928,
+      "loss": 2.8802,
+      "step": 6827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004427437009811877,
+      "loss": 2.8104,
+      "step": 6828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004426139993931701,
+      "loss": 3.1189,
+      "step": 6829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004424843017180842,
+      "loss": 2.9633,
+      "step": 6830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044235460796477325,
+      "loss": 2.6779,
+      "step": 6831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004422249181420809,
+      "loss": 3.2142,
+      "step": 6832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044209523225884994,
+      "loss": 2.8538,
+      "step": 6833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044196555032392315,
+      "loss": 2.8961,
+      "step": 6834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004418358723461433,
+      "loss": 3.1383,
+      "step": 6835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044170619833435235,
+      "loss": 3.139,
+      "step": 6836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004415765282973926,
+      "loss": 3.1506,
+      "step": 6837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004414468622441053,
+      "loss": 2.8873,
+      "step": 6838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004413172001833323,
+      "loss": 3.0429,
+      "step": 6839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004411875421239145,
+      "loss": 3.096,
+      "step": 6840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044105788807469296,
+      "loss": 2.9768,
+      "step": 6841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044092823804450803,
+      "loss": 2.8756,
+      "step": 6842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044079859204220053,
+      "loss": 3.0891,
+      "step": 6843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044066895007660996,
+      "loss": 2.8712,
+      "step": 6844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004405393121565766,
+      "loss": 2.7591,
+      "step": 6845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044040967829093953,
+      "loss": 2.9124,
+      "step": 6846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004402800484885384,
+      "loss": 3.0574,
+      "step": 6847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044015042275821184,
+      "loss": 3.2919,
+      "step": 6848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00044002080110879866,
+      "loss": 2.8418,
+      "step": 6849
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043989118354913713,
+      "loss": 3.0597,
+      "step": 6850
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043976157008806583,
+      "loss": 3.0556,
+      "step": 6851
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000439631960734422,
+      "loss": 2.9822,
+      "step": 6852
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043950235549704373,
+      "loss": 3.1185,
+      "step": 6853
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043937275438476776,
+      "loss": 2.9713,
+      "step": 6854
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004392431574064317,
+      "loss": 2.9863,
+      "step": 6855
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004391135645708717,
+      "loss": 2.9328,
+      "step": 6856
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043898397588692465,
+      "loss": 2.9927,
+      "step": 6857
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004388543913634263,
+      "loss": 2.961,
+      "step": 6858
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000438724811009213,
+      "loss": 3.3853,
+      "step": 6859
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004385952348331199,
+      "loss": 2.8053,
+      "step": 6860
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004384656628439827,
+      "loss": 2.8402,
+      "step": 6861
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.000438336095050636,
+      "loss": 3.0638,
+      "step": 6862
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004382065314619149,
+      "loss": 2.9185,
+      "step": 6863
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004380769720866536,
+      "loss": 3.1252,
+      "step": 6864
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004379474169336862,
+      "loss": 2.9418,
+      "step": 6865
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043781786601184694,
+      "loss": 2.9049,
+      "step": 6866
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043768831932996887,
+      "loss": 2.6915,
+      "step": 6867
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004375587768968859,
+      "loss": 2.9248,
+      "step": 6868
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043742923872143046,
+      "loss": 2.6885,
+      "step": 6869
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043729970481243556,
+      "loss": 3.1748,
+      "step": 6870
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043717017517873357,
+      "loss": 2.9413,
+      "step": 6871
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043704064982915677,
+      "loss": 3.0713,
+      "step": 6872
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043691112877253664,
+      "loss": 2.9438,
+      "step": 6873
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043678161201770514,
+      "loss": 2.8421,
+      "step": 6874
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043665209957349316,
+      "loss": 3.3224,
+      "step": 6875
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043652259144873196,
+      "loss": 3.155,
+      "step": 6876
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043639308765225187,
+      "loss": 3.0953,
+      "step": 6877
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043626358819288355,
+      "loss": 2.9658,
+      "step": 6878
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004361340930794568,
+      "loss": 3.0944,
+      "step": 6879
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.00043600460232080183,
+      "loss": 2.8489,
+      "step": 6880
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004358751159257477,
+      "loss": 2.9751,
+      "step": 6881
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004357456339031238,
+      "loss": 2.854,
+      "step": 6882
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004356161562617589,
+      "loss": 3.0317,
+      "step": 6883
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 0.0004354866830104818,
+      "loss": 3.0304,
+      "step": 6884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043535721415812037,
+      "loss": 3.0703,
+      "step": 6885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004352277497135029,
+      "loss": 2.97,
+      "step": 6886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000435098289685457,
+      "loss": 3.1286,
+      "step": 6887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004349688340828103,
+      "loss": 3.0009,
+      "step": 6888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004348393829143894,
+      "loss": 3.1343,
+      "step": 6889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043470993618902157,
+      "loss": 2.8508,
+      "step": 6890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004345804939155328,
+      "loss": 2.9447,
+      "step": 6891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004344510561027498,
+      "loss": 2.9938,
+      "step": 6892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043432162275949793,
+      "loss": 2.7837,
+      "step": 6893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004341921938946031,
+      "loss": 2.8803,
+      "step": 6894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004340627695168903,
+      "loss": 3.0745,
+      "step": 6895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043393334963518474,
+      "loss": 2.9553,
+      "step": 6896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004338039342583111,
+      "loss": 3.077,
+      "step": 6897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004336745233950933,
+      "loss": 2.8731,
+      "step": 6898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004335451170543559,
+      "loss": 3.1758,
+      "step": 6899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043341571524492214,
+      "loss": 2.8204,
+      "step": 6900
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043328631797561574,
+      "loss": 2.683,
+      "step": 6901
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043315692525525964,
+      "loss": 2.9547,
+      "step": 6902
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004330275370926769,
+      "loss": 3.0945,
+      "step": 6903
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043289815349668967,
+      "loss": 2.9196,
+      "step": 6904
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004327687744761204,
+      "loss": 2.9316,
+      "step": 6905
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004326394000397907,
+      "loss": 2.9756,
+      "step": 6906
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043251003019652236,
+      "loss": 2.7708,
+      "step": 6907
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004323806649551363,
+      "loss": 2.8216,
+      "step": 6908
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004322513043244538,
+      "loss": 3.0498,
+      "step": 6909
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043212194831329523,
+      "loss": 3.0302,
+      "step": 6910
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043199259693048103,
+      "loss": 3.1206,
+      "step": 6911
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043186325018483096,
+      "loss": 3.0752,
+      "step": 6912
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043173390808516507,
+      "loss": 2.9365,
+      "step": 6913
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043160457064030214,
+      "loss": 2.8554,
+      "step": 6914
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004314752378590618,
+      "loss": 2.7984,
+      "step": 6915
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004313459097502622,
+      "loss": 2.7662,
+      "step": 6916
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004312165863227221,
+      "loss": 2.7215,
+      "step": 6917
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004310872675852593,
+      "loss": 2.9387,
+      "step": 6918
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000430957953546692,
+      "loss": 3.0378,
+      "step": 6919
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043082864421583705,
+      "loss": 3.1942,
+      "step": 6920
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004306993396015121,
+      "loss": 2.7346,
+      "step": 6921
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004305700397125335,
+      "loss": 2.8291,
+      "step": 6922
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043044074455771803,
+      "loss": 3.0743,
+      "step": 6923
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004303114541458815,
+      "loss": 2.9591,
+      "step": 6924
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043018216848584006,
+      "loss": 2.9149,
+      "step": 6925
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00043005288758640893,
+      "loss": 2.7694,
+      "step": 6926
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004299236114564037,
+      "loss": 2.8393,
+      "step": 6927
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042979434010463857,
+      "loss": 3.2583,
+      "step": 6928
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004296650735399288,
+      "loss": 2.8055,
+      "step": 6929
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042953581177108794,
+      "loss": 2.8903,
+      "step": 6930
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042940655480693007,
+      "loss": 3.1677,
+      "step": 6931
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.000429277302656269,
+      "loss": 2.9824,
+      "step": 6932
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004291480553279175,
+      "loss": 2.8898,
+      "step": 6933
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004290188128306889,
+      "loss": 2.9691,
+      "step": 6934
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004288895751733952,
+      "loss": 3.0065,
+      "step": 6935
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042876034236484923,
+      "loss": 3.0495,
+      "step": 6936
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042863111441386235,
+      "loss": 2.8093,
+      "step": 6937
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042850189132924644,
+      "loss": 2.851,
+      "step": 6938
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042837267311981254,
+      "loss": 3.0002,
+      "step": 6939
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004282434597943719,
+      "loss": 2.9185,
+      "step": 6940
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004281142513617345,
+      "loss": 3.0266,
+      "step": 6941
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004279850478307112,
+      "loss": 2.8887,
+      "step": 6942
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042785584921011135,
+      "loss": 3.0772,
+      "step": 6943
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004277266555087449,
+      "loss": 3.1219,
+      "step": 6944
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042759746673542076,
+      "loss": 2.8845,
+      "step": 6945
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042746828289894805,
+      "loss": 3.0334,
+      "step": 6946
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004273391040081352,
+      "loss": 3.1026,
+      "step": 6947
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042720993007179064,
+      "loss": 3.0371,
+      "step": 6948
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042708076109872184,
+      "loss": 2.8389,
+      "step": 6949
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004269515970977368,
+      "loss": 2.9327,
+      "step": 6950
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004268224380776423,
+      "loss": 3.0501,
+      "step": 6951
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004266932840472456,
+      "loss": 3.1246,
+      "step": 6952
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004265641350153528,
+      "loss": 3.0106,
+      "step": 6953
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042643499099077035,
+      "loss": 2.8286,
+      "step": 6954
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042630585198230397,
+      "loss": 2.8304,
+      "step": 6955
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042617671799875947,
+      "loss": 2.7003,
+      "step": 6956
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042604758904894144,
+      "loss": 3.0318,
+      "step": 6957
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042591846514165524,
+      "loss": 2.9182,
+      "step": 6958
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004257893462857048,
+      "loss": 2.9054,
+      "step": 6959
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042566023248989475,
+      "loss": 3.2055,
+      "step": 6960
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004255311237630285,
+      "loss": 3.0246,
+      "step": 6961
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042540202011390963,
+      "loss": 2.9953,
+      "step": 6962
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004252729215513414,
+      "loss": 2.7204,
+      "step": 6963
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004251438280841261,
+      "loss": 3.2387,
+      "step": 6964
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042501473972106655,
+      "loss": 2.8367,
+      "step": 6965
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004248856564709645,
+      "loss": 2.861,
+      "step": 6966
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042475657834262193,
+      "loss": 2.8368,
+      "step": 6967
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004246275053448397,
+      "loss": 2.9078,
+      "step": 6968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042449843748641937,
+      "loss": 3.1442,
+      "step": 6969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042436937477616113,
+      "loss": 3.0597,
+      "step": 6970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004242403172228657,
+      "loss": 3.1784,
+      "step": 6971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042411126483533256,
+      "loss": 2.976,
+      "step": 6972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004239822176223618,
+      "loss": 2.8957,
+      "step": 6973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042385317559275217,
+      "loss": 3.0144,
+      "step": 6974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004237241387553029,
+      "loss": 3.0779,
+      "step": 6975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004235951071188122,
+      "loss": 3.231,
+      "step": 6976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042346608069207856,
+      "loss": 2.994,
+      "step": 6977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004233370594838995,
+      "loss": 2.8144,
+      "step": 6978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004232080435030729,
+      "loss": 3.0332,
+      "step": 6979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042307903275839537,
+      "loss": 3.0315,
+      "step": 6980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042295002725866414,
+      "loss": 3.1868,
+      "step": 6981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004228210270126751,
+      "loss": 2.9307,
+      "step": 6982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042269203202922477,
+      "loss": 2.7308,
+      "step": 6983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004225630423171084,
+      "loss": 2.9302,
+      "step": 6984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004224340578851216,
+      "loss": 3.0055,
+      "step": 6985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042230507874205914,
+      "loss": 3.235,
+      "step": 6986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004221761048967159,
+      "loss": 2.8226,
+      "step": 6987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042204713635788564,
+      "loss": 2.9565,
+      "step": 6988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042191817313436277,
+      "loss": 2.9228,
+      "step": 6989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042178921523494027,
+      "loss": 2.8389,
+      "step": 6990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042166026266841173,
+      "loss": 3.1122,
+      "step": 6991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042153131544356965,
+      "loss": 3.0724,
+      "step": 6992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042140237356920663,
+      "loss": 3.0853,
+      "step": 6993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004212734370541145,
+      "loss": 2.912,
+      "step": 6994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004211445059070852,
+      "loss": 3.1041,
+      "step": 6995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042101558013691003,
+      "loss": 3.1104,
+      "step": 6996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004208866597523796,
+      "loss": 3.1752,
+      "step": 6997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042075774476228484,
+      "loss": 2.956,
+      "step": 6998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042062883517541584,
+      "loss": 2.9503,
+      "step": 6999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004204999310005626,
+      "loss": 3.045,
+      "step": 7000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004203710322465143,
+      "loss": 3.0213,
+      "step": 7001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00042024213892206047,
+      "loss": 2.8996,
+      "step": 7002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004201132510359894,
+      "loss": 2.886,
+      "step": 7003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004199843685970899,
+      "loss": 2.7696,
+      "step": 7004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.0004198554916141495,
+      "loss": 3.1948,
+      "step": 7005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041972662009595623,
+      "loss": 2.6165,
+      "step": 7006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041959775405129706,
+      "loss": 2.9123,
+      "step": 7007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 0.00041946889348895924,
+      "loss": 2.951,
+      "step": 7008
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041934003841772883,
+      "loss": 3.0696,
+      "step": 7009
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041921118884639236,
+      "loss": 2.8359,
+      "step": 7010
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004190823447837352,
+      "loss": 2.9334,
+      "step": 7011
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004189535062385432,
+      "loss": 3.0873,
+      "step": 7012
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004188246732196008,
+      "loss": 3.1844,
+      "step": 7013
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041869584573569314,
+      "loss": 2.8457,
+      "step": 7014
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041856702379560406,
+      "loss": 2.9913,
+      "step": 7015
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041843820740811795,
+      "loss": 2.8923,
+      "step": 7016
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004183093965820177,
+      "loss": 2.8097,
+      "step": 7017
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041818059132608704,
+      "loss": 2.8883,
+      "step": 7018
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004180517916491081,
+      "loss": 2.9861,
+      "step": 7019
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004179229975598638,
+      "loss": 3.0176,
+      "step": 7020
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041779420906713554,
+      "loss": 3.1003,
+      "step": 7021
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041766542617970536,
+      "loss": 3.1162,
+      "step": 7022
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004175366489063543,
+      "loss": 2.8187,
+      "step": 7023
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004174078772558634,
+      "loss": 2.9682,
+      "step": 7024
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004172791112370126,
+      "loss": 2.7074,
+      "step": 7025
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004171503508585826,
+      "loss": 3.1206,
+      "step": 7026
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041702159612935254,
+      "loss": 2.7963,
+      "step": 7027
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041689284705810213,
+      "loss": 2.8993,
+      "step": 7028
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041676410365361,
+      "loss": 2.7964,
+      "step": 7029
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004166353659246546,
+      "loss": 2.8816,
+      "step": 7030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004165066338800145,
+      "loss": 3.026,
+      "step": 7031
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004163779075284669,
+      "loss": 2.9207,
+      "step": 7032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004162491868787896,
+      "loss": 2.9068,
+      "step": 7033
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041612047193975916,
+      "loss": 2.6826,
+      "step": 7034
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004159917627201526,
+      "loss": 3.0043,
+      "step": 7035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041586305922874575,
+      "loss": 2.7776,
+      "step": 7036
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004157343614743145,
+      "loss": 3.1051,
+      "step": 7037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004156056694656344,
+      "loss": 3.009,
+      "step": 7038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004154769832114804,
+      "loss": 3.0566,
+      "step": 7039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041534830272062695,
+      "loss": 2.7555,
+      "step": 7040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041521962800184864,
+      "loss": 2.9008,
+      "step": 7041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004150909590639189,
+      "loss": 2.9912,
+      "step": 7042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041496229591561157,
+      "loss": 2.9153,
+      "step": 7043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004148336385656992,
+      "loss": 2.9488,
+      "step": 7044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004147049870229549,
+      "loss": 2.8485,
+      "step": 7045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004145763412961506,
+      "loss": 2.9401,
+      "step": 7046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004144477013940586,
+      "loss": 3.1975,
+      "step": 7047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041431906732544976,
+      "loss": 3.0378,
+      "step": 7048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004141904390990958,
+      "loss": 3.2466,
+      "step": 7049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041406181672376676,
+      "loss": 2.9603,
+      "step": 7050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041393320020823325,
+      "loss": 3.1386,
+      "step": 7051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041380458956126513,
+      "loss": 2.7939,
+      "step": 7052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004136759847916318,
+      "loss": 2.8909,
+      "step": 7053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041354738590810224,
+      "loss": 3.1089,
+      "step": 7054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041341879291944547,
+      "loss": 2.7904,
+      "step": 7055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004132902058344293,
+      "loss": 3.1904,
+      "step": 7056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041316162466182194,
+      "loss": 2.9539,
+      "step": 7057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004130330494103905,
+      "loss": 2.9057,
+      "step": 7058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041290448008890235,
+      "loss": 2.841,
+      "step": 7059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041277591670612403,
+      "loss": 2.9567,
+      "step": 7060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041264735927082173,
+      "loss": 3.0446,
+      "step": 7061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041251880779176155,
+      "loss": 2.7693,
+      "step": 7062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041239026227770845,
+      "loss": 2.5754,
+      "step": 7063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004122617227374279,
+      "loss": 2.8473,
+      "step": 7064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041213318917968414,
+      "loss": 2.7934,
+      "step": 7065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004120046616132416,
+      "loss": 3.0019,
+      "step": 7066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041187614004686406,
+      "loss": 3.0177,
+      "step": 7067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004117476244893151,
+      "loss": 3.1268,
+      "step": 7068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004116191149493573,
+      "loss": 3.001,
+      "step": 7069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004114906114357536,
+      "loss": 3.1422,
+      "step": 7070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041136211395726585,
+      "loss": 2.864,
+      "step": 7071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041123362252265625,
+      "loss": 2.9541,
+      "step": 7072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004111051371406855,
+      "loss": 3.0502,
+      "step": 7073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004109766578201151,
+      "loss": 2.9145,
+      "step": 7074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004108481845697051,
+      "loss": 2.9701,
+      "step": 7075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004107197173982161,
+      "loss": 2.9144,
+      "step": 7076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041059125631440733,
+      "loss": 2.998,
+      "step": 7077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004104628013270385,
+      "loss": 2.7925,
+      "step": 7078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041033435244486793,
+      "loss": 3.1145,
+      "step": 7079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004102059096766546,
+      "loss": 2.9889,
+      "step": 7080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00041007747303115603,
+      "loss": 2.7872,
+      "step": 7081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004099490425171302,
+      "loss": 2.8628,
+      "step": 7082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000409820618143334,
+      "loss": 2.9907,
+      "step": 7083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004096921999185247,
+      "loss": 3.1216,
+      "step": 7084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000409563787851458,
+      "loss": 2.8818,
+      "step": 7085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004094353819508903,
+      "loss": 2.9244,
+      "step": 7086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004093069822255767,
+      "loss": 2.8702,
+      "step": 7087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040917858868427283,
+      "loss": 2.8649,
+      "step": 7088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004090502013357328,
+      "loss": 2.7637,
+      "step": 7089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040892182018871113,
+      "loss": 2.9176,
+      "step": 7090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004087934452519616,
+      "loss": 2.8582,
+      "step": 7091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004086650765342378,
+      "loss": 3.093,
+      "step": 7092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040853671404429226,
+      "loss": 3.1982,
+      "step": 7093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000408408357790878,
+      "loss": 2.8889,
+      "step": 7094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040828000778274695,
+      "loss": 2.9512,
+      "step": 7095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004081516640286505,
+      "loss": 2.9072,
+      "step": 7096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040802332653734033,
+      "loss": 3.0613,
+      "step": 7097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040789499531756703,
+      "loss": 2.8524,
+      "step": 7098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040776667037808133,
+      "loss": 2.795,
+      "step": 7099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004076383517276328,
+      "loss": 2.9952,
+      "step": 7100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004075100393749714,
+      "loss": 3.1196,
+      "step": 7101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040738173332884595,
+      "loss": 3.1753,
+      "step": 7102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004072534335980054,
+      "loss": 2.9242,
+      "step": 7103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004071251401911977,
+      "loss": 2.8253,
+      "step": 7104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000406996853117171,
+      "loss": 3.038,
+      "step": 7105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004068685723846724,
+      "loss": 2.6953,
+      "step": 7106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004067402980024493,
+      "loss": 2.9934,
+      "step": 7107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040661202997924775,
+      "loss": 2.6452,
+      "step": 7108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040648376832381437,
+      "loss": 3.1064,
+      "step": 7109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004063555130448943,
+      "loss": 2.9365,
+      "step": 7110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040622726415123306,
+      "loss": 3.034,
+      "step": 7111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004060990216515755,
+      "loss": 2.9125,
+      "step": 7112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004059707855546658,
+      "loss": 2.9836,
+      "step": 7113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000405842555869248,
+      "loss": 2.8248,
+      "step": 7114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040571433260406576,
+      "loss": 2.9251,
+      "step": 7115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004055861157678617,
+      "loss": 3.155,
+      "step": 7116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040545790536937893,
+      "loss": 2.8899,
+      "step": 7117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040532970141735915,
+      "loss": 3.1118,
+      "step": 7118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040520150392054445,
+      "loss": 2.9787,
+      "step": 7119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000405073312887676,
+      "loss": 3.0221,
+      "step": 7120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004049451283274947,
+      "loss": 2.8559,
+      "step": 7121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040481695024874076,
+      "loss": 3.0836,
+      "step": 7122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040468877866015464,
+      "loss": 2.9251,
+      "step": 7123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040456061357047523,
+      "loss": 2.7221,
+      "step": 7124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004044324549884422,
+      "loss": 3.1006,
+      "step": 7125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040430430292279364,
+      "loss": 2.8413,
+      "step": 7126
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004041761573822682,
+      "loss": 2.7809,
+      "step": 7127
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.00040404801837560356,
+      "loss": 3.1746,
+      "step": 7128
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004039198859115367,
+      "loss": 2.8043,
+      "step": 7129
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.000403791759998805,
+      "loss": 3.0157,
+      "step": 7130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004036636406461444,
+      "loss": 3.0926,
+      "step": 7131
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 0.0004035355278622912,
+      "loss": 3.0349,
+      "step": 7132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004034074216559805,
+      "loss": 3.0349,
+      "step": 7133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004032793220359479,
+      "loss": 2.7158,
+      "step": 7134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004031512290109276,
+      "loss": 2.895,
+      "step": 7135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040302314258965424,
+      "loss": 2.7679,
+      "step": 7136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004028950627808609,
+      "loss": 3.2509,
+      "step": 7137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004027669895932815,
+      "loss": 2.9367,
+      "step": 7138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004026389230356483,
+      "loss": 2.9901,
+      "step": 7139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040251086311669424,
+      "loss": 3.0734,
+      "step": 7140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040238280984515064,
+      "loss": 2.8692,
+      "step": 7141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040225476322974937,
+      "loss": 2.9225,
+      "step": 7142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004021267232792212,
+      "loss": 2.8919,
+      "step": 7143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000401998690002297,
+      "loss": 3.1428,
+      "step": 7144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040187066340770645,
+      "loss": 2.9437,
+      "step": 7145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040174264350417965,
+      "loss": 2.6826,
+      "step": 7146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004016146303004453,
+      "loss": 2.8089,
+      "step": 7147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004014866238052326,
+      "loss": 3.1163,
+      "step": 7148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004013586240272694,
+      "loss": 3.2638,
+      "step": 7149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040123063097528383,
+      "loss": 3.2191,
+      "step": 7150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000401102644658003,
+      "loss": 3.0838,
+      "step": 7151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040097466508415416,
+      "loss": 2.8825,
+      "step": 7152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004008466922624633,
+      "loss": 3.0767,
+      "step": 7153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040071872620165685,
+      "loss": 2.6642,
+      "step": 7154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040059076691045987,
+      "loss": 2.7349,
+      "step": 7155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040046281439759784,
+      "loss": 2.9944,
+      "step": 7156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000400334868671795,
+      "loss": 3.0204,
+      "step": 7157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00040020692974177566,
+      "loss": 2.9556,
+      "step": 7158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0004000789976162636,
+      "loss": 3.013,
+      "step": 7159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039995107230398166,
+      "loss": 2.9734,
+      "step": 7160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039982315381365295,
+      "loss": 3.0714,
+      "step": 7161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003996952421539994,
+      "loss": 3.111,
+      "step": 7162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003995673373337433,
+      "loss": 3.3662,
+      "step": 7163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003994394393616054,
+      "loss": 2.7962,
+      "step": 7164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000399311548246307,
+      "loss": 3.0023,
+      "step": 7165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003991836639965682,
+      "loss": 2.8932,
+      "step": 7166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003990557866211094,
+      "loss": 2.742,
+      "step": 7167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003989279161286495,
+      "loss": 3.0259,
+      "step": 7168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000398800052527908,
+      "loss": 2.8058,
+      "step": 7169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000398672195827603,
+      "loss": 3.0479,
+      "step": 7170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003985443460364528,
+      "loss": 3.015,
+      "step": 7171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003984165031631751,
+      "loss": 3.1337,
+      "step": 7172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003982886672164867,
+      "loss": 2.9887,
+      "step": 7173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003981608382051044,
+      "loss": 2.8293,
+      "step": 7174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039803301613774457,
+      "loss": 2.905,
+      "step": 7175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003979052010231225,
+      "loss": 2.9719,
+      "step": 7176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039777739286995384,
+      "loss": 3.0754,
+      "step": 7177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039764959168695293,
+      "loss": 2.987,
+      "step": 7178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003975217974828343,
+      "loss": 3.0252,
+      "step": 7179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003973940102663117,
+      "loss": 2.859,
+      "step": 7180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039726623004609845,
+      "loss": 2.8428,
+      "step": 7181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003971384568309073,
+      "loss": 2.8207,
+      "step": 7182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039701069062945087,
+      "loss": 2.7817,
+      "step": 7183
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003968829314504407,
+      "loss": 2.8367,
+      "step": 7184
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039675517930258856,
+      "loss": 2.9955,
+      "step": 7185
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039662743419460495,
+      "loss": 2.919,
+      "step": 7186
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003964996961352007,
+      "loss": 3.1385,
+      "step": 7187
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039637196513308566,
+      "loss": 2.9353,
+      "step": 7188
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039624424119696927,
+      "loss": 2.9383,
+      "step": 7189
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039611652433556044,
+      "loss": 3.0235,
+      "step": 7190
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003959888145575681,
+      "loss": 2.971,
+      "step": 7191
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039586111187169996,
+      "loss": 3.0661,
+      "step": 7192
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003957334162866634,
+      "loss": 2.7569,
+      "step": 7193
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003956057278111658,
+      "loss": 2.9069,
+      "step": 7194
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003954780464539137,
+      "loss": 3.0126,
+      "step": 7195
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039535037222361325,
+      "loss": 2.7797,
+      "step": 7196
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003952227051289698,
+      "loss": 2.9933,
+      "step": 7197
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003950950451786889,
+      "loss": 3.018,
+      "step": 7198
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003949673923814747,
+      "loss": 2.8096,
+      "step": 7199
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039483974674603184,
+      "loss": 2.9222,
+      "step": 7200
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039471210828106356,
+      "loss": 3.0615,
+      "step": 7201
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003945844769952734,
+      "loss": 2.751,
+      "step": 7202
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039445685289736374,
+      "loss": 2.8699,
+      "step": 7203
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039432923599603707,
+      "loss": 2.9736,
+      "step": 7204
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003942016262999948,
+      "loss": 2.7164,
+      "step": 7205
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039407402381793857,
+      "loss": 3.015,
+      "step": 7206
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039394642855856857,
+      "loss": 2.8518,
+      "step": 7207
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003938188405305855,
+      "loss": 2.9751,
+      "step": 7208
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003936912597426886,
+      "loss": 2.7469,
+      "step": 7209
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003935636862035776,
+      "loss": 2.9566,
+      "step": 7210
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039343611992195094,
+      "loss": 2.9253,
+      "step": 7211
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039330856090650726,
+      "loss": 2.937,
+      "step": 7212
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003931810091659438,
+      "loss": 2.8325,
+      "step": 7213
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039305346470895825,
+      "loss": 2.8988,
+      "step": 7214
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039292592754424705,
+      "loss": 3.0446,
+      "step": 7215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039279839768050683,
+      "loss": 2.7983,
+      "step": 7216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000392670875126433,
+      "loss": 2.7872,
+      "step": 7217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000392543359890721,
+      "loss": 3.256,
+      "step": 7218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003924158519820656,
+      "loss": 3.0597,
+      "step": 7219
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003922883514091612,
+      "loss": 2.8206,
+      "step": 7220
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003921608581807013,
+      "loss": 3.0179,
+      "step": 7221
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003920333723053795,
+      "loss": 2.8785,
+      "step": 7222
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003919058937918882,
+      "loss": 2.8693,
+      "step": 7223
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003917784226489201,
+      "loss": 2.866,
+      "step": 7224
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039165095888516664,
+      "loss": 2.9127,
+      "step": 7225
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003915235025093191,
+      "loss": 3.0235,
+      "step": 7226
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003913960535300685,
+      "loss": 3.2271,
+      "step": 7227
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003912686119561047,
+      "loss": 2.9018,
+      "step": 7228
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039114117779611796,
+      "loss": 3.0214,
+      "step": 7229
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000391013751058797,
+      "loss": 2.7875,
+      "step": 7230
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039088633175283095,
+      "loss": 3.1819,
+      "step": 7231
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003907589198869078,
+      "loss": 2.9496,
+      "step": 7232
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003906315154697155,
+      "loss": 2.7689,
+      "step": 7233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003905041185099409,
+      "loss": 2.8209,
+      "step": 7234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003903767290162713,
+      "loss": 2.9805,
+      "step": 7235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00039024934699739236,
+      "loss": 2.9439,
+      "step": 7236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003901219724619901,
+      "loss": 2.747,
+      "step": 7237
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038999460541874943,
+      "loss": 2.6705,
+      "step": 7238
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003898672458763552,
+      "loss": 2.9965,
+      "step": 7239
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003897398938434916,
+      "loss": 2.6701,
+      "step": 7240
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038961254932884214,
+      "loss": 3.1023,
+      "step": 7241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038948521234109,
+      "loss": 3.0817,
+      "step": 7242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000389357882888918,
+      "loss": 3.1926,
+      "step": 7243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038923056098100786,
+      "loss": 3.0116,
+      "step": 7244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038910324662604155,
+      "loss": 2.9439,
+      "step": 7245
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038897593983269985,
+      "loss": 3.226,
+      "step": 7246
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038884864060966353,
+      "loss": 3.1887,
+      "step": 7247
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003887213489656126,
+      "loss": 3.1999,
+      "step": 7248
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038859406490922657,
+      "loss": 2.9185,
+      "step": 7249
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038846678844918427,
+      "loss": 2.823,
+      "step": 7250
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003883395195941646,
+      "loss": 2.9006,
+      "step": 7251
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.000388212258352845,
+      "loss": 2.8863,
+      "step": 7252
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038808500473390357,
+      "loss": 2.9838,
+      "step": 7253
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003879577587460166,
+      "loss": 3.0253,
+      "step": 7254
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.00038783052039786096,
+      "loss": 3.0768,
+      "step": 7255
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 0.0003877032896981124,
+      "loss": 3.0096,
+      "step": 7256
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003875760666554462,
+      "loss": 3.0405,
+      "step": 7257
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038744885127853744,
+      "loss": 2.8837,
+      "step": 7258
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038732164357606016,
+      "loss": 3.0932,
+      "step": 7259
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003871944435566885,
+      "loss": 2.8904,
+      "step": 7260
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003870672512290953,
+      "loss": 3.1221,
+      "step": 7261
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038694006660195367,
+      "loss": 2.849,
+      "step": 7262
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003868128896839357,
+      "loss": 2.9349,
+      "step": 7263
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038668572048371333,
+      "loss": 3.0841,
+      "step": 7264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038655855900995736,
+      "loss": 3.062,
+      "step": 7265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003864314052713388,
+      "loss": 3.1829,
+      "step": 7266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003863042592765274,
+      "loss": 3.0336,
+      "step": 7267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003861771210341932,
+      "loss": 2.904,
+      "step": 7268
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003860499905530048,
+      "loss": 3.0821,
+      "step": 7269
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003859228678416311,
+      "loss": 2.9082,
+      "step": 7270
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038579575290873994,
+      "loss": 3.2375,
+      "step": 7271
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038566864576299904,
+      "loss": 3.0391,
+      "step": 7272
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038554154641307493,
+      "loss": 2.6349,
+      "step": 7273
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038541445486763444,
+      "loss": 3.1713,
+      "step": 7274
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038528737113534305,
+      "loss": 2.9646,
+      "step": 7275
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038516029522486664,
+      "loss": 2.9979,
+      "step": 7276
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038503322714486944,
+      "loss": 3.1255,
+      "step": 7277
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003849061669040161,
+      "loss": 2.773,
+      "step": 7278
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003847791145109703,
+      "loss": 2.6473,
+      "step": 7279
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003846520699743954,
+      "loss": 2.9924,
+      "step": 7280
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038452503330295375,
+      "loss": 2.8506,
+      "step": 7281
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038439800450530784,
+      "loss": 2.9152,
+      "step": 7282
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038427098359011884,
+      "loss": 2.9217,
+      "step": 7283
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038414397056604836,
+      "loss": 3.0794,
+      "step": 7284
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003840169654417565,
+      "loss": 2.8489,
+      "step": 7285
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003838899682259035,
+      "loss": 2.9583,
+      "step": 7286
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038376297892714855,
+      "loss": 2.9383,
+      "step": 7287
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000383635997554151,
+      "loss": 2.9265,
+      "step": 7288
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003835090241155688,
+      "loss": 2.8454,
+      "step": 7289
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038338205862006,
+      "loss": 2.7717,
+      "step": 7290
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038325510107628194,
+      "loss": 2.7806,
+      "step": 7291
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038312815149289114,
+      "loss": 3.0183,
+      "step": 7292
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003830012098785439,
+      "loss": 3.0816,
+      "step": 7293
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003828742762418959,
+      "loss": 3.1662,
+      "step": 7294
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003827473505916025,
+      "loss": 2.9211,
+      "step": 7295
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003826204329363178,
+      "loss": 2.999,
+      "step": 7296
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003824935232846962,
+      "loss": 3.1054,
+      "step": 7297
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038236662164539096,
+      "loss": 2.7592,
+      "step": 7298
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038223972802705513,
+      "loss": 2.9054,
+      "step": 7299
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003821128424383411,
+      "loss": 2.8062,
+      "step": 7300
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003819859648879007,
+      "loss": 2.8958,
+      "step": 7301
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003818590953843852,
+      "loss": 2.8985,
+      "step": 7302
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038173223393644554,
+      "loss": 3.0767,
+      "step": 7303
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038160538055273145,
+      "loss": 2.7818,
+      "step": 7304
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003814785352418931,
+      "loss": 2.7551,
+      "step": 7305
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003813516980125792,
+      "loss": 3.0015,
+      "step": 7306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003812248688734385,
+      "loss": 3.0962,
+      "step": 7307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038109804783311905,
+      "loss": 2.8767,
+      "step": 7308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003809712349002681,
+      "loss": 2.8468,
+      "step": 7309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038084443008353246,
+      "loss": 2.7626,
+      "step": 7310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003807176333915589,
+      "loss": 2.7931,
+      "step": 7311
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003805908448329927,
+      "loss": 2.891,
+      "step": 7312
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003804640644164796,
+      "loss": 2.9415,
+      "step": 7313
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038033729215066365,
+      "loss": 2.7878,
+      "step": 7314
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00038021052804418945,
+      "loss": 2.9496,
+      "step": 7315
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003800837721057004,
+      "loss": 3.1636,
+      "step": 7316
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037995702434383955,
+      "loss": 2.8283,
+      "step": 7317
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037983028476724917,
+      "loss": 3.0898,
+      "step": 7318
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037970355338457146,
+      "loss": 2.8244,
+      "step": 7319
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003795768302044474,
+      "loss": 2.8332,
+      "step": 7320
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003794501152355181,
+      "loss": 2.658,
+      "step": 7321
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037932340848642344,
+      "loss": 3.0779,
+      "step": 7322
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037919670996580326,
+      "loss": 2.9782,
+      "step": 7323
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003790700196822967,
+      "loss": 3.171,
+      "step": 7324
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000378943337644542,
+      "loss": 2.8558,
+      "step": 7325
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003788166638611776,
+      "loss": 2.9084,
+      "step": 7326
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003786899983408403,
+      "loss": 2.9913,
+      "step": 7327
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037856334109216753,
+      "loss": 2.9797,
+      "step": 7328
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003784366921237951,
+      "loss": 3.1907,
+      "step": 7329
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037831005144435895,
+      "loss": 2.8964,
+      "step": 7330
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037818341906249405,
+      "loss": 2.9343,
+      "step": 7331
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003780567949868353,
+      "loss": 2.8206,
+      "step": 7332
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037793017922601636,
+      "loss": 2.849,
+      "step": 7333
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000377803571788671,
+      "loss": 2.9796,
+      "step": 7334
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037767697268343176,
+      "loss": 3.162,
+      "step": 7335
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003775503819189313,
+      "loss": 2.9883,
+      "step": 7336
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037742379950380094,
+      "loss": 3.1965,
+      "step": 7337
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003772972254466722,
+      "loss": 3.1493,
+      "step": 7338
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003771706597561755,
+      "loss": 3.2945,
+      "step": 7339
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037704410244094103,
+      "loss": 2.7288,
+      "step": 7340
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000376917553509598,
+      "loss": 2.7242,
+      "step": 7341
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003767910129707757,
+      "loss": 3.1917,
+      "step": 7342
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003766644808331019,
+      "loss": 2.9883,
+      "step": 7343
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037653795710520485,
+      "loss": 3.0766,
+      "step": 7344
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003764114417957113,
+      "loss": 3.2037,
+      "step": 7345
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003762849349132481,
+      "loss": 2.9401,
+      "step": 7346
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003761584364664412,
+      "loss": 3.1208,
+      "step": 7347
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003760319464639163,
+      "loss": 3.0126,
+      "step": 7348
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037590546491429765,
+      "loss": 2.701,
+      "step": 7349
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037577899182621025,
+      "loss": 3.0913,
+      "step": 7350
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003756525272082772,
+      "loss": 2.8889,
+      "step": 7351
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037552607106912215,
+      "loss": 2.8411,
+      "step": 7352
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037539962341736726,
+      "loss": 2.9642,
+      "step": 7353
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037527318426163484,
+      "loss": 2.907,
+      "step": 7354
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003751467536105459,
+      "loss": 2.986,
+      "step": 7355
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003750203314727217,
+      "loss": 2.8531,
+      "step": 7356
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003748939178567824,
+      "loss": 3.0654,
+      "step": 7357
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003747675127713472,
+      "loss": 2.8939,
+      "step": 7358
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003746411162250357,
+      "loss": 2.7786,
+      "step": 7359
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037451472822646627,
+      "loss": 3.0197,
+      "step": 7360
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037438834878425665,
+      "loss": 2.923,
+      "step": 7361
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003742619779070242,
+      "loss": 3.113,
+      "step": 7362
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037413561560338597,
+      "loss": 2.8869,
+      "step": 7363
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003740092618819577,
+      "loss": 2.7726,
+      "step": 7364
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037388291675135546,
+      "loss": 2.8047,
+      "step": 7365
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003737565802201936,
+      "loss": 2.9938,
+      "step": 7366
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037363025229708704,
+      "loss": 2.9822,
+      "step": 7367
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037350393299064944,
+      "loss": 3.1623,
+      "step": 7368
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037337762230949396,
+      "loss": 3.1287,
+      "step": 7369
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037325132026223325,
+      "loss": 2.8093,
+      "step": 7370
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003731250268574796,
+      "loss": 3.2366,
+      "step": 7371
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003729987421038441,
+      "loss": 2.9441,
+      "step": 7372
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037287246600993794,
+      "loss": 2.8365,
+      "step": 7373
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003727461985843712,
+      "loss": 2.8826,
+      "step": 7374
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037261993983575363,
+      "loss": 2.8633,
+      "step": 7375
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037249368977269437,
+      "loss": 2.8932,
+      "step": 7376
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.000372367448403802,
+      "loss": 3.0288,
+      "step": 7377
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003722412157376841,
+      "loss": 2.8074,
+      "step": 7378
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.0003721149917829485,
+      "loss": 2.9693,
+      "step": 7379
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 0.00037198877654820146,
+      "loss": 2.9726,
+      "step": 7380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037186257004204946,
+      "loss": 3.0131,
+      "step": 7381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037173637227309777,
+      "loss": 2.9351,
+      "step": 7382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037161018324995154,
+      "loss": 3.148,
+      "step": 7383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000371484002981215,
+      "loss": 2.9551,
+      "step": 7384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000371357831475492,
+      "loss": 2.9588,
+      "step": 7385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037123166874138537,
+      "loss": 2.9101,
+      "step": 7386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003711055147874982,
+      "loss": 2.9431,
+      "step": 7387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037097936962243196,
+      "loss": 2.9451,
+      "step": 7388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037085323325478835,
+      "loss": 2.781,
+      "step": 7389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003707271056931679,
+      "loss": 2.9703,
+      "step": 7390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037060098694617085,
+      "loss": 2.9444,
+      "step": 7391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037047487702239685,
+      "loss": 2.8278,
+      "step": 7392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037034877593044457,
+      "loss": 2.8039,
+      "step": 7393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003702226836789127,
+      "loss": 2.9151,
+      "step": 7394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00037009660027639867,
+      "loss": 3.0863,
+      "step": 7395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036997052573149996,
+      "loss": 2.7427,
+      "step": 7396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003698444600528128,
+      "loss": 2.9656,
+      "step": 7397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036971840324893327,
+      "loss": 2.8477,
+      "step": 7398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036959235532845657,
+      "loss": 3.0761,
+      "step": 7399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036946631629997784,
+      "loss": 2.8495,
+      "step": 7400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003693402861720907,
+      "loss": 2.9133,
+      "step": 7401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000369214264953389,
+      "loss": 2.9616,
+      "step": 7402
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003690882526524653,
+      "loss": 2.6762,
+      "step": 7403
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036896224927791235,
+      "loss": 2.9308,
+      "step": 7404
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003688362548383214,
+      "loss": 2.9228,
+      "step": 7405
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003687102693422838,
+      "loss": 3.1492,
+      "step": 7406
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003685842927983898,
+      "loss": 2.8885,
+      "step": 7407
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036845832521522964,
+      "loss": 3.1125,
+      "step": 7408
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003683323666013921,
+      "loss": 2.7425,
+      "step": 7409
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003682064169654663,
+      "loss": 2.9521,
+      "step": 7410
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003680804763160397,
+      "loss": 3.1006,
+      "step": 7411
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003679545446617003,
+      "loss": 3.3013,
+      "step": 7412
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003678286220110343,
+      "loss": 3.1271,
+      "step": 7413
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036770270837262843,
+      "loss": 2.7254,
+      "step": 7414
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003675768037550678,
+      "loss": 2.7347,
+      "step": 7415
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003674509081669378,
+      "loss": 3.0438,
+      "step": 7416
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003673250216168225,
+      "loss": 2.9215,
+      "step": 7417
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036719914411330567,
+      "loss": 3.1079,
+      "step": 7418
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003670732756649703,
+      "loss": 2.9973,
+      "step": 7419
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003669474162803992,
+      "loss": 2.771,
+      "step": 7420
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036682156596817383,
+      "loss": 3.0377,
+      "step": 7421
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036669572473687576,
+      "loss": 3.1206,
+      "step": 7422
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003665698925950856,
+      "loss": 2.7895,
+      "step": 7423
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003664440695513831,
+      "loss": 3.193,
+      "step": 7424
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036631825561434796,
+      "loss": 2.786,
+      "step": 7425
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003661924507925586,
+      "loss": 2.8768,
+      "step": 7426
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036606665509459355,
+      "loss": 2.7854,
+      "step": 7427
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003659408685290301,
+      "loss": 2.8764,
+      "step": 7428
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036581509110444524,
+      "loss": 3.2085,
+      "step": 7429
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036568932282941515,
+      "loss": 3.0104,
+      "step": 7430
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036556356371251575,
+      "loss": 3.0809,
+      "step": 7431
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036543781376232167,
+      "loss": 2.9723,
+      "step": 7432
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036531207298740775,
+      "loss": 2.7036,
+      "step": 7433
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003651863413963473,
+      "loss": 2.7806,
+      "step": 7434
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036506061899771393,
+      "loss": 3.0431,
+      "step": 7435
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036493490580007993,
+      "loss": 2.8906,
+      "step": 7436
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036480920181201725,
+      "loss": 3.0327,
+      "step": 7437
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003646835070420971,
+      "loss": 2.9999,
+      "step": 7438
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036455782149889034,
+      "loss": 2.9152,
+      "step": 7439
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036443214519096655,
+      "loss": 2.9648,
+      "step": 7440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003643064781268958,
+      "loss": 2.9998,
+      "step": 7441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036418082031524604,
+      "loss": 2.6674,
+      "step": 7442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003640551717645859,
+      "loss": 2.9858,
+      "step": 7443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036392953248348285,
+      "loss": 3.0642,
+      "step": 7444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036380390248050363,
+      "loss": 3.1267,
+      "step": 7445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003636782817642145,
+      "loss": 2.6744,
+      "step": 7446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003635526703431812,
+      "loss": 2.9939,
+      "step": 7447
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036342706822596827,
+      "loss": 2.8483,
+      "step": 7448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003633014754211406,
+      "loss": 2.6933,
+      "step": 7449
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036317589193726144,
+      "loss": 2.9135,
+      "step": 7450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000363050317782894,
+      "loss": 3.1902,
+      "step": 7451
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003629247529666008,
+      "loss": 3.0402,
+      "step": 7452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036279919749694346,
+      "loss": 3.0593,
+      "step": 7453
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036267365138248337,
+      "loss": 3.3078,
+      "step": 7454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003625481146317806,
+      "loss": 2.7338,
+      "step": 7455
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003624225872533956,
+      "loss": 2.9383,
+      "step": 7456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000362297069255887,
+      "loss": 3.1271,
+      "step": 7457
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036217156064781376,
+      "loss": 2.9091,
+      "step": 7458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036204606143773373,
+      "loss": 2.7433,
+      "step": 7459
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036192057163420444,
+      "loss": 3.07,
+      "step": 7460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003617950912457821,
+      "loss": 3.166,
+      "step": 7461
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003616696202810232,
+      "loss": 2.8368,
+      "step": 7462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003615441587484828,
+      "loss": 2.8658,
+      "step": 7463
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003614187066567159,
+      "loss": 2.998,
+      "step": 7464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036129326401427626,
+      "loss": 2.7023,
+      "step": 7465
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036116783082971765,
+      "loss": 2.914,
+      "step": 7466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036104240711159265,
+      "loss": 2.9621,
+      "step": 7467
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003609169928684538,
+      "loss": 2.9094,
+      "step": 7468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036079158810885205,
+      "loss": 2.7524,
+      "step": 7469
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003606661928413388,
+      "loss": 2.9694,
+      "step": 7470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003605408070744639,
+      "loss": 2.9255,
+      "step": 7471
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003604154308167773,
+      "loss": 2.8799,
+      "step": 7472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036029006407682743,
+      "loss": 2.9577,
+      "step": 7473
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036016470686316295,
+      "loss": 2.7517,
+      "step": 7474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00036003935918433124,
+      "loss": 2.8626,
+      "step": 7475
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003599140210488796,
+      "loss": 2.9917,
+      "step": 7476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035978869246535404,
+      "loss": 2.9594,
+      "step": 7477
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035966337344230055,
+      "loss": 2.925,
+      "step": 7478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003595380639882637,
+      "loss": 2.8787,
+      "step": 7479
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003594127641117884,
+      "loss": 2.9112,
+      "step": 7480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003592874738214179,
+      "loss": 3.0989,
+      "step": 7481
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035916219312569556,
+      "loss": 2.669,
+      "step": 7482
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003590369220331635,
+      "loss": 2.7531,
+      "step": 7483
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000358911660552364,
+      "loss": 2.9772,
+      "step": 7484
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003587864086918375,
+      "loss": 3.0018,
+      "step": 7485
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003586611664601251,
+      "loss": 3.0609,
+      "step": 7486
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003585359338657662,
+      "loss": 3.2234,
+      "step": 7487
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003584107109172999,
+      "loss": 2.7256,
+      "step": 7488
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035828549762326477,
+      "loss": 3.2126,
+      "step": 7489
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035816029399219853,
+      "loss": 3.1529,
+      "step": 7490
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003580351000326386,
+      "loss": 3.0703,
+      "step": 7491
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003579099157531211,
+      "loss": 2.861,
+      "step": 7492
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035778474116218234,
+      "loss": 2.8542,
+      "step": 7493
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.000357659576268357,
+      "loss": 2.9947,
+      "step": 7494
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035753442108017985,
+      "loss": 2.8227,
+      "step": 7495
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003574092756061848,
+      "loss": 2.6991,
+      "step": 7496
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003572841398549049,
+      "loss": 2.9144,
+      "step": 7497
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.0003571590138348727,
+      "loss": 2.9494,
+      "step": 7498
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035703389755462024,
+      "loss": 3.0288,
+      "step": 7499
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035690879102267837,
+      "loss": 2.9252,
+      "step": 7500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035678369424757805,
+      "loss": 2.8804,
+      "step": 7501
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035665860723784875,
+      "loss": 3.1688,
+      "step": 7502
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035653353000201994,
+      "loss": 2.7418,
+      "step": 7503
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 0.00035640846254862007,
+      "loss": 2.9288,
+      "step": 7504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000356283404886177,
+      "loss": 3.0204,
+      "step": 7505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003561583570232179,
+      "loss": 2.7533,
+      "step": 7506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035603331896826945,
+      "loss": 2.8693,
+      "step": 7507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003559082907298573,
+      "loss": 2.6886,
+      "step": 7508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035578327231650694,
+      "loss": 2.746,
+      "step": 7509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003556582637367425,
+      "loss": 3.0778,
+      "step": 7510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003555332649990881,
+      "loss": 3.0031,
+      "step": 7511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000355408276112067,
+      "loss": 2.8981,
+      "step": 7512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035528329708420155,
+      "loss": 2.7659,
+      "step": 7513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035515832792401347,
+      "loss": 2.7697,
+      "step": 7514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035503336864002436,
+      "loss": 3.0965,
+      "step": 7515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003549084192407542,
+      "loss": 3.014,
+      "step": 7516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035478347973472336,
+      "loss": 2.8537,
+      "step": 7517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003546585501304504,
+      "loss": 3.0462,
+      "step": 7518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003545336304364542,
+      "loss": 2.9756,
+      "step": 7519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035440872066125253,
+      "loss": 2.962,
+      "step": 7520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003542838208133622,
+      "loss": 2.955,
+      "step": 7521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035415893090130003,
+      "loss": 3.2504,
+      "step": 7522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003540340509335815,
+      "loss": 2.9855,
+      "step": 7523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003539091809187219,
+      "loss": 2.9286,
+      "step": 7524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003537843208652355,
+      "loss": 3.0355,
+      "step": 7525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003536594707816361,
+      "loss": 2.9791,
+      "step": 7526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035353463067643655,
+      "loss": 3.0327,
+      "step": 7527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003534098005581497,
+      "loss": 3.0576,
+      "step": 7528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035328498043528666,
+      "loss": 2.9012,
+      "step": 7529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035316017031635904,
+      "loss": 3.0194,
+      "step": 7530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003530353702098765,
+      "loss": 3.0426,
+      "step": 7531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003529105801243493,
+      "loss": 2.8997,
+      "step": 7532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035278580006828586,
+      "loss": 2.9399,
+      "step": 7533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035266103005019485,
+      "loss": 3.0321,
+      "step": 7534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003525362700785836,
+      "loss": 3.0765,
+      "step": 7535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003524115201619593,
+      "loss": 2.7568,
+      "step": 7536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035228678030882786,
+      "loss": 2.8108,
+      "step": 7537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035216205052769514,
+      "loss": 2.9914,
+      "step": 7538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003520373308270656,
+      "loss": 2.9534,
+      "step": 7539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035191262121544374,
+      "loss": 2.9684,
+      "step": 7540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003517879217013327,
+      "loss": 2.7544,
+      "step": 7541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003516632322932356,
+      "loss": 3.0916,
+      "step": 7542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035153855299965424,
+      "loss": 2.9411,
+      "step": 7543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035141388382909044,
+      "loss": 2.8796,
+      "step": 7544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035128922479004434,
+      "loss": 3.2176,
+      "step": 7545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035116457589101657,
+      "loss": 2.9353,
+      "step": 7546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035103993714050586,
+      "loss": 2.9311,
+      "step": 7547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003509153085470114,
+      "loss": 3.0204,
+      "step": 7548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035079069011903063,
+      "loss": 2.812,
+      "step": 7549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003506660818650612,
+      "loss": 2.735,
+      "step": 7550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003505414837935994,
+      "loss": 2.9907,
+      "step": 7551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003504168959131414,
+      "loss": 3.1002,
+      "step": 7552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035029231823218224,
+      "loss": 2.882,
+      "step": 7553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00035016775075921605,
+      "loss": 2.8663,
+      "step": 7554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003500431935027371,
+      "loss": 2.7703,
+      "step": 7555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034991864647123815,
+      "loss": 2.8496,
+      "step": 7556
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003497941096732118,
+      "loss": 2.8498,
+      "step": 7557
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003496695831171495,
+      "loss": 3.1558,
+      "step": 7558
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003495450668115426,
+      "loss": 3.0164,
+      "step": 7559
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034942056076488106,
+      "loss": 3.0864,
+      "step": 7560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003492960649856547,
+      "loss": 3.0435,
+      "step": 7561
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034917157948235203,
+      "loss": 2.6061,
+      "step": 7562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034904710426346164,
+      "loss": 3.035,
+      "step": 7563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003489226393374709,
+      "loss": 2.9369,
+      "step": 7564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034879818471286663,
+      "loss": 3.0026,
+      "step": 7565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003486737403981347,
+      "loss": 3.0479,
+      "step": 7566
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034854930640176087,
+      "loss": 3.0507,
+      "step": 7567
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003484248827322294,
+      "loss": 2.8035,
+      "step": 7568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003483004693980247,
+      "loss": 3.0969,
+      "step": 7569
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034817606640762957,
+      "loss": 3.0244,
+      "step": 7570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034805167376952687,
+      "loss": 2.6845,
+      "step": 7571
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003479272914921984,
+      "loss": 2.792,
+      "step": 7572
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034780291958412535,
+      "loss": 2.8596,
+      "step": 7573
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000347678558053788,
+      "loss": 2.8466,
+      "step": 7574
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003475542069096664,
+      "loss": 2.9915,
+      "step": 7575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003474298661602392,
+      "loss": 2.7267,
+      "step": 7576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034730553581398504,
+      "loss": 3.0946,
+      "step": 7577
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003471812158793812,
+      "loss": 2.8991,
+      "step": 7578
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034705690636490494,
+      "loss": 2.9835,
+      "step": 7579
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003469326072790322,
+      "loss": 2.9522,
+      "step": 7580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034680831863023867,
+      "loss": 2.934,
+      "step": 7581
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034668404042699876,
+      "loss": 2.9907,
+      "step": 7582
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003465597726777869,
+      "loss": 2.8822,
+      "step": 7583
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003464355153910762,
+      "loss": 2.8105,
+      "step": 7584
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034631126857533957,
+      "loss": 2.8325,
+      "step": 7585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003461870322390486,
+      "loss": 2.9483,
+      "step": 7586
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034606280639067457,
+      "loss": 3.1114,
+      "step": 7587
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034593859103868805,
+      "loss": 2.989,
+      "step": 7588
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003458143861915587,
+      "loss": 2.843,
+      "step": 7589
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034569019185775587,
+      "loss": 2.9395,
+      "step": 7590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034556600804574734,
+      "loss": 2.9281,
+      "step": 7591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034544183476400135,
+      "loss": 2.8519,
+      "step": 7592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003453176720209843,
+      "loss": 2.9612,
+      "step": 7593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034519351982516266,
+      "loss": 2.7836,
+      "step": 7594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003450693781850017,
+      "loss": 3.1973,
+      "step": 7595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034494524710896647,
+      "loss": 3.1083,
+      "step": 7596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034482112660552054,
+      "loss": 2.7842,
+      "step": 7597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003446970166831278,
+      "loss": 3.0016,
+      "step": 7598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003445729173502502,
+      "loss": 2.8099,
+      "step": 7599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003444488286153501,
+      "loss": 2.7899,
+      "step": 7600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034432475048688826,
+      "loss": 2.8445,
+      "step": 7601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034420068297332543,
+      "loss": 3.0277,
+      "step": 7602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003440766260831209,
+      "loss": 2.8195,
+      "step": 7603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034395257982473414,
+      "loss": 3.1191,
+      "step": 7604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003438285442066229,
+      "loss": 3.1772,
+      "step": 7605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003437045192372451,
+      "loss": 2.9069,
+      "step": 7606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034358050492505713,
+      "loss": 3.2743,
+      "step": 7607
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034345650127851545,
+      "loss": 2.9733,
+      "step": 7608
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.000343332508306075,
+      "loss": 2.8492,
+      "step": 7609
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003432085260161907,
+      "loss": 3.0894,
+      "step": 7610
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003430845544173162,
+      "loss": 2.8822,
+      "step": 7611
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034296059351790497,
+      "loss": 2.9204,
+      "step": 7612
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003428366433264091,
+      "loss": 2.9451,
+      "step": 7613
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003427127038512805,
+      "loss": 3.0023,
+      "step": 7614
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034258877510096985,
+      "loss": 3.1006,
+      "step": 7615
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003424648570839277,
+      "loss": 2.9449,
+      "step": 7616
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003423409498086033,
+      "loss": 2.8615,
+      "step": 7617
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034221705328344555,
+      "loss": 2.842,
+      "step": 7618
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034209316751690255,
+      "loss": 2.882,
+      "step": 7619
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003419692925174213,
+      "loss": 2.9154,
+      "step": 7620
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003418454282934488,
+      "loss": 2.9014,
+      "step": 7621
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003417215748534303,
+      "loss": 2.9362,
+      "step": 7622
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034159773220581135,
+      "loss": 2.869,
+      "step": 7623
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.00034147390035903626,
+      "loss": 2.969,
+      "step": 7624
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003413500793215485,
+      "loss": 2.8783,
+      "step": 7625
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003412262691017909,
+      "loss": 2.7867,
+      "step": 7626
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003411024697082059,
+      "loss": 3.0731,
+      "step": 7627
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 0.0003409786811492346,
+      "loss": 2.9534,
+      "step": 7628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034085490343331793,
+      "loss": 3.0022,
+      "step": 7629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034073113656889545,
+      "loss": 2.933,
+      "step": 7630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034060738056440674,
+      "loss": 3.2294,
+      "step": 7631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003404836354282901,
+      "loss": 2.7085,
+      "step": 7632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003403599011689831,
+      "loss": 3.0742,
+      "step": 7633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003402361777949229,
+      "loss": 2.7902,
+      "step": 7634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00034011246531454574,
+      "loss": 3.0156,
+      "step": 7635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033998876373628683,
+      "loss": 2.7804,
+      "step": 7636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033986507306858125,
+      "loss": 2.9722,
+      "step": 7637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003397413933198627,
+      "loss": 2.8867,
+      "step": 7638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033961772449856467,
+      "loss": 3.0851,
+      "step": 7639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003394940666131195,
+      "loss": 2.739,
+      "step": 7640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033937041967195905,
+      "loss": 2.9489,
+      "step": 7641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033924678368351416,
+      "loss": 2.9994,
+      "step": 7642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003391231586562153,
+      "loss": 3.0973,
+      "step": 7643
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033899954459849183,
+      "loss": 2.8112,
+      "step": 7644
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033887594151877276,
+      "loss": 2.9585,
+      "step": 7645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003387523494254857,
+      "loss": 2.8276,
+      "step": 7646
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033862876832705833,
+      "loss": 2.8171,
+      "step": 7647
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000338505198231917,
+      "loss": 2.8555,
+      "step": 7648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033838163914848746,
+      "loss": 3.0946,
+      "step": 7649
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003382580910851947,
+      "loss": 2.8904,
+      "step": 7650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003381345540504632,
+      "loss": 2.8458,
+      "step": 7651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033801102805271633,
+      "loss": 2.8885,
+      "step": 7652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033788751310037657,
+      "loss": 2.9447,
+      "step": 7653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003377640092018664,
+      "loss": 2.7088,
+      "step": 7654
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003376405163656068,
+      "loss": 3.0037,
+      "step": 7655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003375170346000185,
+      "loss": 2.8708,
+      "step": 7656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033739356391352093,
+      "loss": 2.8595,
+      "step": 7657
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033727010431453344,
+      "loss": 3.174,
+      "step": 7658
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003371466558114739,
+      "loss": 2.7173,
+      "step": 7659
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003370232184127602,
+      "loss": 3.0424,
+      "step": 7660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033689979212680863,
+      "loss": 2.9288,
+      "step": 7661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033677637696203543,
+      "loss": 2.967,
+      "step": 7662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003366529729268557,
+      "loss": 2.8717,
+      "step": 7663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033652958002968413,
+      "loss": 2.8334,
+      "step": 7664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000336406198278934,
+      "loss": 2.9982,
+      "step": 7665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033628282768301867,
+      "loss": 3.0103,
+      "step": 7666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033615946825034984,
+      "loss": 2.9749,
+      "step": 7667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003360361199893394,
+      "loss": 2.8823,
+      "step": 7668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003359127829083975,
+      "loss": 2.8856,
+      "step": 7669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033578945701593445,
+      "loss": 3.1595,
+      "step": 7670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003356661423203591,
+      "loss": 2.8875,
+      "step": 7671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003355428388300801,
+      "loss": 3.1181,
+      "step": 7672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003354195465535046,
+      "loss": 2.9109,
+      "step": 7673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033529626549903997,
+      "loss": 3.029,
+      "step": 7674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033517299567509175,
+      "loss": 2.9712,
+      "step": 7675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033504973709006566,
+      "loss": 2.4738,
+      "step": 7676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003349264897523658,
+      "loss": 2.7681,
+      "step": 7677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003348032536703963,
+      "loss": 3.2288,
+      "step": 7678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033468002885256,
+      "loss": 2.8963,
+      "step": 7679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033455681530725933,
+      "loss": 2.7798,
+      "step": 7680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033443361304289534,
+      "loss": 2.9166,
+      "step": 7681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003343104220678692,
+      "loss": 3.0826,
+      "step": 7682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033418724239058043,
+      "loss": 3.0338,
+      "step": 7683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033406407401942846,
+      "loss": 3.0292,
+      "step": 7684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033394091696281147,
+      "loss": 2.6674,
+      "step": 7685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033381777122912704,
+      "loss": 2.7684,
+      "step": 7686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033369463682677235,
+      "loss": 2.6459,
+      "step": 7687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033357151376414295,
+      "loss": 2.8907,
+      "step": 7688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003334484020496347,
+      "loss": 3.0,
+      "step": 7689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003333253016916416,
+      "loss": 2.8721,
+      "step": 7690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033320221269855766,
+      "loss": 2.7944,
+      "step": 7691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033307913507877587,
+      "loss": 2.8601,
+      "step": 7692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033295606884068833,
+      "loss": 3.0439,
+      "step": 7693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003328330139926864,
+      "loss": 2.9083,
+      "step": 7694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033270997054316096,
+      "loss": 2.7538,
+      "step": 7695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033258693850050155,
+      "loss": 2.7978,
+      "step": 7696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003324639178730976,
+      "loss": 2.8584,
+      "step": 7697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033234090866933696,
+      "loss": 3.0839,
+      "step": 7698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033221791089760755,
+      "loss": 2.8644,
+      "step": 7699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003320949245662961,
+      "loss": 3.1308,
+      "step": 7700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033197194968378834,
+      "loss": 2.9887,
+      "step": 7701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033184898625846946,
+      "loss": 2.9602,
+      "step": 7702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033172603429872436,
+      "loss": 2.9665,
+      "step": 7703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033160309381293597,
+      "loss": 2.8079,
+      "step": 7704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003314801648094876,
+      "loss": 3.0972,
+      "step": 7705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003313572472967611,
+      "loss": 3.1237,
+      "step": 7706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003312343412831378,
+      "loss": 2.9694,
+      "step": 7707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033111144677699825,
+      "loss": 3.0865,
+      "step": 7708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033098856378672206,
+      "loss": 2.8485,
+      "step": 7709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003308656923206881,
+      "loss": 2.8609,
+      "step": 7710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003307428323872748,
+      "loss": 2.9155,
+      "step": 7711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033061998399485916,
+      "loss": 2.7753,
+      "step": 7712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000330497147151818,
+      "loss": 2.8363,
+      "step": 7713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00033037432186652684,
+      "loss": 2.9824,
+      "step": 7714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003302515081473608,
+      "loss": 2.8945,
+      "step": 7715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003301287060026942,
+      "loss": 2.8605,
+      "step": 7716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003300059154409003,
+      "loss": 2.7025,
+      "step": 7717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032988313647035184,
+      "loss": 2.7492,
+      "step": 7718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003297603690994203,
+      "loss": 3.0536,
+      "step": 7719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032963761333647723,
+      "loss": 2.9869,
+      "step": 7720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032951486918989236,
+      "loss": 2.8095,
+      "step": 7721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003293921366680356,
+      "loss": 2.796,
+      "step": 7722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003292694157792752,
+      "loss": 2.9935,
+      "step": 7723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003291467065319795,
+      "loss": 3.0451,
+      "step": 7724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003290240089345151,
+      "loss": 3.1837,
+      "step": 7725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032890132299524864,
+      "loss": 3.0151,
+      "step": 7726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003287786487225453,
+      "loss": 2.7237,
+      "step": 7727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003286559861247701,
+      "loss": 3.039,
+      "step": 7728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003285333352102866,
+      "loss": 2.7225,
+      "step": 7729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003284106959874581,
+      "loss": 2.7466,
+      "step": 7730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032828806846464675,
+      "loss": 2.8649,
+      "step": 7731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003281654526502144,
+      "loss": 3.0156,
+      "step": 7732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003280428485525213,
+      "loss": 3.0689,
+      "step": 7733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003279202561799277,
+      "loss": 2.9171,
+      "step": 7734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003277976755407924,
+      "loss": 2.9484,
+      "step": 7735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003276751066434741,
+      "loss": 2.8618,
+      "step": 7736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003275525494963298,
+      "loss": 2.6728,
+      "step": 7737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032743000410771664,
+      "loss": 2.9788,
+      "step": 7738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032730747048599016,
+      "loss": 3.1787,
+      "step": 7739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003271849486395059,
+      "loss": 2.7933,
+      "step": 7740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032706243857661766,
+      "loss": 2.7834,
+      "step": 7741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003269399403056794,
+      "loss": 2.747,
+      "step": 7742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032681745383504346,
+      "loss": 3.0901,
+      "step": 7743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000326694979173062,
+      "loss": 2.8675,
+      "step": 7744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003265725163280857,
+      "loss": 3.0437,
+      "step": 7745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003264500653084652,
+      "loss": 2.8717,
+      "step": 7746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.0003263276261225498,
+      "loss": 2.9532,
+      "step": 7747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032620519877868837,
+      "loss": 3.0957,
+      "step": 7748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032608278328522856,
+      "loss": 3.0692,
+      "step": 7749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032596037965051727,
+      "loss": 2.8059,
+      "step": 7750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.000325837987882901,
+      "loss": 2.885,
+      "step": 7751
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 0.00032571560799072506,
+      "loss": 3.0991,
+      "step": 7752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003255932399823342,
+      "loss": 3.0329,
+      "step": 7753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003254708838660719,
+      "loss": 2.8156,
+      "step": 7754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032534853965028164,
+      "loss": 2.7374,
+      "step": 7755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003252262073433052,
+      "loss": 3.0405,
+      "step": 7756
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003251038869534842,
+      "loss": 3.1023,
+      "step": 7757
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003249815784891589,
+      "loss": 3.0479,
+      "step": 7758
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003248592819586693,
+      "loss": 2.8959,
+      "step": 7759
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032473699737035433,
+      "loss": 2.9569,
+      "step": 7760
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032461472473255205,
+      "loss": 2.8102,
+      "step": 7761
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003244924640535996,
+      "loss": 2.8623,
+      "step": 7762
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032437021534183387,
+      "loss": 2.9877,
+      "step": 7763
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032424797860559006,
+      "loss": 3.1812,
+      "step": 7764
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003241257538532034,
+      "loss": 2.955,
+      "step": 7765
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032400354109300765,
+      "loss": 2.7247,
+      "step": 7766
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032388134033333626,
+      "loss": 3.0948,
+      "step": 7767
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003237591515825215,
+      "loss": 2.9456,
+      "step": 7768
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000323636974848895,
+      "loss": 3.1015,
+      "step": 7769
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003235148101407874,
+      "loss": 2.8663,
+      "step": 7770
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003233926574665289,
+      "loss": 3.0664,
+      "step": 7771
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003232705168344483,
+      "loss": 2.9332,
+      "step": 7772
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003231483882528743,
+      "loss": 2.9225,
+      "step": 7773
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032302627173013397,
+      "loss": 2.9492,
+      "step": 7774
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003229041672745542,
+      "loss": 2.9532,
+      "step": 7775
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032278207489446067,
+      "loss": 2.9368,
+      "step": 7776
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032265999459817876,
+      "loss": 3.0167,
+      "step": 7777
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032253792639403214,
+      "loss": 2.9865,
+      "step": 7778
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003224158702903446,
+      "loss": 2.7908,
+      "step": 7779
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032229382629543835,
+      "loss": 2.9335,
+      "step": 7780
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003221717944176355,
+      "loss": 3.0506,
+      "step": 7781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003220497746652565,
+      "loss": 2.6258,
+      "step": 7782
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032192776704662165,
+      "loss": 3.0762,
+      "step": 7783
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003218057715700502,
+      "loss": 2.9018,
+      "step": 7784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032168378824386034,
+      "loss": 3.0888,
+      "step": 7785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032156181707637,
+      "loss": 2.9625,
+      "step": 7786
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032143985807589546,
+      "loss": 3.1323,
+      "step": 7787
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032131791125075313,
+      "loss": 2.8617,
+      "step": 7788
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032119597660925774,
+      "loss": 3.0255,
+      "step": 7789
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003210740541597237,
+      "loss": 3.2058,
+      "step": 7790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032095214391046433,
+      "loss": 2.9582,
+      "step": 7791
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032083024586979255,
+      "loss": 2.7814,
+      "step": 7792
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032070836004601954,
+      "loss": 2.7221,
+      "step": 7793
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003205864864474569,
+      "loss": 2.7829,
+      "step": 7794
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003204646250824141,
+      "loss": 2.9362,
+      "step": 7795
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003203427759592009,
+      "loss": 2.945,
+      "step": 7796
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00032022093908612534,
+      "loss": 2.8561,
+      "step": 7797
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003200991144714952,
+      "loss": 3.1222,
+      "step": 7798
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003199773021236172,
+      "loss": 3.0808,
+      "step": 7799
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003198555020507974,
+      "loss": 2.8974,
+      "step": 7800
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031973371426134055,
+      "loss": 2.9486,
+      "step": 7801
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031961193876355135,
+      "loss": 2.7798,
+      "step": 7802
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031949017556573265,
+      "loss": 2.7729,
+      "step": 7803
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031936842467618755,
+      "loss": 3.1075,
+      "step": 7804
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003192466861032174,
+      "loss": 2.9487,
+      "step": 7805
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031912495985512323,
+      "loss": 2.7312,
+      "step": 7806
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031900324594020505,
+      "loss": 2.984,
+      "step": 7807
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031888154436676225,
+      "loss": 2.9509,
+      "step": 7808
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031875985514309284,
+      "loss": 2.813,
+      "step": 7809
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031863817827749476,
+      "loss": 2.9583,
+      "step": 7810
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003185165137782643,
+      "loss": 2.9335,
+      "step": 7811
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003183948616536976,
+      "loss": 2.8386,
+      "step": 7812
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003182732219120894,
+      "loss": 3.1122,
+      "step": 7813
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000318151594561734,
+      "loss": 2.8178,
+      "step": 7814
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003180299796109248,
+      "loss": 2.8754,
+      "step": 7815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031790837706795395,
+      "loss": 3.1551,
+      "step": 7816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003177867869411134,
+      "loss": 3.0529,
+      "step": 7817
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031766520923869355,
+      "loss": 3.0825,
+      "step": 7818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031754364396898474,
+      "loss": 3.0569,
+      "step": 7819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031742209114027565,
+      "loss": 3.1778,
+      "step": 7820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003173005507608548,
+      "loss": 2.8525,
+      "step": 7821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031717902283900923,
+      "loss": 3.1049,
+      "step": 7822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000317057507383026,
+      "loss": 2.7504,
+      "step": 7823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003169360044011902,
+      "loss": 2.7776,
+      "step": 7824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031681451390178707,
+      "loss": 2.8648,
+      "step": 7825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031669303589310023,
+      "loss": 3.0152,
+      "step": 7826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003165715703834131,
+      "loss": 2.8993,
+      "step": 7827
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003164501173810078,
+      "loss": 2.8891,
+      "step": 7828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003163286768941658,
+      "loss": 2.8198,
+      "step": 7829
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003162072489311676,
+      "loss": 2.8534,
+      "step": 7830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031608583350029306,
+      "loss": 2.7621,
+      "step": 7831
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003159644306098208,
+      "loss": 2.9812,
+      "step": 7832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031584304026802915,
+      "loss": 3.0002,
+      "step": 7833
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003157216624831948,
+      "loss": 3.0708,
+      "step": 7834
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003156002972635944,
+      "loss": 3.1172,
+      "step": 7835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031547894461750315,
+      "loss": 2.749,
+      "step": 7836
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003153576045531959,
+      "loss": 2.7843,
+      "step": 7837
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000315236277078946,
+      "loss": 2.7229,
+      "step": 7838
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003151149622030267,
+      "loss": 3.0273,
+      "step": 7839
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031499365993370963,
+      "loss": 3.0328,
+      "step": 7840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003148723702792663,
+      "loss": 2.842,
+      "step": 7841
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003147510932479666,
+      "loss": 2.7421,
+      "step": 7842
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031462982884808015,
+      "loss": 2.7544,
+      "step": 7843
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031450857708787544,
+      "loss": 3.0953,
+      "step": 7844
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003143873379756204,
+      "loss": 2.6471,
+      "step": 7845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031426611151958144,
+      "loss": 2.9848,
+      "step": 7846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031414489772802503,
+      "loss": 2.897,
+      "step": 7847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003140236966092159,
+      "loss": 2.9777,
+      "step": 7848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003139025081714182,
+      "loss": 2.9555,
+      "step": 7849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031378133242289564,
+      "loss": 3.0871,
+      "step": 7850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031366016937191043,
+      "loss": 2.8752,
+      "step": 7851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031353901902672463,
+      "loss": 3.0581,
+      "step": 7852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003134178813955985,
+      "loss": 3.1757,
+      "step": 7853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031329675648679244,
+      "loss": 2.8821,
+      "step": 7854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031317564430856506,
+      "loss": 2.9595,
+      "step": 7855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031305454486917495,
+      "loss": 2.9437,
+      "step": 7856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003129334581768789,
+      "loss": 3.1055,
+      "step": 7857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031281238423993386,
+      "loss": 2.8551,
+      "step": 7858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003126913230665951,
+      "loss": 2.9612,
+      "step": 7859
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003125702746651175,
+      "loss": 2.8289,
+      "step": 7860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031244923904375464,
+      "loss": 3.093,
+      "step": 7861
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031232821621075986,
+      "loss": 2.9827,
+      "step": 7862
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003122072061743848,
+      "loss": 2.9181,
+      "step": 7863
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031208620894288106,
+      "loss": 3.0796,
+      "step": 7864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003119652245244986,
+      "loss": 2.9721,
+      "step": 7865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003118442529274872,
+      "loss": 3.0144,
+      "step": 7866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003117232941600951,
+      "loss": 2.8257,
+      "step": 7867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031160234823057064,
+      "loss": 3.1814,
+      "step": 7868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031148141514716,
+      "loss": 2.683,
+      "step": 7869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003113604949181096,
+      "loss": 2.9892,
+      "step": 7870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031123958755166416,
+      "loss": 2.8975,
+      "step": 7871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003111186930560682,
+      "loss": 3.2161,
+      "step": 7872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.0003109978114395648,
+      "loss": 2.8335,
+      "step": 7873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.00031087694271039664,
+      "loss": 2.9891,
+      "step": 7874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000310756086876805,
+      "loss": 2.8492,
+      "step": 7875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 0.000310635243947031,
+      "loss": 3.3574,
+      "step": 7876
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003105144139293141,
+      "loss": 2.9739,
+      "step": 7877
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003103935968318935,
+      "loss": 2.8025,
+      "step": 7878
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003102727926630069,
+      "loss": 2.7703,
+      "step": 7879
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003101520014308919,
+      "loss": 2.8062,
+      "step": 7880
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00031003122314378454,
+      "loss": 2.7629,
+      "step": 7881
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030991045780992007,
+      "loss": 3.0289,
+      "step": 7882
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003097897054375334,
+      "loss": 3.0101,
+      "step": 7883
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003096689660348578,
+      "loss": 2.9195,
+      "step": 7884
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030954823961012654,
+      "loss": 2.8043,
+      "step": 7885
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030942752617157086,
+      "loss": 3.0903,
+      "step": 7886
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003093068257274221,
+      "loss": 2.7091,
+      "step": 7887
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003091861382859104,
+      "loss": 2.7142,
+      "step": 7888
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030906546385526493,
+      "loss": 2.6832,
+      "step": 7889
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030894480244371376,
+      "loss": 2.8519,
+      "step": 7890
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030882415405948476,
+      "loss": 2.8183,
+      "step": 7891
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003087035187108041,
+      "loss": 2.8877,
+      "step": 7892
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030858289640589776,
+      "loss": 2.8444,
+      "step": 7893
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003084622871529902,
+      "loss": 3.0089,
+      "step": 7894
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003083416909603056,
+      "loss": 2.9037,
+      "step": 7895
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003082211078360666,
+      "loss": 3.1052,
+      "step": 7896
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003081005377884958,
+      "loss": 3.0606,
+      "step": 7897
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003079799808258139,
+      "loss": 2.9723,
+      "step": 7898
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003078594369562417,
+      "loss": 2.805,
+      "step": 7899
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003077389061879981,
+      "loss": 3.0212,
+      "step": 7900
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003076183885293021,
+      "loss": 3.0113,
+      "step": 7901
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030749788398837095,
+      "loss": 2.8813,
+      "step": 7902
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003073773925734217,
+      "loss": 3.0154,
+      "step": 7903
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030725691429266995,
+      "loss": 2.7507,
+      "step": 7904
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000307136449154331,
+      "loss": 2.8197,
+      "step": 7905
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030701599716661843,
+      "loss": 2.8085,
+      "step": 7906
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003068955583377459,
+      "loss": 3.0739,
+      "step": 7907
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003067751326759252,
+      "loss": 3.0661,
+      "step": 7908
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003066547201893681,
+      "loss": 3.0357,
+      "step": 7909
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003065343208862848,
+      "loss": 2.9811,
+      "step": 7910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000306413934774885,
+      "loss": 3.1046,
+      "step": 7911
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003062935618633772,
+      "loss": 2.9684,
+      "step": 7912
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030617320215996945,
+      "loss": 2.9849,
+      "step": 7913
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030605285567286857,
+      "loss": 2.8396,
+      "step": 7914
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030593252241028015,
+      "loss": 3.0474,
+      "step": 7915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030581220238040966,
+      "loss": 3.0678,
+      "step": 7916
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000305691895591461,
+      "loss": 2.9054,
+      "step": 7917
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003055716020516376,
+      "loss": 3.0056,
+      "step": 7918
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003054513217691417,
+      "loss": 3.0955,
+      "step": 7919
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000305331054752175,
+      "loss": 2.9582,
+      "step": 7920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030521080100893766,
+      "loss": 2.8282,
+      "step": 7921
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030509056054762974,
+      "loss": 2.7729,
+      "step": 7922
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030497033337644955,
+      "loss": 2.8739,
+      "step": 7923
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003048501195035953,
+      "loss": 2.8253,
+      "step": 7924
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003047299189372636,
+      "loss": 2.9113,
+      "step": 7925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030460973168565066,
+      "loss": 2.9632,
+      "step": 7926
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003044895577569515,
+      "loss": 3.0103,
+      "step": 7927
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003043693971593605,
+      "loss": 2.9326,
+      "step": 7928
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030424924990107063,
+      "loss": 3.0301,
+      "step": 7929
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030412911599027466,
+      "loss": 2.9454,
+      "step": 7930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030400899543516364,
+      "loss": 2.7797,
+      "step": 7931
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003038888882439286,
+      "loss": 2.8593,
+      "step": 7932
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003037687944247586,
+      "loss": 2.9986,
+      "step": 7933
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030364871398584295,
+      "loss": 3.0191,
+      "step": 7934
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000303528646935369,
+      "loss": 2.7857,
+      "step": 7935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003034085932815242,
+      "loss": 3.0298,
+      "step": 7936
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000303288553032494,
+      "loss": 2.8657,
+      "step": 7937
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030316852619646393,
+      "loss": 2.909,
+      "step": 7938
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030304851278161783,
+      "loss": 2.8459,
+      "step": 7939
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003029285127961392,
+      "loss": 2.9519,
+      "step": 7940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003028085262482101,
+      "loss": 2.7447,
+      "step": 7941
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003026885531460122,
+      "loss": 2.9263,
+      "step": 7942
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003025685934977258,
+      "loss": 2.8223,
+      "step": 7943
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003024486473115309,
+      "loss": 3.0729,
+      "step": 7944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003023287145956056,
+      "loss": 3.1586,
+      "step": 7945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003022087953581282,
+      "loss": 2.7446,
+      "step": 7946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003020888896072751,
+      "loss": 2.8857,
+      "step": 7947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003019689973512224,
+      "loss": 3.1365,
+      "step": 7948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003018491185981451,
+      "loss": 2.9759,
+      "step": 7949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003017292533562172,
+      "loss": 2.9563,
+      "step": 7950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030160940163361194,
+      "loss": 2.6368,
+      "step": 7951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003014895634385014,
+      "loss": 2.9958,
+      "step": 7952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030136973877905725,
+      "loss": 2.8058,
+      "step": 7953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003012499276634494,
+      "loss": 2.7818,
+      "step": 7954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003011301300998476,
+      "loss": 2.9322,
+      "step": 7955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003010103460964202,
+      "loss": 2.9271,
+      "step": 7956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030089057566133517,
+      "loss": 2.9232,
+      "step": 7957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030077081880275873,
+      "loss": 3.2199,
+      "step": 7958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003006510755288571,
+      "loss": 2.768,
+      "step": 7959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030053134584779467,
+      "loss": 2.8593,
+      "step": 7960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003004116297677358,
+      "loss": 2.9013,
+      "step": 7961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.000300291927296843,
+      "loss": 2.8287,
+      "step": 7962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00030017223844327866,
+      "loss": 2.633,
+      "step": 7963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0003000525632152037,
+      "loss": 2.9802,
+      "step": 7964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002999329016207787,
+      "loss": 2.7263,
+      "step": 7965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029981325366816233,
+      "loss": 2.8942,
+      "step": 7966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029969361936551343,
+      "loss": 2.8902,
+      "step": 7967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029957399872098913,
+      "loss": 2.742,
+      "step": 7968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002994543917427461,
+      "loss": 2.7994,
+      "step": 7969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002993347984389396,
+      "loss": 3.0549,
+      "step": 7970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029921521881772444,
+      "loss": 2.6827,
+      "step": 7971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002990956528872543,
+      "loss": 2.9389,
+      "step": 7972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029897610065568193,
+      "loss": 2.9916,
+      "step": 7973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029885656213115896,
+      "loss": 3.1659,
+      "step": 7974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002987370373218367,
+      "loss": 2.9061,
+      "step": 7975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029861752623586445,
+      "loss": 2.9536,
+      "step": 7976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029849802888139176,
+      "loss": 3.1024,
+      "step": 7977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002983785452665664,
+      "loss": 2.8941,
+      "step": 7978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029825907539953557,
+      "loss": 2.9775,
+      "step": 7979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002981396192884456,
+      "loss": 3.0809,
+      "step": 7980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029802017694144145,
+      "loss": 2.7208,
+      "step": 7981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002979007483666678,
+      "loss": 2.9796,
+      "step": 7982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002977813335722677,
+      "loss": 3.1052,
+      "step": 7983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002976619325663839,
+      "loss": 2.842,
+      "step": 7984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029754254535715747,
+      "loss": 3.0351,
+      "step": 7985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002974231719527294,
+      "loss": 2.8819,
+      "step": 7986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029730381236123894,
+      "loss": 2.8862,
+      "step": 7987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029718446659082515,
+      "loss": 2.8329,
+      "step": 7988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002970651346496254,
+      "loss": 3.2424,
+      "step": 7989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002969458165457768,
+      "loss": 2.7342,
+      "step": 7990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002968265122874149,
+      "loss": 2.9502,
+      "step": 7991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029670722188267483,
+      "loss": 2.9599,
+      "step": 7992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002965879453396904,
+      "loss": 2.875,
+      "step": 7993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002964686826665946,
+      "loss": 3.0163,
+      "step": 7994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002963494338715195,
+      "loss": 2.9584,
+      "step": 7995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.00029623019896259654,
+      "loss": 2.9053,
+      "step": 7996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002961109779479554,
+      "loss": 2.9429,
+      "step": 7997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002959917708357257,
+      "loss": 3.0859,
+      "step": 7998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002958725776340354,
+      "loss": 2.9483,
+      "step": 7999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 0.0002957533983510122,
+      "loss": 3.029,
+      "step": 8000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029563423299478196,
+      "loss": 2.5424,
+      "step": 8001
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002955150815734706,
+      "loss": 2.8832,
+      "step": 8002
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002953959440952023,
+      "loss": 2.7927,
+      "step": 8003
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002952768205681009,
+      "loss": 2.9007,
+      "step": 8004
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029515771100028853,
+      "loss": 2.802,
+      "step": 8005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002950386153998873,
+      "loss": 3.0527,
+      "step": 8006
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002949195337750174,
+      "loss": 2.9238,
+      "step": 8007
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000294800466133799,
+      "loss": 2.6556,
+      "step": 8008
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002946814124843504,
+      "loss": 2.8051,
+      "step": 8009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002945623728347899,
+      "loss": 2.6953,
+      "step": 8010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029444334719323393,
+      "loss": 2.8935,
+      "step": 8011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002943243355677988,
+      "loss": 2.9755,
+      "step": 8012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029420533796659933,
+      "loss": 2.7371,
+      "step": 8013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002940863543977491,
+      "loss": 2.9807,
+      "step": 8014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029396738486936165,
+      "loss": 2.9609,
+      "step": 8015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002938484293895487,
+      "loss": 2.8737,
+      "step": 8016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029372948796642183,
+      "loss": 2.9516,
+      "step": 8017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002936105606080907,
+      "loss": 2.6745,
+      "step": 8018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029349164732266487,
+      "loss": 2.9693,
+      "step": 8019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029337274811825227,
+      "loss": 3.0334,
+      "step": 8020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029325386300296054,
+      "loss": 2.8971,
+      "step": 8021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029313499198489556,
+      "loss": 3.1439,
+      "step": 8022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029301613507216315,
+      "loss": 3.0846,
+      "step": 8023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029289729227286734,
+      "loss": 2.9722,
+      "step": 8024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002927784635951119,
+      "loss": 2.9975,
+      "step": 8025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029265964904699894,
+      "loss": 2.7124,
+      "step": 8026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029254084863663036,
+      "loss": 2.86,
+      "step": 8027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002924220623721062,
+      "loss": 2.8989,
+      "step": 8028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029230329026152667,
+      "loss": 3.1265,
+      "step": 8029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002921845323129898,
+      "loss": 2.7259,
+      "step": 8030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029206578853459357,
+      "loss": 2.909,
+      "step": 8031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002919470589344345,
+      "loss": 2.889,
+      "step": 8032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002918283435206087,
+      "loss": 3.062,
+      "step": 8033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029170964230121023,
+      "loss": 3.1019,
+      "step": 8034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029159095528433345,
+      "loss": 3.0386,
+      "step": 8035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002914722824780709,
+      "loss": 2.9377,
+      "step": 8036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029135362389051455,
+      "loss": 3.194,
+      "step": 8037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029123497952975524,
+      "loss": 2.6855,
+      "step": 8038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002911163494038829,
+      "loss": 2.9589,
+      "step": 8039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029099773352098633,
+      "loss": 2.8329,
+      "step": 8040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002908791318891536,
+      "loss": 2.8847,
+      "step": 8041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002907605445164717,
+      "loss": 3.0918,
+      "step": 8042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029064197141102666,
+      "loss": 2.9934,
+      "step": 8043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002905234125809034,
+      "loss": 2.8595,
+      "step": 8044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002904048680341862,
+      "loss": 3.1835,
+      "step": 8045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000290286337778958,
+      "loss": 2.9634,
+      "step": 8046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00029016782182330106,
+      "loss": 3.0879,
+      "step": 8047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002900493201752963,
+      "loss": 2.9832,
+      "step": 8048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028993083284302413,
+      "loss": 2.8406,
+      "step": 8049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002898123598345636,
+      "loss": 2.9098,
+      "step": 8050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028969390115799284,
+      "loss": 2.8615,
+      "step": 8051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028957545682138956,
+      "loss": 2.8647,
+      "step": 8052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002894570268328294,
+      "loss": 3.1335,
+      "step": 8053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002893386112003883,
+      "loss": 3.0644,
+      "step": 8054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028922020993213984,
+      "loss": 3.1336,
+      "step": 8055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028910182303615805,
+      "loss": 3.0321,
+      "step": 8056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028898345052051465,
+      "loss": 2.8904,
+      "step": 8057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028886509239328147,
+      "loss": 2.9549,
+      "step": 8058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002887467486625287,
+      "loss": 2.8753,
+      "step": 8059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002886284193363259,
+      "loss": 2.9678,
+      "step": 8060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002885101044227413,
+      "loss": 2.7612,
+      "step": 8061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028839180392984244,
+      "loss": 2.9912,
+      "step": 8062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028827351786569577,
+      "loss": 2.9196,
+      "step": 8063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002881552462383667,
+      "loss": 3.0497,
+      "step": 8064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028803698905591977,
+      "loss": 3.0078,
+      "step": 8065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028791874632641854,
+      "loss": 3.1371,
+      "step": 8066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002878005180579253,
+      "loss": 2.8875,
+      "step": 8067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002876823042585021,
+      "loss": 2.9079,
+      "step": 8068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002875641049362088,
+      "loss": 3.1842,
+      "step": 8069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002874459200991056,
+      "loss": 2.9877,
+      "step": 8070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028732774975525043,
+      "loss": 2.854,
+      "step": 8071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028720959391270165,
+      "loss": 2.8652,
+      "step": 8072
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028709145257951494,
+      "loss": 2.7471,
+      "step": 8073
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002869733257637467,
+      "loss": 3.2024,
+      "step": 8074
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002868552134734512,
+      "loss": 2.9961,
+      "step": 8075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000286737115716682,
+      "loss": 2.9981,
+      "step": 8076
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002866190325014921,
+      "loss": 2.8343,
+      "step": 8077
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002865009638359325,
+      "loss": 2.8195,
+      "step": 8078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028638290972805435,
+      "loss": 3.0394,
+      "step": 8079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002862648701859072,
+      "loss": 2.9478,
+      "step": 8080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028614684521753976,
+      "loss": 3.1357,
+      "step": 8081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002860288348309995,
+      "loss": 2.8329,
+      "step": 8082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028591083903433334,
+      "loss": 3.0522,
+      "step": 8083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028579285783558684,
+      "loss": 2.7212,
+      "step": 8084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002856748912428047,
+      "loss": 3.0823,
+      "step": 8085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002855569392640305,
+      "loss": 2.935,
+      "step": 8086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002854390019073071,
+      "loss": 2.8415,
+      "step": 8087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028532107918067596,
+      "loss": 2.7975,
+      "step": 8088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002852031710921783,
+      "loss": 3.0306,
+      "step": 8089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028508527764985317,
+      "loss": 2.9743,
+      "step": 8090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002849673988617399,
+      "loss": 2.8141,
+      "step": 8091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002848495347358755,
+      "loss": 3.0105,
+      "step": 8092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002847316852802975,
+      "loss": 3.0507,
+      "step": 8093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002846138505030407,
+      "loss": 3.2054,
+      "step": 8094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028449603041214033,
+      "loss": 3.1642,
+      "step": 8095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028437822501563016,
+      "loss": 2.9027,
+      "step": 8096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002842604343215427,
+      "loss": 3.0056,
+      "step": 8097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028414265833790966,
+      "loss": 2.8283,
+      "step": 8098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002840248970727618,
+      "loss": 3.1498,
+      "step": 8099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028390715053412875,
+      "loss": 3.0776,
+      "step": 8100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028378941873003914,
+      "loss": 2.9789,
+      "step": 8101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028367170166852085,
+      "loss": 2.787,
+      "step": 8102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028355399935760034,
+      "loss": 3.07,
+      "step": 8103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002834363118053032,
+      "loss": 3.0722,
+      "step": 8104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028331863901965464,
+      "loss": 2.9523,
+      "step": 8105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002832009810086775,
+      "loss": 2.9751,
+      "step": 8106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002830833377803954,
+      "loss": 2.9455,
+      "step": 8107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002829657093428289,
+      "loss": 3.1204,
+      "step": 8108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002828480957039997,
+      "loss": 2.669,
+      "step": 8109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028273049687192666,
+      "loss": 2.9801,
+      "step": 8110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028261291285462843,
+      "loss": 2.831,
+      "step": 8111
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002824953436601232,
+      "loss": 3.1199,
+      "step": 8112
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028237778929642677,
+      "loss": 2.7091,
+      "step": 8113
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002822602497715556,
+      "loss": 2.9063,
+      "step": 8114
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028214272509352335,
+      "loss": 2.95,
+      "step": 8115
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002820252152703444,
+      "loss": 2.9289,
+      "step": 8116
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002819077203100306,
+      "loss": 3.0363,
+      "step": 8117
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.000281790240220594,
+      "loss": 2.8432,
+      "step": 8118
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028167277501004486,
+      "loss": 2.6656,
+      "step": 8119
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002815553246863927,
+      "loss": 3.0997,
+      "step": 8120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028143788925764614,
+      "loss": 2.8341,
+      "step": 8121
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002813204687318125,
+      "loss": 3.1125,
+      "step": 8122
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.00028120306311689826,
+      "loss": 2.6555,
+      "step": 8123
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 0.0002810856724209089,
+      "loss": 2.6561,
+      "step": 8124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002809682966518489,
+      "loss": 2.9371,
+      "step": 8125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028085093581772157,
+      "loss": 2.7624,
+      "step": 8126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028073358992652915,
+      "loss": 2.7131,
+      "step": 8127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028061625898627353,
+      "loss": 2.7955,
+      "step": 8128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00028049894300495435,
+      "loss": 3.0063,
+      "step": 8129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002803816419905717,
+      "loss": 2.8954,
+      "step": 8130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002802643559511231,
+      "loss": 2.9945,
+      "step": 8131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002801470848946064,
+      "loss": 2.9155,
+      "step": 8132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002800298288290176,
+      "loss": 2.7587,
+      "step": 8133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027991258776235213,
+      "loss": 2.9885,
+      "step": 8134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027979536170260407,
+      "loss": 2.7799,
+      "step": 8135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002796781506577666,
+      "loss": 2.7539,
+      "step": 8136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027956095463583196,
+      "loss": 2.9437,
+      "step": 8137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027944377364479134,
+      "loss": 2.9927,
+      "step": 8138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002793266076926348,
+      "loss": 2.8856,
+      "step": 8139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002792094567873514,
+      "loss": 3.0398,
+      "step": 8140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027909232093692926,
+      "loss": 2.6724,
+      "step": 8141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002789752001493554,
+      "loss": 2.8242,
+      "step": 8142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002788580944326159,
+      "loss": 2.9014,
+      "step": 8143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002787410037946956,
+      "loss": 2.9501,
+      "step": 8144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027862392824357866,
+      "loss": 2.794,
+      "step": 8145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002785068677872479,
+      "loss": 3.127,
+      "step": 8146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027838982243368526,
+      "loss": 3.0414,
+      "step": 8147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027827279219087137,
+      "loss": 3.1022,
+      "step": 8148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027815577706678677,
+      "loss": 2.8519,
+      "step": 8149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002780387770694093,
+      "loss": 3.0367,
+      "step": 8150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027792179220671764,
+      "loss": 2.973,
+      "step": 8151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002778048224866878,
+      "loss": 2.806,
+      "step": 8152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002776878679172963,
+      "loss": 3.0408,
+      "step": 8153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002775709285065169,
+      "loss": 2.6869,
+      "step": 8154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002774540042623239,
+      "loss": 3.2172,
+      "step": 8155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002773370951926897,
+      "loss": 2.8548,
+      "step": 8156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027722020130558603,
+      "loss": 2.9166,
+      "step": 8157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027710332260898326,
+      "loss": 2.9437,
+      "step": 8158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027698645911085096,
+      "loss": 2.9036,
+      "step": 8159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027686961081915763,
+      "loss": 3.0153,
+      "step": 8160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002767527777418706,
+      "loss": 2.9629,
+      "step": 8161
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027663595988695644,
+      "loss": 2.9755,
+      "step": 8162
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002765191572623804,
+      "loss": 2.9919,
+      "step": 8163
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002764023698761066,
+      "loss": 2.9364,
+      "step": 8164
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000276285597736099,
+      "loss": 2.7397,
+      "step": 8165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000276168840850319,
+      "loss": 3.0905,
+      "step": 8166
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002760520992267286,
+      "loss": 3.0284,
+      "step": 8167
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002759353728732871,
+      "loss": 2.9235,
+      "step": 8168
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027581866179795454,
+      "loss": 2.9048,
+      "step": 8169
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027570196600868813,
+      "loss": 2.9039,
+      "step": 8170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002755852855134455,
+      "loss": 3.0465,
+      "step": 8171
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002754686203201826,
+      "loss": 2.7444,
+      "step": 8172
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002753519704368541,
+      "loss": 2.9149,
+      "step": 8173
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002752353358714142,
+      "loss": 3.0571,
+      "step": 8174
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027511871663181555,
+      "loss": 3.0584,
+      "step": 8175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027500211272601027,
+      "loss": 2.8982,
+      "step": 8176
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027488552416194854,
+      "loss": 2.9103,
+      "step": 8177
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027476895094758066,
+      "loss": 3.026,
+      "step": 8178
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002746523930908551,
+      "loss": 3.0998,
+      "step": 8179
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027453585059971947,
+      "loss": 2.9923,
+      "step": 8180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002744193234821204,
+      "loss": 3.0125,
+      "step": 8181
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027430281174600345,
+      "loss": 2.8345,
+      "step": 8182
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027418631539931305,
+      "loss": 2.965,
+      "step": 8183
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027406983444999266,
+      "loss": 2.8364,
+      "step": 8184
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027395336890598476,
+      "loss": 2.9316,
+      "step": 8185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027383691877523055,
+      "loss": 2.9236,
+      "step": 8186
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002737204840656701,
+      "loss": 2.7962,
+      "step": 8187
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027360406478524346,
+      "loss": 2.914,
+      "step": 8188
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002734876609418878,
+      "loss": 2.8048,
+      "step": 8189
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027337127254354113,
+      "loss": 2.9373,
+      "step": 8190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002732548995981387,
+      "loss": 2.923,
+      "step": 8191
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027313854211361623,
+      "loss": 2.9021,
+      "step": 8192
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027302220009790736,
+      "loss": 3.0923,
+      "step": 8193
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002729058735589451,
+      "loss": 3.0482,
+      "step": 8194
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002727895625046613,
+      "loss": 2.7954,
+      "step": 8195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027267326694298673,
+      "loss": 3.0161,
+      "step": 8196
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027255698688185117,
+      "loss": 2.9377,
+      "step": 8197
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027244072232918327,
+      "loss": 3.0956,
+      "step": 8198
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027232447329291067,
+      "loss": 2.9224,
+      "step": 8199
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027220823978096,
+      "loss": 2.7004,
+      "step": 8200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027209202180125675,
+      "loss": 2.4892,
+      "step": 8201
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027197581936172537,
+      "loss": 3.0355,
+      "step": 8202
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002718596324702891,
+      "loss": 2.83,
+      "step": 8203
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002717434611348708,
+      "loss": 2.9153,
+      "step": 8204
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027162730536339097,
+      "loss": 3.0688,
+      "step": 8205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002715111651637707,
+      "loss": 3.1008,
+      "step": 8206
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027139504054392825,
+      "loss": 3.0153,
+      "step": 8207
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002712789315117824,
+      "loss": 2.9219,
+      "step": 8208
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002711628380752501,
+      "loss": 2.7894,
+      "step": 8209
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027104676024224675,
+      "loss": 2.8285,
+      "step": 8210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000270930698020688,
+      "loss": 2.8868,
+      "step": 8211
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027081465141848705,
+      "loss": 3.1689,
+      "step": 8212
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002706986204435573,
+      "loss": 2.6825,
+      "step": 8213
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027058260510380984,
+      "loss": 2.8799,
+      "step": 8214
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002704666054071557,
+      "loss": 2.9409,
+      "step": 8215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027035062136150433,
+      "loss": 3.2569,
+      "step": 8216
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002702346529747642,
+      "loss": 2.783,
+      "step": 8217
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00027011870025484295,
+      "loss": 2.836,
+      "step": 8218
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002700027632096468,
+      "loss": 2.6966,
+      "step": 8219
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026988684184708103,
+      "loss": 3.0328,
+      "step": 8220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002697709361750499,
+      "loss": 2.9709,
+      "step": 8221
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026965504620145655,
+      "loss": 2.7702,
+      "step": 8222
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026953917193420316,
+      "loss": 3.1164,
+      "step": 8223
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026942331338119044,
+      "loss": 3.0228,
+      "step": 8224
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000269307470550319,
+      "loss": 3.0113,
+      "step": 8225
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000269191643449487,
+      "loss": 2.9228,
+      "step": 8226
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026907583208659285,
+      "loss": 2.9818,
+      "step": 8227
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002689600364695327,
+      "loss": 2.7941,
+      "step": 8228
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026884425660620283,
+      "loss": 2.5998,
+      "step": 8229
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002687284925044972,
+      "loss": 2.8255,
+      "step": 8230
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002686127441723097,
+      "loss": 2.8241,
+      "step": 8231
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002684970116175327,
+      "loss": 3.3144,
+      "step": 8232
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026838129484805764,
+      "loss": 2.9706,
+      "step": 8233
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002682655938717746,
+      "loss": 2.956,
+      "step": 8234
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.000268149908696573,
+      "loss": 2.9725,
+      "step": 8235
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026803423933034076,
+      "loss": 2.9498,
+      "step": 8236
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026791858578096504,
+      "loss": 2.9351,
+      "step": 8237
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002678029480563319,
+      "loss": 2.8953,
+      "step": 8238
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002676873261643261,
+      "loss": 2.9924,
+      "step": 8239
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026757172011283127,
+      "loss": 2.9402,
+      "step": 8240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026745612990973077,
+      "loss": 3.031,
+      "step": 8241
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026734055556290566,
+      "loss": 2.9542,
+      "step": 8242
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026722499708023667,
+      "loss": 2.8787,
+      "step": 8243
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002671094544696034,
+      "loss": 2.9671,
+      "step": 8244
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026699392773888407,
+      "loss": 3.0349,
+      "step": 8245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.00026687841689595625,
+      "loss": 2.8869,
+      "step": 8246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002667629219486958,
+      "loss": 2.8577,
+      "step": 8247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 0.0002666474429049785,
+      "loss": 3.1039,
+      "step": 8248
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002665319797726776,
+      "loss": 2.8118,
+      "step": 8249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000266416532559667,
+      "loss": 2.8164,
+      "step": 8250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026630110127381777,
+      "loss": 2.9926,
+      "step": 8251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002661856859230013,
+      "loss": 2.9136,
+      "step": 8252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026607028651508713,
+      "loss": 2.9781,
+      "step": 8253
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026595490305794385,
+      "loss": 3.1209,
+      "step": 8254
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026583953555943906,
+      "loss": 2.8159,
+      "step": 8255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026572418402743923,
+      "loss": 2.9168,
+      "step": 8256
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002656088484698097,
+      "loss": 3.2106,
+      "step": 8257
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026549352889441486,
+      "loss": 2.7375,
+      "step": 8258
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002653782253091178,
+      "loss": 2.6868,
+      "step": 8259
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026526293772178065,
+      "loss": 2.9594,
+      "step": 8260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002651476661402645,
+      "loss": 2.9503,
+      "step": 8261
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002650324105724293,
+      "loss": 3.0343,
+      "step": 8262
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026491717102613356,
+      "loss": 2.8351,
+      "step": 8263
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002648019475092356,
+      "loss": 2.808,
+      "step": 8264
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026468674002959146,
+      "loss": 3.1058,
+      "step": 8265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002645715485950574,
+      "loss": 3.0987,
+      "step": 8266
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026445637321348696,
+      "loss": 3.0653,
+      "step": 8267
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002643412138927342,
+      "loss": 2.975,
+      "step": 8268
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002642260706406514,
+      "loss": 2.7234,
+      "step": 8269
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002641109434650894,
+      "loss": 2.701,
+      "step": 8270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002639958323738985,
+      "loss": 2.8855,
+      "step": 8271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002638807373749276,
+      "loss": 2.8964,
+      "step": 8272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002637656584760245,
+      "loss": 3.0217,
+      "step": 8273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026365059568503617,
+      "loss": 2.8737,
+      "step": 8274
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002635355490098082,
+      "loss": 2.851,
+      "step": 8275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002634205184581853,
+      "loss": 2.6789,
+      "step": 8276
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026330550403801074,
+      "loss": 3.1657,
+      "step": 8277
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002631905057571271,
+      "loss": 2.9224,
+      "step": 8278
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026307552362337555,
+      "loss": 2.8313,
+      "step": 8279
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026296055764459637,
+      "loss": 2.9065,
+      "step": 8280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002628456078286287,
+      "loss": 2.8115,
+      "step": 8281
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026273067418331033,
+      "loss": 3.0154,
+      "step": 8282
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026261575671647833,
+      "loss": 2.7345,
+      "step": 8283
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026250085543596825,
+      "loss": 2.7664,
+      "step": 8284
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002623859703496152,
+      "loss": 2.8713,
+      "step": 8285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002622711014652522,
+      "loss": 2.8265,
+      "step": 8286
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002621562487907124,
+      "loss": 2.8536,
+      "step": 8287
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026204141233382634,
+      "loss": 2.7387,
+      "step": 8288
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002619265921024251,
+      "loss": 3.0019,
+      "step": 8289
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026181178810433703,
+      "loss": 2.8684,
+      "step": 8290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026169700034739087,
+      "loss": 2.8777,
+      "step": 8291
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002615822288394132,
+      "loss": 2.9356,
+      "step": 8292
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002614674735882299,
+      "loss": 2.9736,
+      "step": 8293
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026135273460166575,
+      "loss": 2.9986,
+      "step": 8294
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026123801188754434,
+      "loss": 2.802,
+      "step": 8295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002611233054536881,
+      "loss": 3.1206,
+      "step": 8296
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002610086153079185,
+      "loss": 3.0497,
+      "step": 8297
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002608939414580559,
+      "loss": 3.0213,
+      "step": 8298
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002607792839119193,
+      "loss": 3.1181,
+      "step": 8299
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002606646426773266,
+      "loss": 2.9193,
+      "step": 8300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026055001776209536,
+      "loss": 3.0148,
+      "step": 8301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002604354091740407,
+      "loss": 3.2203,
+      "step": 8302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000260320816920978,
+      "loss": 2.7955,
+      "step": 8303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002602062410107202,
+      "loss": 2.6798,
+      "step": 8304
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00026009168145108046,
+      "loss": 3.1073,
+      "step": 8305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025997713824986945,
+      "loss": 2.6847,
+      "step": 8306
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025986261141489807,
+      "loss": 2.7898,
+      "step": 8307
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002597481009539754,
+      "loss": 3.0957,
+      "step": 8308
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002596336068749088,
+      "loss": 3.0792,
+      "step": 8309
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025951912918550614,
+      "loss": 2.6917,
+      "step": 8310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025940466789357223,
+      "loss": 2.8764,
+      "step": 8311
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025929022300691254,
+      "loss": 2.9647,
+      "step": 8312
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002591757945333303,
+      "loss": 3.1007,
+      "step": 8313
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000259061382480628,
+      "loss": 2.9827,
+      "step": 8314
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025894698685660693,
+      "loss": 2.9927,
+      "step": 8315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025883260766906737,
+      "loss": 2.9449,
+      "step": 8316
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002587182449258082,
+      "loss": 3.0947,
+      "step": 8317
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025860389863462764,
+      "loss": 2.6662,
+      "step": 8318
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002584895688033223,
+      "loss": 3.0824,
+      "step": 8319
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002583752554396881,
+      "loss": 2.9378,
+      "step": 8320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002582609585515194,
+      "loss": 2.78,
+      "step": 8321
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025814667814660984,
+      "loss": 3.0379,
+      "step": 8322
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002580324142327516,
+      "loss": 2.9734,
+      "step": 8323
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025791816681773627,
+      "loss": 3.0765,
+      "step": 8324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025780393590935335,
+      "loss": 3.2076,
+      "step": 8325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002576897215153926,
+      "loss": 3.1283,
+      "step": 8326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002575755236436409,
+      "loss": 3.0489,
+      "step": 8327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025746134230188574,
+      "loss": 3.1818,
+      "step": 8328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025734717749791237,
+      "loss": 2.7434,
+      "step": 8329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025723302923950534,
+      "loss": 2.7103,
+      "step": 8330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.000257118897534448,
+      "loss": 2.8107,
+      "step": 8331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002570047823905225,
+      "loss": 2.7976,
+      "step": 8332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002568906838155099,
+      "loss": 2.7426,
+      "step": 8333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002567766018171902,
+      "loss": 2.8647,
+      "step": 8334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025666253640334216,
+      "loss": 2.8537,
+      "step": 8335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002565484875817435,
+      "loss": 3.0362,
+      "step": 8336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025643445536017073,
+      "loss": 2.8113,
+      "step": 8337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025632043974639933,
+      "loss": 2.7472,
+      "step": 8338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025620644074820343,
+      "loss": 2.9473,
+      "step": 8339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002560924583733566,
+      "loss": 3.0265,
+      "step": 8340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025597849262963047,
+      "loss": 2.9216,
+      "step": 8341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025586454352479593,
+      "loss": 2.6577,
+      "step": 8342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025575061106662293,
+      "loss": 2.847,
+      "step": 8343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002556366952628798,
+      "loss": 2.9878,
+      "step": 8344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025552279612133454,
+      "loss": 2.8071,
+      "step": 8345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025540891364975285,
+      "loss": 3.1175,
+      "step": 8346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002552950478559006,
+      "loss": 2.8795,
+      "step": 8347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002551811987475412,
+      "loss": 2.9397,
+      "step": 8348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002550673663324382,
+      "loss": 2.8804,
+      "step": 8349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002549535506183528,
+      "loss": 2.9232,
+      "step": 8350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002548397516130462,
+      "loss": 2.6958,
+      "step": 8351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002547259693242776,
+      "loss": 2.9385,
+      "step": 8352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025461220375980545,
+      "loss": 2.8446,
+      "step": 8353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025449845492738697,
+      "loss": 2.9289,
+      "step": 8354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025438472283477844,
+      "loss": 2.9911,
+      "step": 8355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002542710074897345,
+      "loss": 2.9788,
+      "step": 8356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002541573089000092,
+      "loss": 2.8907,
+      "step": 8357
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002540436270733552,
+      "loss": 3.0322,
+      "step": 8358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002539299620175238,
+      "loss": 3.1723,
+      "step": 8359
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025381631374026545,
+      "loss": 2.8727,
+      "step": 8360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002537026822493299,
+      "loss": 3.1899,
+      "step": 8361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025358906755246436,
+      "loss": 2.7365,
+      "step": 8362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025347546965741677,
+      "loss": 2.8364,
+      "step": 8363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002533618885719319,
+      "loss": 2.893,
+      "step": 8364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002532483243037554,
+      "loss": 2.8308,
+      "step": 8365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025313477686062994,
+      "loss": 2.8963,
+      "step": 8366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025302124625029845,
+      "loss": 2.9983,
+      "step": 8367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025290773248050184,
+      "loss": 2.9904,
+      "step": 8368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002527942355589804,
+      "loss": 2.9718,
+      "step": 8369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025268075549347293,
+      "loss": 2.8967,
+      "step": 8370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.00025256729229171723,
+      "loss": 2.7845,
+      "step": 8371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 0.0002524538459614499,
+      "loss": 2.9907,
+      "step": 8372
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002523404165104064,
+      "loss": 2.8427,
+      "step": 8373
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000252227003946321,
+      "loss": 2.7756,
+      "step": 8374
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002521136082769271,
+      "loss": 2.8846,
+      "step": 8375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002520002295099564,
+      "loss": 2.9254,
+      "step": 8376
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025188686765314,
+      "loss": 3.0495,
+      "step": 8377
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002517735227142075,
+      "loss": 2.8078,
+      "step": 8378
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002516601947008875,
+      "loss": 2.6794,
+      "step": 8379
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025154688362090735,
+      "loss": 2.8249,
+      "step": 8380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025143358948199335,
+      "loss": 2.8225,
+      "step": 8381
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025132031229187056,
+      "loss": 2.9045,
+      "step": 8382
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002512070520582628,
+      "loss": 2.9123,
+      "step": 8383
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002510938087888933,
+      "loss": 3.011,
+      "step": 8384
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025098058249148304,
+      "loss": 2.9178,
+      "step": 8385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002508673731737532,
+      "loss": 2.9784,
+      "step": 8386
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002507541808434223,
+      "loss": 2.7096,
+      "step": 8387
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025064100550820914,
+      "loss": 3.0845,
+      "step": 8388
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00025052784717583036,
+      "loss": 2.8521,
+      "step": 8389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000250414705854002,
+      "loss": 3.1805,
+      "step": 8390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002503015815504386,
+      "loss": 2.9538,
+      "step": 8391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002501884742728537,
+      "loss": 2.9033,
+      "step": 8392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002500753840289597,
+      "loss": 3.0527,
+      "step": 8393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002499623108264678,
+      "loss": 2.9436,
+      "step": 8394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024984925467308797,
+      "loss": 2.7333,
+      "step": 8395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000249736215576529,
+      "loss": 2.9208,
+      "step": 8396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024962319354449877,
+      "loss": 2.9285,
+      "step": 8397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024951018858470367,
+      "loss": 2.8299,
+      "step": 8398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024939720070484895,
+      "loss": 2.9924,
+      "step": 8399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002492842299126394,
+      "loss": 2.7672,
+      "step": 8400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024917127621577726,
+      "loss": 2.5317,
+      "step": 8401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002490583396219653,
+      "loss": 2.8808,
+      "step": 8402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002489454201389033,
+      "loss": 3.1173,
+      "step": 8403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002488325177742914,
+      "loss": 2.9419,
+      "step": 8404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002487196325358282,
+      "loss": 2.9988,
+      "step": 8405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002486067644312101,
+      "loss": 2.84,
+      "step": 8406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002484939134681339,
+      "loss": 2.937,
+      "step": 8407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024838107965429384,
+      "loss": 3.1569,
+      "step": 8408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024826826299738434,
+      "loss": 2.9892,
+      "step": 8409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024815546350509716,
+      "loss": 2.9012,
+      "step": 8410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002480426811851243,
+      "loss": 3.0559,
+      "step": 8411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024792991604515567,
+      "loss": 2.986,
+      "step": 8412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024781716809288033,
+      "loss": 2.9061,
+      "step": 8413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002477044373359862,
+      "loss": 2.913,
+      "step": 8414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002475917237821598,
+      "loss": 2.8928,
+      "step": 8415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002474790274390868,
+      "loss": 2.8194,
+      "step": 8416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024736634831445145,
+      "loss": 2.9945,
+      "step": 8417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024725368641593685,
+      "loss": 2.8765,
+      "step": 8418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024714104175122513,
+      "loss": 2.9567,
+      "step": 8419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002470284143279968,
+      "loss": 2.9757,
+      "step": 8420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024691580415393214,
+      "loss": 2.9005,
+      "step": 8421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002468032112367088,
+      "loss": 2.8838,
+      "step": 8422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002466906355840048,
+      "loss": 2.7942,
+      "step": 8423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002465780772034955,
+      "loss": 2.8991,
+      "step": 8424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002464655361028566,
+      "loss": 3.0522,
+      "step": 8425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024635301228976105,
+      "loss": 2.8312,
+      "step": 8426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000246240505771882,
+      "loss": 2.802,
+      "step": 8427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002461280165568907,
+      "loss": 3.1605,
+      "step": 8428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002460155446524573,
+      "loss": 2.9596,
+      "step": 8429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024590309006625085,
+      "loss": 3.0113,
+      "step": 8430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002457906528059392,
+      "loss": 2.8786,
+      "step": 8431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.000245678232879189,
+      "loss": 2.9372,
+      "step": 8432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024556583029366574,
+      "loss": 2.9779,
+      "step": 8433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024545344505703373,
+      "loss": 3.0106,
+      "step": 8434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024534107717695606,
+      "loss": 2.7455,
+      "step": 8435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002452287266610946,
+      "loss": 2.8002,
+      "step": 8436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002451163935171105,
+      "loss": 2.8767,
+      "step": 8437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002450040777526628,
+      "loss": 3.0384,
+      "step": 8438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024489177937541014,
+      "loss": 2.995,
+      "step": 8439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002447794983930097,
+      "loss": 3.0403,
+      "step": 8440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024466723481311736,
+      "loss": 2.8679,
+      "step": 8441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024455498864338815,
+      "loss": 2.9834,
+      "step": 8442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002444427598914754,
+      "loss": 2.994,
+      "step": 8443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002443305485650321,
+      "loss": 2.9829,
+      "step": 8444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002442183546717088,
+      "loss": 2.8988,
+      "step": 8445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024410617821915638,
+      "loss": 3.132,
+      "step": 8446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002439940192150229,
+      "loss": 2.8177,
+      "step": 8447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024388187766695663,
+      "loss": 2.7725,
+      "step": 8448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024376975358260383,
+      "loss": 3.0526,
+      "step": 8449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002436576469696099,
+      "loss": 2.7396,
+      "step": 8450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024354555783561894,
+      "loss": 2.5828,
+      "step": 8451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024343348618827387,
+      "loss": 2.9463,
+      "step": 8452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002433214320352164,
+      "loss": 2.9053,
+      "step": 8453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002432093953840871,
+      "loss": 2.8885,
+      "step": 8454
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024309737624252531,
+      "loss": 2.9893,
+      "step": 8455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024298537461816923,
+      "loss": 3.0654,
+      "step": 8456
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024287339051865575,
+      "loss": 3.0632,
+      "step": 8457
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024276142395162066,
+      "loss": 2.922,
+      "step": 8458
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024264947492469836,
+      "loss": 2.7082,
+      "step": 8459
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024253754344552282,
+      "loss": 3.1033,
+      "step": 8460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024242562952172535,
+      "loss": 3.117,
+      "step": 8461
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024231373316093785,
+      "loss": 2.9874,
+      "step": 8462
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024220185437078917,
+      "loss": 2.8377,
+      "step": 8463
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024208999315890862,
+      "loss": 2.7443,
+      "step": 8464
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024197814953292335,
+      "loss": 2.9246,
+      "step": 8465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002418663235004595,
+      "loss": 2.9112,
+      "step": 8466
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002417545150691421,
+      "loss": 3.1917,
+      "step": 8467
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024164272424659488,
+      "loss": 3.0441,
+      "step": 8468
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024153095104044055,
+      "loss": 2.9268,
+      "step": 8469
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024141919545830038,
+      "loss": 2.9824,
+      "step": 8470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024130745750779466,
+      "loss": 2.803,
+      "step": 8471
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024119573719654227,
+      "loss": 2.9001,
+      "step": 8472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024108403453216098,
+      "loss": 2.8085,
+      "step": 8473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002409723495222675,
+      "loss": 3.0552,
+      "step": 8474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024086068217447703,
+      "loss": 2.8899,
+      "step": 8475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024074903249640396,
+      "loss": 2.8588,
+      "step": 8476
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024063740049566103,
+      "loss": 3.0568,
+      "step": 8477
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002405257861798601,
+      "loss": 2.8518,
+      "step": 8478
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024041418955661177,
+      "loss": 2.8585,
+      "step": 8479
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024030261063352504,
+      "loss": 2.8807,
+      "step": 8480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002401910494182088,
+      "loss": 2.7618,
+      "step": 8481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00024007950591826917,
+      "loss": 2.881,
+      "step": 8482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023996798014131254,
+      "loss": 2.8872,
+      "step": 8483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023985647209494277,
+      "loss": 2.8071,
+      "step": 8484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023974498178676385,
+      "loss": 2.9603,
+      "step": 8485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023963350922437716,
+      "loss": 2.8558,
+      "step": 8486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023952205441538416,
+      "loss": 2.9277,
+      "step": 8487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023941061736738433,
+      "loss": 3.1774,
+      "step": 8488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023929919808797617,
+      "loss": 3.005,
+      "step": 8489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023918779658475687,
+      "loss": 3.1808,
+      "step": 8490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002390764128653225,
+      "loss": 2.8901,
+      "step": 8491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002389650469372679,
+      "loss": 3.1424,
+      "step": 8492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023885369880818663,
+      "loss": 2.9118,
+      "step": 8493
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023874236848567121,
+      "loss": 2.9511,
+      "step": 8494
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.0002386310559773127,
+      "loss": 3.137,
+      "step": 8495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 0.00023851976129070097,
+      "loss": 3.247,
+      "step": 8496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023840848443342527,
+      "loss": 2.9081,
+      "step": 8497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023829722541307248,
+      "loss": 3.0901,
+      "step": 8498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023818598423722958,
+      "loss": 3.0014,
+      "step": 8499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023807476091348103,
+      "loss": 2.8031,
+      "step": 8500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002379635554494114,
+      "loss": 2.8715,
+      "step": 8501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023785236785260267,
+      "loss": 2.9844,
+      "step": 8502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023774119813063676,
+      "loss": 2.8838,
+      "step": 8503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023763004629109402,
+      "loss": 3.083,
+      "step": 8504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023751891234155288,
+      "loss": 2.9331,
+      "step": 8505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023740779628959185,
+      "loss": 3.0436,
+      "step": 8506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023729669814278676,
+      "loss": 3.1477,
+      "step": 8507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002371856179087135,
+      "loss": 3.0629,
+      "step": 8508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023707455559494618,
+      "loss": 3.011,
+      "step": 8509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002369635112090575,
+      "loss": 2.9082,
+      "step": 8510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023685248475861925,
+      "loss": 2.7987,
+      "step": 8511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023674147625120197,
+      "loss": 3.2066,
+      "step": 8512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023663048569437474,
+      "loss": 2.7619,
+      "step": 8513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023651951309570574,
+      "loss": 2.5567,
+      "step": 8514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023640855846276165,
+      "loss": 3.2125,
+      "step": 8515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023629762180310809,
+      "loss": 2.8072,
+      "step": 8516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023618670312430945,
+      "loss": 3.1134,
+      "step": 8517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002360758024339288,
+      "loss": 2.9011,
+      "step": 8518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023596491973952783,
+      "loss": 2.9543,
+      "step": 8519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002358540550486678,
+      "loss": 3.0393,
+      "step": 8520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023574320836890734,
+      "loss": 2.8171,
+      "step": 8521
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023563237970780548,
+      "loss": 2.9979,
+      "step": 8522
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002355215690729185,
+      "loss": 2.9936,
+      "step": 8523
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002354107764718026,
+      "loss": 3.0378,
+      "step": 8524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002353000019120121,
+      "loss": 2.9582,
+      "step": 8525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023518924540110042,
+      "loss": 2.6124,
+      "step": 8526
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023507850694661953,
+      "loss": 3.0873,
+      "step": 8527
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023496778655612033,
+      "loss": 2.9289,
+      "step": 8528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023485708423715235,
+      "loss": 3.005,
+      "step": 8529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023474639999726394,
+      "loss": 3.0779,
+      "step": 8530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023463573384400232,
+      "loss": 3.0386,
+      "step": 8531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023452508578491332,
+      "loss": 2.7079,
+      "step": 8532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023441445582754168,
+      "loss": 3.119,
+      "step": 8533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023430384397943077,
+      "loss": 2.9473,
+      "step": 8534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002341932502481226,
+      "loss": 3.2187,
+      "step": 8535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023408267464115873,
+      "loss": 2.9256,
+      "step": 8536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023397211716607831,
+      "loss": 3.0613,
+      "step": 8537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023386157783042,
+      "loss": 2.6932,
+      "step": 8538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023375105664172108,
+      "loss": 2.7897,
+      "step": 8539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002336405536075174,
+      "loss": 2.8569,
+      "step": 8540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002335300687353442,
+      "loss": 2.9306,
+      "step": 8541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023341960203273444,
+      "loss": 2.7718,
+      "step": 8542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023330915350722097,
+      "loss": 2.7325,
+      "step": 8543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023319872316633427,
+      "loss": 2.8754,
+      "step": 8544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023308831101760485,
+      "loss": 2.8883,
+      "step": 8545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023297791706856048,
+      "loss": 3.0232,
+      "step": 8546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023286754132672915,
+      "loss": 2.7588,
+      "step": 8547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023275718379963677,
+      "loss": 2.8474,
+      "step": 8548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023264684449480816,
+      "loss": 2.9169,
+      "step": 8549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023253652341976688,
+      "loss": 2.8873,
+      "step": 8550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023242622058203545,
+      "loss": 3.0739,
+      "step": 8551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023231593598913486,
+      "loss": 3.0559,
+      "step": 8552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002322056696485851,
+      "loss": 2.7105,
+      "step": 8553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002320954215679047,
+      "loss": 2.9921,
+      "step": 8554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002319851917546111,
+      "loss": 3.0472,
+      "step": 8555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023187498021622028,
+      "loss": 2.7719,
+      "step": 8556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023176478696024773,
+      "loss": 2.7505,
+      "step": 8557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002316546119942063,
+      "loss": 2.8102,
+      "step": 8558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023154445532560914,
+      "loss": 2.5489,
+      "step": 8559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023143431696196677,
+      "loss": 3.0375,
+      "step": 8560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002313241969107897,
+      "loss": 2.8867,
+      "step": 8561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023121409517958596,
+      "loss": 2.8984,
+      "step": 8562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023110401177586348,
+      "loss": 3.0163,
+      "step": 8563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023099394670712825,
+      "loss": 2.8624,
+      "step": 8564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023088389998088515,
+      "loss": 2.8869,
+      "step": 8565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023077387160463792,
+      "loss": 3.0211,
+      "step": 8566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023066386158588888,
+      "loss": 2.8116,
+      "step": 8567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023055386993213928,
+      "loss": 2.9798,
+      "step": 8568
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000230443896650889,
+      "loss": 2.8884,
+      "step": 8569
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023033394174963663,
+      "loss": 2.9726,
+      "step": 8570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023022400523587967,
+      "loss": 2.6121,
+      "step": 8571
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023011408711711424,
+      "loss": 3.0692,
+      "step": 8572
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00023000418740083518,
+      "loss": 2.7329,
+      "step": 8573
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002298943060945362,
+      "loss": 3.0562,
+      "step": 8574
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022978444320570968,
+      "loss": 2.8594,
+      "step": 8575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002296745987418467,
+      "loss": 2.7423,
+      "step": 8576
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002295647727104372,
+      "loss": 3.0541,
+      "step": 8577
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022945496511896974,
+      "loss": 2.9751,
+      "step": 8578
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022934517597493148,
+      "loss": 2.7129,
+      "step": 8579
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022923540528580917,
+      "loss": 2.8515,
+      "step": 8580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002291256530590868,
+      "loss": 2.8711,
+      "step": 8581
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002290159193022488,
+      "loss": 2.9981,
+      "step": 8582
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002289062040227766,
+      "loss": 2.8709,
+      "step": 8583
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022879650722815203,
+      "loss": 2.8829,
+      "step": 8584
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022868682892585458,
+      "loss": 2.7796,
+      "step": 8585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002285771691233628,
+      "loss": 2.7342,
+      "step": 8586
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022846752782815395,
+      "loss": 2.9673,
+      "step": 8587
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022835790504770415,
+      "loss": 2.7097,
+      "step": 8588
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022824830078948806,
+      "loss": 2.9432,
+      "step": 8589
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022813871506097917,
+      "loss": 2.9015,
+      "step": 8590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002280291478696498,
+      "loss": 2.9122,
+      "step": 8591
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022791959922297084,
+      "loss": 2.8017,
+      "step": 8592
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000227810069128412,
+      "loss": 2.9895,
+      "step": 8593
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022770055759344177,
+      "loss": 3.0123,
+      "step": 8594
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002275910646255271,
+      "loss": 3.0556,
+      "step": 8595
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022748159023213443,
+      "loss": 2.8853,
+      "step": 8596
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022737213442072769,
+      "loss": 3.0508,
+      "step": 8597
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022726269719877096,
+      "loss": 2.7657,
+      "step": 8598
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002271532785737257,
+      "loss": 2.9191,
+      "step": 8599
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022704387855305313,
+      "loss": 2.8182,
+      "step": 8600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022693449714421282,
+      "loss": 2.9416,
+      "step": 8601
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022682513435466294,
+      "loss": 2.8607,
+      "step": 8602
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022671579019186078,
+      "loss": 3.0999,
+      "step": 8603
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002266064646632615,
+      "loss": 3.1215,
+      "step": 8604
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022649715777632034,
+      "loss": 2.861,
+      "step": 8605
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022638786953848978,
+      "loss": 2.6243,
+      "step": 8606
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022627859995722233,
+      "loss": 2.6725,
+      "step": 8607
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022616934903996845,
+      "loss": 2.9031,
+      "step": 8608
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002260601167941776,
+      "loss": 2.7554,
+      "step": 8609
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022595090322729782,
+      "loss": 3.036,
+      "step": 8610
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022584170834677602,
+      "loss": 2.817,
+      "step": 8611
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002257325321600578,
+      "loss": 3.0852,
+      "step": 8612
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022562337467458743,
+      "loss": 3.1464,
+      "step": 8613
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022551423589780794,
+      "loss": 2.8958,
+      "step": 8614
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.0002254051158371611,
+      "loss": 2.9483,
+      "step": 8615
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022529601450008724,
+      "loss": 2.8078,
+      "step": 8616
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.000225186931894026,
+      "loss": 3.0426,
+      "step": 8617
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022507786802641473,
+      "loss": 2.8944,
+      "step": 8618
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022496882290469068,
+      "loss": 2.6129,
+      "step": 8619
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 0.00022485979653628853,
+      "loss": 2.9912,
+      "step": 8620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022475078892864316,
+      "loss": 3.0996,
+      "step": 8621
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002246418000891865,
+      "loss": 2.9636,
+      "step": 8622
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002245328300253508,
+      "loss": 2.8222,
+      "step": 8623
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022442387874456603,
+      "loss": 3.0145,
+      "step": 8624
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022431494625426118,
+      "loss": 2.8518,
+      "step": 8625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022420603256186395,
+      "loss": 3.1435,
+      "step": 8626
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002240971376748007,
+      "loss": 3.0017,
+      "step": 8627
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002239882616004967,
+      "loss": 2.9608,
+      "step": 8628
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002238794043463756,
+      "loss": 2.9358,
+      "step": 8629
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022377056591986005,
+      "loss": 2.9779,
+      "step": 8630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022366174632837134,
+      "loss": 3.0491,
+      "step": 8631
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022355294557932926,
+      "loss": 2.769,
+      "step": 8632
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002234441636801531,
+      "loss": 3.0335,
+      "step": 8633
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022333540063825945,
+      "loss": 3.1034,
+      "step": 8634
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022322665646106526,
+      "loss": 2.8379,
+      "step": 8635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022311793115598484,
+      "loss": 2.9024,
+      "step": 8636
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002230092247304319,
+      "loss": 2.778,
+      "step": 8637
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002229005371918187,
+      "loss": 2.8918,
+      "step": 8638
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022279186854755608,
+      "loss": 2.7907,
+      "step": 8639
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022268321880505426,
+      "loss": 2.7827,
+      "step": 8640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002225745879717209,
+      "loss": 2.8381,
+      "step": 8641
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022246597605496394,
+      "loss": 2.936,
+      "step": 8642
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002223573830621884,
+      "loss": 2.8682,
+      "step": 8643
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002222488090007994,
+      "loss": 2.997,
+      "step": 8644
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022214025387819997,
+      "loss": 3.0924,
+      "step": 8645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022203171770179215,
+      "loss": 2.8306,
+      "step": 8646
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022192320047897647,
+      "loss": 2.8277,
+      "step": 8647
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022181470221715243,
+      "loss": 2.9697,
+      "step": 8648
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022170622292371807,
+      "loss": 2.9349,
+      "step": 8649
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022159776260607017,
+      "loss": 2.9451,
+      "step": 8650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022148932127160422,
+      "loss": 2.8068,
+      "step": 8651
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002213808989277144,
+      "loss": 2.9942,
+      "step": 8652
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022127249558179368,
+      "loss": 2.8468,
+      "step": 8653
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022116411124123358,
+      "loss": 3.0214,
+      "step": 8654
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022105574591342432,
+      "loss": 2.9914,
+      "step": 8655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022094739960575545,
+      "loss": 2.8646,
+      "step": 8656
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022083907232561389,
+      "loss": 3.2275,
+      "step": 8657
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022073076408038683,
+      "loss": 2.9218,
+      "step": 8658
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002206224748774587,
+      "loss": 2.9003,
+      "step": 8659
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002205142047242138,
+      "loss": 2.93,
+      "step": 8660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022040595362803452,
+      "loss": 3.0859,
+      "step": 8661
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022029772159630213,
+      "loss": 2.7969,
+      "step": 8662
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00022018950863639649,
+      "loss": 2.9362,
+      "step": 8663
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002200813147556962,
+      "loss": 2.8486,
+      "step": 8664
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002199731399615787,
+      "loss": 2.9984,
+      "step": 8665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021986498426141988,
+      "loss": 2.7508,
+      "step": 8666
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021975684766259456,
+      "loss": 3.045,
+      "step": 8667
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021964873017247616,
+      "loss": 2.9962,
+      "step": 8668
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002195406317984368,
+      "loss": 2.8974,
+      "step": 8669
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021943255254784722,
+      "loss": 2.6604,
+      "step": 8670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021932449242807696,
+      "loss": 2.8801,
+      "step": 8671
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021921645144649428,
+      "loss": 2.8709,
+      "step": 8672
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021910842961046602,
+      "loss": 2.9977,
+      "step": 8673
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002190004269273578,
+      "loss": 3.1584,
+      "step": 8674
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021889244340453384,
+      "loss": 2.7635,
+      "step": 8675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021878447904935712,
+      "loss": 2.9343,
+      "step": 8676
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021867653386918967,
+      "loss": 2.8525,
+      "step": 8677
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002185686078713912,
+      "loss": 3.1156,
+      "step": 8678
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021846070106332156,
+      "loss": 3.106,
+      "step": 8679
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021835281345233771,
+      "loss": 2.8041,
+      "step": 8680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021824494504579678,
+      "loss": 2.8291,
+      "step": 8681
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021813709585105328,
+      "loss": 2.8805,
+      "step": 8682
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021802926587546152,
+      "loss": 2.7921,
+      "step": 8683
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021792145512637378,
+      "loss": 2.8473,
+      "step": 8684
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002178136636111414,
+      "loss": 2.8417,
+      "step": 8685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021770589133711415,
+      "loss": 3.0018,
+      "step": 8686
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021759813831164065,
+      "loss": 2.9274,
+      "step": 8687
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021749040454206814,
+      "loss": 3.1211,
+      "step": 8688
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021738269003574262,
+      "loss": 2.8007,
+      "step": 8689
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021727499480000872,
+      "loss": 2.8375,
+      "step": 8690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002171673188422097,
+      "loss": 3.1405,
+      "step": 8691
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002170596621696875,
+      "loss": 3.0926,
+      "step": 8692
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021695202478978326,
+      "loss": 2.843,
+      "step": 8693
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021684440670983568,
+      "loss": 2.9628,
+      "step": 8694
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021673680793718358,
+      "loss": 2.9722,
+      "step": 8695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002166292284791629,
+      "loss": 2.8538,
+      "step": 8696
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021652166834310987,
+      "loss": 2.938,
+      "step": 8697
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021641412753635786,
+      "loss": 2.8579,
+      "step": 8698
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002163066060662402,
+      "loss": 2.764,
+      "step": 8699
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021619910394008834,
+      "loss": 2.9834,
+      "step": 8700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021609162116523195,
+      "loss": 2.8186,
+      "step": 8701
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002159841577490006,
+      "loss": 2.9534,
+      "step": 8702
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021587671369872102,
+      "loss": 3.0786,
+      "step": 8703
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021576928902171998,
+      "loss": 2.8797,
+      "step": 8704
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002156618837253222,
+      "loss": 3.0681,
+      "step": 8705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021555449781685115,
+      "loss": 3.016,
+      "step": 8706
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002154471313036292,
+      "loss": 2.9109,
+      "step": 8707
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021533978419297718,
+      "loss": 2.842,
+      "step": 8708
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021523245649221468,
+      "loss": 2.9928,
+      "step": 8709
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021512514820865998,
+      "loss": 2.8069,
+      "step": 8710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021501785934962997,
+      "loss": 2.9516,
+      "step": 8711
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021491058992244034,
+      "loss": 2.7595,
+      "step": 8712
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021480333993440537,
+      "loss": 3.1422,
+      "step": 8713
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.000214696109392838,
+      "loss": 3.0893,
+      "step": 8714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002145888983050497,
+      "loss": 2.5912,
+      "step": 8715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021448170667835126,
+      "loss": 2.9779,
+      "step": 8716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021437453452005106,
+      "loss": 3.1061,
+      "step": 8717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021426738183745743,
+      "loss": 3.0362,
+      "step": 8718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021416024863787596,
+      "loss": 2.8578,
+      "step": 8719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021405313492861222,
+      "loss": 2.7331,
+      "step": 8720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021394604071696971,
+      "loss": 2.9636,
+      "step": 8721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002138389660102507,
+      "loss": 2.6531,
+      "step": 8722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021373191081575633,
+      "loss": 2.9836,
+      "step": 8723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021362487514078616,
+      "loss": 3.0478,
+      "step": 8724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021351785899263858,
+      "loss": 2.8877,
+      "step": 8725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002134108623786107,
+      "loss": 2.9914,
+      "step": 8726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021330388530599808,
+      "loss": 2.9432,
+      "step": 8727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021319692778209515,
+      "loss": 3.0945,
+      "step": 8728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021308998981419493,
+      "loss": 3.021,
+      "step": 8729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002129830714095891,
+      "loss": 2.8159,
+      "step": 8730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002128761725755678,
+      "loss": 3.0778,
+      "step": 8731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021276929331942064,
+      "loss": 2.8423,
+      "step": 8732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002126624336484348,
+      "loss": 3.2146,
+      "step": 8733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021255559356989667,
+      "loss": 3.1068,
+      "step": 8734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021244877309109145,
+      "loss": 2.8122,
+      "step": 8735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021234197221930252,
+      "loss": 3.0316,
+      "step": 8736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021223519096181276,
+      "loss": 2.7238,
+      "step": 8737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021212842932590248,
+      "loss": 2.9033,
+      "step": 8738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021202168731885212,
+      "loss": 3.0684,
+      "step": 8739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.0002119149649479392,
+      "loss": 2.9345,
+      "step": 8740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021180826222044153,
+      "loss": 3.064,
+      "step": 8741
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021170157914363397,
+      "loss": 2.8666,
+      "step": 8742
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021159491572479133,
+      "loss": 2.8103,
+      "step": 8743
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 0.00021148827197118648,
+      "loss": 2.9976,
+      "step": 8744
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000211381647890091,
+      "loss": 2.7563,
+      "step": 8745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021127504348877524,
+      "loss": 3.0364,
+      "step": 8746
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021116845877450807,
+      "loss": 2.9475,
+      "step": 8747
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002110618937545572,
+      "loss": 2.887,
+      "step": 8748
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021095534843618873,
+      "loss": 2.5588,
+      "step": 8749
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021084882282666767,
+      "loss": 2.8528,
+      "step": 8750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021074231693325762,
+      "loss": 2.8975,
+      "step": 8751
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021063583076322062,
+      "loss": 3.0076,
+      "step": 8752
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021052936432381804,
+      "loss": 3.2745,
+      "step": 8753
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002104229176223087,
+      "loss": 2.8295,
+      "step": 8754
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002103164906659516,
+      "loss": 2.8707,
+      "step": 8755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00021021008346200275,
+      "loss": 3.2877,
+      "step": 8756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002101036960177185,
+      "loss": 2.9859,
+      "step": 8757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020999732834035218,
+      "loss": 2.9328,
+      "step": 8758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020989098043715715,
+      "loss": 2.6283,
+      "step": 8759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020978465231538467,
+      "loss": 3.0264,
+      "step": 8760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002096783439822849,
+      "loss": 3.0738,
+      "step": 8761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020957205544510661,
+      "loss": 2.9575,
+      "step": 8762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020946578671109716,
+      "loss": 2.9743,
+      "step": 8763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002093595377875026,
+      "loss": 2.8053,
+      "step": 8764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020925330868156773,
+      "loss": 2.793,
+      "step": 8765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020914709940053573,
+      "loss": 3.0524,
+      "step": 8766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020904090995164877,
+      "loss": 2.7838,
+      "step": 8767
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020893474034214743,
+      "loss": 2.9048,
+      "step": 8768
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000208828590579271,
+      "loss": 2.885,
+      "step": 8769
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020872246067025753,
+      "loss": 2.6879,
+      "step": 8770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002086163506223434,
+      "loss": 2.8771,
+      "step": 8771
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020851026044276406,
+      "loss": 2.9442,
+      "step": 8772
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020840419013875327,
+      "loss": 2.8726,
+      "step": 8773
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020829813971754362,
+      "loss": 3.1284,
+      "step": 8774
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020819210918636606,
+      "loss": 2.8567,
+      "step": 8775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020808609855245096,
+      "loss": 2.8421,
+      "step": 8776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020798010782302606,
+      "loss": 2.9974,
+      "step": 8777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002078741370053192,
+      "loss": 2.9785,
+      "step": 8778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002077681861065554,
+      "loss": 3.1113,
+      "step": 8779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002076622551339596,
+      "loss": 2.9438,
+      "step": 8780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020755634409475455,
+      "loss": 2.8972,
+      "step": 8781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020745045299616199,
+      "loss": 2.9891,
+      "step": 8782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020734458184540227,
+      "loss": 2.7378,
+      "step": 8783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020723873064969428,
+      "loss": 3.0719,
+      "step": 8784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020713289941625562,
+      "loss": 2.7143,
+      "step": 8785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020702708815230252,
+      "loss": 3.0115,
+      "step": 8786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020692129686504984,
+      "loss": 2.8227,
+      "step": 8787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020681552556171107,
+      "loss": 3.3104,
+      "step": 8788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020670977424949838,
+      "loss": 2.9801,
+      "step": 8789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020660404293562258,
+      "loss": 2.9281,
+      "step": 8790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020649833162729282,
+      "loss": 2.9663,
+      "step": 8791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020639264033171767,
+      "loss": 2.7473,
+      "step": 8792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020628696905610323,
+      "loss": 2.8002,
+      "step": 8793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020618131780765543,
+      "loss": 2.9913,
+      "step": 8794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002060756865935775,
+      "loss": 2.9581,
+      "step": 8795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020597007542107266,
+      "loss": 2.724,
+      "step": 8796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002058644842973419,
+      "loss": 2.9478,
+      "step": 8797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000205758913229585,
+      "loss": 2.7417,
+      "step": 8798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020565336222500076,
+      "loss": 2.8708,
+      "step": 8799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020554783129078563,
+      "loss": 2.9031,
+      "step": 8800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002054423204341362,
+      "loss": 2.9284,
+      "step": 8801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000205336829662246,
+      "loss": 2.7608,
+      "step": 8802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020523135898230872,
+      "loss": 2.9417,
+      "step": 8803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020512590840151575,
+      "loss": 2.8222,
+      "step": 8804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002050204779270573,
+      "loss": 2.856,
+      "step": 8805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002049150675661224,
+      "loss": 2.8301,
+      "step": 8806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002048096773258985,
+      "loss": 3.0676,
+      "step": 8807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020470430721357174,
+      "loss": 3.1275,
+      "step": 8808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000204598957236327,
+      "loss": 2.7745,
+      "step": 8809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020449362740134757,
+      "loss": 2.942,
+      "step": 8810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002043883177158156,
+      "loss": 2.7408,
+      "step": 8811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020428302818691153,
+      "loss": 2.9091,
+      "step": 8812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002041777588218151,
+      "loss": 2.8997,
+      "step": 8813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002040725096277037,
+      "loss": 3.1974,
+      "step": 8814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020396728061175434,
+      "loss": 3.0284,
+      "step": 8815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020386207178114168,
+      "loss": 2.8861,
+      "step": 8816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020375688314304015,
+      "loss": 3.0092,
+      "step": 8817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002036517147046214,
+      "loss": 2.9516,
+      "step": 8818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020354656647305703,
+      "loss": 2.7001,
+      "step": 8819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020344143845551656,
+      "loss": 2.7879,
+      "step": 8820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020333633065916818,
+      "loss": 2.9765,
+      "step": 8821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020323124309117892,
+      "loss": 2.8438,
+      "step": 8822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020312617575871412,
+      "loss": 3.1208,
+      "step": 8823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020302112866893806,
+      "loss": 2.8542,
+      "step": 8824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002029161018290135,
+      "loss": 2.776,
+      "step": 8825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002028110952461017,
+      "loss": 3.0285,
+      "step": 8826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020270610892736274,
+      "loss": 3.0001,
+      "step": 8827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020260114287995508,
+      "loss": 2.9198,
+      "step": 8828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002024961971110364,
+      "loss": 2.9996,
+      "step": 8829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020239127162776188,
+      "loss": 2.8997,
+      "step": 8830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020228636643728677,
+      "loss": 3.0184,
+      "step": 8831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020218148154676352,
+      "loss": 2.8119,
+      "step": 8832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020207661696334406,
+      "loss": 3.028,
+      "step": 8833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020197177269417872,
+      "loss": 3.0646,
+      "step": 8834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020186694874641615,
+      "loss": 3.1888,
+      "step": 8835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020176214512720452,
+      "loss": 2.9249,
+      "step": 8836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020165736184368926,
+      "loss": 2.9856,
+      "step": 8837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020155259890301586,
+      "loss": 3.0934,
+      "step": 8838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020144785631232693,
+      "loss": 2.9981,
+      "step": 8839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020134313407876509,
+      "loss": 2.8278,
+      "step": 8840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002012384322094707,
+      "loss": 2.775,
+      "step": 8841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020113375071158303,
+      "loss": 2.8118,
+      "step": 8842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020102908959223988,
+      "loss": 3.116,
+      "step": 8843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002009244488585777,
+      "loss": 3.0425,
+      "step": 8844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002008198285177315,
+      "loss": 2.8802,
+      "step": 8845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.000200715228576835,
+      "loss": 2.8713,
+      "step": 8846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020061064904302047,
+      "loss": 3.0449,
+      "step": 8847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020050608992341874,
+      "loss": 3.0195,
+      "step": 8848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002004015512251593,
+      "loss": 2.929,
+      "step": 8849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020029703295537028,
+      "loss": 2.9085,
+      "step": 8850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0002001925351211782,
+      "loss": 2.8247,
+      "step": 8851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00020008805772970883,
+      "loss": 3.0906,
+      "step": 8852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019998360078808547,
+      "loss": 2.88,
+      "step": 8853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019987916430343135,
+      "loss": 2.904,
+      "step": 8854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001997747482828668,
+      "loss": 2.8929,
+      "step": 8855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001996703527335121,
+      "loss": 2.8395,
+      "step": 8856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019956597766248546,
+      "loss": 2.8613,
+      "step": 8857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001994616230769038,
+      "loss": 2.7667,
+      "step": 8858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001993572889838826,
+      "loss": 2.7858,
+      "step": 8859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019925297539053605,
+      "loss": 3.1631,
+      "step": 8860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019914868230397692,
+      "loss": 2.6605,
+      "step": 8861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001990444097313165,
+      "loss": 3.0093,
+      "step": 8862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019894015767966472,
+      "loss": 2.7136,
+      "step": 8863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001988359261561302,
+      "loss": 2.8133,
+      "step": 8864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019873171516782012,
+      "loss": 2.7804,
+      "step": 8865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019862752472184008,
+      "loss": 2.9818,
+      "step": 8866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.0001985233548252946,
+      "loss": 2.8213,
+      "step": 8867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 0.00019841920548528647,
+      "loss": 2.6973,
+      "step": 8868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019831507670891736,
+      "loss": 3.0089,
+      "step": 8869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001982109685032873,
+      "loss": 3.0641,
+      "step": 8870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001981068808754951,
+      "loss": 2.7552,
+      "step": 8871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019800281383263796,
+      "loss": 2.8866,
+      "step": 8872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019789876738181224,
+      "loss": 2.9324,
+      "step": 8873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019779474153011178,
+      "loss": 2.838,
+      "step": 8874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019769073628463053,
+      "loss": 3.1067,
+      "step": 8875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019758675165245932,
+      "loss": 3.0496,
+      "step": 8876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019748278764068934,
+      "loss": 3.0013,
+      "step": 8877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019737884425640869,
+      "loss": 2.916,
+      "step": 8878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019727492150670538,
+      "loss": 3.0548,
+      "step": 8879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019717101939866545,
+      "loss": 2.9341,
+      "step": 8880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019706713793937348,
+      "loss": 2.8395,
+      "step": 8881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019696327713591277,
+      "loss": 2.8624,
+      "step": 8882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019685943699536523,
+      "loss": 2.9727,
+      "step": 8883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019675561752481124,
+      "loss": 2.9673,
+      "step": 8884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019665181873133,
+      "loss": 3.0056,
+      "step": 8885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019654804062199893,
+      "loss": 2.9589,
+      "step": 8886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019644428320389446,
+      "loss": 3.0859,
+      "step": 8887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019634054648409117,
+      "loss": 2.8541,
+      "step": 8888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019623683046966295,
+      "loss": 2.8818,
+      "step": 8889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019613313516768117,
+      "loss": 2.6719,
+      "step": 8890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001960294605852171,
+      "loss": 3.0433,
+      "step": 8891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019592580672933913,
+      "loss": 2.9696,
+      "step": 8892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019582217360711586,
+      "loss": 2.8987,
+      "step": 8893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019571856122561278,
+      "loss": 2.8502,
+      "step": 8894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019561496959189545,
+      "loss": 3.1589,
+      "step": 8895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001955113987130272,
+      "loss": 3.0578,
+      "step": 8896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019540784859607013,
+      "loss": 3.0554,
+      "step": 8897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019530431924808507,
+      "loss": 2.8826,
+      "step": 8898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019520081067613077,
+      "loss": 2.7245,
+      "step": 8899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019509732288726562,
+      "loss": 3.0587,
+      "step": 8900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001949938558885459,
+      "loss": 2.8664,
+      "step": 8901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019489040968702665,
+      "loss": 2.9927,
+      "step": 8902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001947869842897614,
+      "loss": 3.1099,
+      "step": 8903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001946835797038023,
+      "loss": 2.8301,
+      "step": 8904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019458019593620023,
+      "loss": 3.1164,
+      "step": 8905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001944768329940045,
+      "loss": 3.1265,
+      "step": 8906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019437349088426297,
+      "loss": 3.0307,
+      "step": 8907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019427016961402215,
+      "loss": 2.8528,
+      "step": 8908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019416686919032723,
+      "loss": 3.057,
+      "step": 8909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019406358962022175,
+      "loss": 2.9723,
+      "step": 8910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019396033091074783,
+      "loss": 3.0561,
+      "step": 8911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019385709306894683,
+      "loss": 2.9322,
+      "step": 8912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019375387610185736,
+      "loss": 2.6944,
+      "step": 8913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001936506800165182,
+      "loss": 3.1294,
+      "step": 8914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019354750481996507,
+      "loss": 2.8867,
+      "step": 8915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001934443505192337,
+      "loss": 2.9121,
+      "step": 8916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019334121712135765,
+      "loss": 2.9281,
+      "step": 8917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019323810463336912,
+      "loss": 3.0812,
+      "step": 8918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019313501306229892,
+      "loss": 2.8762,
+      "step": 8919
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019303194241517653,
+      "loss": 3.0443,
+      "step": 8920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019292889269902997,
+      "loss": 2.8438,
+      "step": 8921
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019282586392088574,
+      "loss": 3.0969,
+      "step": 8922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019272285608776902,
+      "loss": 2.779,
+      "step": 8923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001926198692067035,
+      "loss": 2.9274,
+      "step": 8924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019251690328471127,
+      "loss": 2.9175,
+      "step": 8925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019241395832881374,
+      "loss": 2.9494,
+      "step": 8926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001923110343460297,
+      "loss": 3.0418,
+      "step": 8927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019220813134337766,
+      "loss": 2.8705,
+      "step": 8928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019210524932787365,
+      "loss": 3.0001,
+      "step": 8929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001920023883065334,
+      "loss": 3.0243,
+      "step": 8930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019189954828637012,
+      "loss": 2.8838,
+      "step": 8931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019179672927439607,
+      "loss": 2.8398,
+      "step": 8932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019169393127762264,
+      "loss": 3.1371,
+      "step": 8933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019159115430305847,
+      "loss": 2.8101,
+      "step": 8934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019148839835771232,
+      "loss": 2.731,
+      "step": 8935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019138566344858993,
+      "loss": 2.9074,
+      "step": 8936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019128294958269714,
+      "loss": 3.187,
+      "step": 8937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019118025676703694,
+      "loss": 2.8658,
+      "step": 8938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000191077585008612,
+      "loss": 2.7119,
+      "step": 8939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019097493431442303,
+      "loss": 2.8534,
+      "step": 8940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019087230469146937,
+      "loss": 2.9134,
+      "step": 8941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019076969614674895,
+      "loss": 2.9483,
+      "step": 8942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019066710868725816,
+      "loss": 2.6917,
+      "step": 8943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019056454231999216,
+      "loss": 2.828,
+      "step": 8944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001904619970519445,
+      "loss": 2.9674,
+      "step": 8945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019035947289010736,
+      "loss": 3.0478,
+      "step": 8946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001902569698414715,
+      "loss": 2.8876,
+      "step": 8947
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019015448791302598,
+      "loss": 2.8654,
+      "step": 8948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00019005202711175917,
+      "loss": 2.9668,
+      "step": 8949
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001899495874446569,
+      "loss": 3.4141,
+      "step": 8950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001898471689187047,
+      "loss": 2.804,
+      "step": 8951
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018974477154088542,
+      "loss": 2.9352,
+      "step": 8952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001896423953181819,
+      "loss": 2.8086,
+      "step": 8953
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018954004025757405,
+      "loss": 2.8387,
+      "step": 8954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018943770636604157,
+      "loss": 3.1489,
+      "step": 8955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018933539365056213,
+      "loss": 3.0399,
+      "step": 8956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018923310211811195,
+      "loss": 2.9152,
+      "step": 8957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018913083177566592,
+      "loss": 2.8285,
+      "step": 8958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018902858263019744,
+      "loss": 3.0029,
+      "step": 8959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001889263546886786,
+      "loss": 3.1068,
+      "step": 8960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001888241479580798,
+      "loss": 2.9817,
+      "step": 8961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001887219624453702,
+      "loss": 2.8119,
+      "step": 8962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001886197981575174,
+      "loss": 2.9168,
+      "step": 8963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001885176551014876,
+      "loss": 3.027,
+      "step": 8964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018841553328424556,
+      "loss": 3.0457,
+      "step": 8965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018831343271275458,
+      "loss": 2.9912,
+      "step": 8966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001882113533939765,
+      "loss": 2.9064,
+      "step": 8967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018810929533487165,
+      "loss": 3.0026,
+      "step": 8968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001880072585423991,
+      "loss": 2.9293,
+      "step": 8969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018790524302351625,
+      "loss": 2.9629,
+      "step": 8970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.000187803248785179,
+      "loss": 2.9976,
+      "step": 8971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018770127583434253,
+      "loss": 2.7238,
+      "step": 8972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001875993241779592,
+      "loss": 2.8801,
+      "step": 8973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018749739382298143,
+      "loss": 3.0447,
+      "step": 8974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018739548477635882,
+      "loss": 2.7154,
+      "step": 8975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018729359704504057,
+      "loss": 2.9325,
+      "step": 8976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018719173063597394,
+      "loss": 2.8564,
+      "step": 8977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018708988555610472,
+      "loss": 2.8961,
+      "step": 8978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018698806181237742,
+      "loss": 2.9312,
+      "step": 8979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018688625941173497,
+      "loss": 2.9765,
+      "step": 8980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001867844783611189,
+      "loss": 2.614,
+      "step": 8981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018668271866746928,
+      "loss": 2.8272,
+      "step": 8982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001865809803377247,
+      "loss": 3.0229,
+      "step": 8983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018647926337882225,
+      "loss": 2.9766,
+      "step": 8984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018637756779769755,
+      "loss": 3.0935,
+      "step": 8985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018627589360128527,
+      "loss": 2.9558,
+      "step": 8986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018617424079651751,
+      "loss": 2.7925,
+      "step": 8987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018607260939032633,
+      "loss": 2.8272,
+      "step": 8988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001859709993896408,
+      "loss": 2.704,
+      "step": 8989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018586941080139008,
+      "loss": 3.0037,
+      "step": 8990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.00018576784363250043,
+      "loss": 2.9869,
+      "step": 8991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 0.0001856662978898978,
+      "loss": 3.0647,
+      "step": 8992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018556477358050606,
+      "loss": 3.1305,
+      "step": 8993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018546327071124769,
+      "loss": 2.9675,
+      "step": 8994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018536178928904408,
+      "loss": 2.9813,
+      "step": 8995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018526032932081421,
+      "loss": 2.929,
+      "step": 8996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018515889081347704,
+      "loss": 2.8755,
+      "step": 8997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018505747377394854,
+      "loss": 3.0308,
+      "step": 8998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018495607820914452,
+      "loss": 3.0072,
+      "step": 8999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018485470412597845,
+      "loss": 2.8162,
+      "step": 9000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018475335153136273,
+      "loss": 2.9595,
+      "step": 9001
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018465202043220825,
+      "loss": 2.9297,
+      "step": 9002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001845507108354243,
+      "loss": 2.7672,
+      "step": 9003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018444942274791886,
+      "loss": 3.028,
+      "step": 9004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018434815617659832,
+      "loss": 2.7543,
+      "step": 9005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001842469111283677,
+      "loss": 2.7615,
+      "step": 9006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018414568761013052,
+      "loss": 2.7157,
+      "step": 9007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001840444856287886,
+      "loss": 2.7809,
+      "step": 9008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018394330519124307,
+      "loss": 2.7848,
+      "step": 9009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018384214630439238,
+      "loss": 2.8574,
+      "step": 9010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001837410089751348,
+      "loss": 2.9515,
+      "step": 9011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001836398932103658,
+      "loss": 3.0439,
+      "step": 9012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001835387990169808,
+      "loss": 2.7992,
+      "step": 9013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018343772640187233,
+      "loss": 3.0563,
+      "step": 9014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018333667537193265,
+      "loss": 2.8517,
+      "step": 9015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001832356459340519,
+      "loss": 2.8639,
+      "step": 9016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018313463809511882,
+      "loss": 3.0142,
+      "step": 9017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018303365186202086,
+      "loss": 3.0436,
+      "step": 9018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001829326872416438,
+      "loss": 2.9637,
+      "step": 9019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000182831744240872,
+      "loss": 3.0749,
+      "step": 9020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001827308228665885,
+      "loss": 2.7619,
+      "step": 9021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001826299231256746,
+      "loss": 3.2985,
+      "step": 9022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001825290450250104,
+      "loss": 2.9098,
+      "step": 9023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018242818857147413,
+      "loss": 2.8679,
+      "step": 9024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018232735377194339,
+      "loss": 3.0934,
+      "step": 9025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018222654063329292,
+      "loss": 2.9309,
+      "step": 9026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018212574916239756,
+      "loss": 3.1148,
+      "step": 9027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018202497936612938,
+      "loss": 2.81,
+      "step": 9028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001819242312513596,
+      "loss": 2.9774,
+      "step": 9029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018182350482495792,
+      "loss": 2.8051,
+      "step": 9030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001817228000937922,
+      "loss": 2.7528,
+      "step": 9031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001816221170647298,
+      "loss": 2.9808,
+      "step": 9032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001815214557446351,
+      "loss": 2.9414,
+      "step": 9033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018142081614037247,
+      "loss": 2.7623,
+      "step": 9034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018132019825880359,
+      "loss": 2.7414,
+      "step": 9035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018121960210678957,
+      "loss": 2.7339,
+      "step": 9036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001811190276911896,
+      "loss": 2.7967,
+      "step": 9037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018101847501886148,
+      "loss": 3.2181,
+      "step": 9038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001809179440966614,
+      "loss": 2.8093,
+      "step": 9039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018081743493144432,
+      "loss": 2.7339,
+      "step": 9040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001807169475300634,
+      "loss": 3.082,
+      "step": 9041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018061648189937068,
+      "loss": 3.1233,
+      "step": 9042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001805160380462164,
+      "loss": 2.7968,
+      "step": 9043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018041561597744948,
+      "loss": 3.1195,
+      "step": 9044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018031521569991722,
+      "loss": 3.1027,
+      "step": 9045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018021483722046588,
+      "loss": 2.9301,
+      "step": 9046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018011448054593932,
+      "loss": 2.8803,
+      "step": 9047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00018001414568318114,
+      "loss": 2.9943,
+      "step": 9048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001799138326390321,
+      "loss": 2.7564,
+      "step": 9049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001798135414203328,
+      "loss": 3.0118,
+      "step": 9050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017971327203392112,
+      "loss": 3.0013,
+      "step": 9051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017961302448663452,
+      "loss": 2.8395,
+      "step": 9052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001795127987853083,
+      "loss": 2.9876,
+      "step": 9053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017941259493677646,
+      "loss": 2.9629,
+      "step": 9054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017931241294787153,
+      "loss": 2.8709,
+      "step": 9055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001792122528254246,
+      "loss": 3.2141,
+      "step": 9056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017911211457626513,
+      "loss": 2.9021,
+      "step": 9057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017901199820722114,
+      "loss": 2.9581,
+      "step": 9058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017891190372511928,
+      "loss": 2.7274,
+      "step": 9059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017881183113678455,
+      "loss": 2.8674,
+      "step": 9060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001787117804490405,
+      "loss": 2.9243,
+      "step": 9061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017861175166870925,
+      "loss": 2.966,
+      "step": 9062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017851174480261134,
+      "loss": 2.922,
+      "step": 9063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017841175985756585,
+      "loss": 2.9313,
+      "step": 9064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001783117968403904,
+      "loss": 3.0045,
+      "step": 9065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001782118557579011,
+      "loss": 2.9403,
+      "step": 9066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001781119366169125,
+      "loss": 3.2308,
+      "step": 9067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017801203942423755,
+      "loss": 2.9795,
+      "step": 9068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017791216418668837,
+      "loss": 2.9469,
+      "step": 9069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001778123109110744,
+      "loss": 2.9017,
+      "step": 9070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.000177712479604205,
+      "loss": 2.6549,
+      "step": 9071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017761267027288646,
+      "loss": 3.0579,
+      "step": 9072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017751288292392514,
+      "loss": 2.8735,
+      "step": 9073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001774131175641245,
+      "loss": 2.9582,
+      "step": 9074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017731337420028766,
+      "loss": 3.0423,
+      "step": 9075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017721365283921553,
+      "loss": 2.9791,
+      "step": 9076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017711395348770783,
+      "loss": 2.6723,
+      "step": 9077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017701427615256248,
+      "loss": 2.8729,
+      "step": 9078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001769146208405763,
+      "loss": 2.9467,
+      "step": 9079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017681498755854432,
+      "loss": 2.8997,
+      "step": 9080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017671537631326018,
+      "loss": 3.1012,
+      "step": 9081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001766157871115159,
+      "loss": 2.7794,
+      "step": 9082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001765162199601022,
+      "loss": 2.6804,
+      "step": 9083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017641667486580792,
+      "loss": 3.0345,
+      "step": 9084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017631715183542124,
+      "loss": 3.0391,
+      "step": 9085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017621765087572755,
+      "loss": 2.7712,
+      "step": 9086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001761181719935121,
+      "loss": 2.8427,
+      "step": 9087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017601871519555735,
+      "loss": 2.9507,
+      "step": 9088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017591928048864546,
+      "loss": 3.2193,
+      "step": 9089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001758198678795559,
+      "loss": 2.7165,
+      "step": 9090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017572047737506768,
+      "loss": 2.7164,
+      "step": 9091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017562110898195784,
+      "loss": 2.8776,
+      "step": 9092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001755217627070017,
+      "loss": 2.7114,
+      "step": 9093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017542243855697364,
+      "loss": 2.9046,
+      "step": 9094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017532313653864567,
+      "loss": 2.9338,
+      "step": 9095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017522385665878925,
+      "loss": 3.0996,
+      "step": 9096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017512459892417382,
+      "loss": 2.9699,
+      "step": 9097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017502536334156727,
+      "loss": 3.0028,
+      "step": 9098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017492614991773616,
+      "loss": 2.8794,
+      "step": 9099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017482695865944547,
+      "loss": 2.6979,
+      "step": 9100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001747277895734587,
+      "loss": 2.7403,
+      "step": 9101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017462864266653777,
+      "loss": 2.877,
+      "step": 9102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017452951794544314,
+      "loss": 2.9283,
+      "step": 9103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017443041541693373,
+      "loss": 3.0179,
+      "step": 9104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017433133508776676,
+      "loss": 2.9555,
+      "step": 9105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017423227696469873,
+      "loss": 2.6262,
+      "step": 9106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001741332410544833,
+      "loss": 2.7235,
+      "step": 9107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001740342273638741,
+      "loss": 2.9446,
+      "step": 9108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017393523589962173,
+      "loss": 2.8344,
+      "step": 9109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017383626666847674,
+      "loss": 3.0717,
+      "step": 9110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017373731967718675,
+      "loss": 2.9857,
+      "step": 9111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017363839493249916,
+      "loss": 2.8262,
+      "step": 9112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017353949244115906,
+      "loss": 2.9556,
+      "step": 9113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.0001734406122099103,
+      "loss": 2.7911,
+      "step": 9114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017334175424549503,
+      "loss": 2.9743,
+      "step": 9115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 0.00017324291855465402,
+      "loss": 3.0789,
+      "step": 9116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001731441051441266,
+      "loss": 2.8973,
+      "step": 9117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001730453140206503,
+      "loss": 2.8915,
+      "step": 9118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001729465451909615,
+      "loss": 3.068,
+      "step": 9119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017284779866179472,
+      "loss": 2.7202,
+      "step": 9120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017274907443988303,
+      "loss": 2.9785,
+      "step": 9121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017265037253195854,
+      "loss": 2.8549,
+      "step": 9122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017255169294475066,
+      "loss": 2.9237,
+      "step": 9123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017245303568498867,
+      "loss": 2.8765,
+      "step": 9124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001723544007593989,
+      "loss": 3.0867,
+      "step": 9125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017225578817470767,
+      "loss": 2.7416,
+      "step": 9126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001721571979376384,
+      "loss": 2.8523,
+      "step": 9127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017205863005491357,
+      "loss": 3.1447,
+      "step": 9128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017196008453325475,
+      "loss": 2.9337,
+      "step": 9129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017186156137938058,
+      "loss": 2.7817,
+      "step": 9130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017176306060000973,
+      "loss": 2.951,
+      "step": 9131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017166458220185793,
+      "loss": 2.6883,
+      "step": 9132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001715661261916407,
+      "loss": 3.0242,
+      "step": 9133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017146769257607065,
+      "loss": 2.9144,
+      "step": 9134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017136928136186015,
+      "loss": 2.9783,
+      "step": 9135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001712708925557193,
+      "loss": 2.8698,
+      "step": 9136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001711725261643569,
+      "loss": 2.825,
+      "step": 9137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017107418219448013,
+      "loss": 2.682,
+      "step": 9138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001709758606527947,
+      "loss": 2.9796,
+      "step": 9139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017087756154600476,
+      "loss": 2.7127,
+      "step": 9140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.000170779284880813,
+      "loss": 2.9322,
+      "step": 9141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001706810306639205,
+      "loss": 2.8668,
+      "step": 9142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017058279890202677,
+      "loss": 2.5984,
+      "step": 9143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017048458960182983,
+      "loss": 2.8793,
+      "step": 9144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001703864027700266,
+      "loss": 2.9556,
+      "step": 9145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017028823841331136,
+      "loss": 2.9912,
+      "step": 9146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017019009653837836,
+      "loss": 3.1975,
+      "step": 9147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00017009197715191877,
+      "loss": 3.0732,
+      "step": 9148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016999388026062363,
+      "loss": 2.8816,
+      "step": 9149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016989580587118113,
+      "loss": 2.6927,
+      "step": 9150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016979775399027913,
+      "loss": 2.5933,
+      "step": 9151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001696997246246032,
+      "loss": 2.8555,
+      "step": 9152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001696017177808376,
+      "loss": 3.1157,
+      "step": 9153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016950373346566507,
+      "loss": 2.7206,
+      "step": 9154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016940577168576677,
+      "loss": 2.8583,
+      "step": 9155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001693078324478223,
+      "loss": 2.8766,
+      "step": 9156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001692099157585098,
+      "loss": 2.9666,
+      "step": 9157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016911202162450583,
+      "loss": 2.9065,
+      "step": 9158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016901415005248538,
+      "loss": 2.7925,
+      "step": 9159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016891630104912199,
+      "loss": 3.002,
+      "step": 9160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016881847462108763,
+      "loss": 2.8584,
+      "step": 9161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001687206707750526,
+      "loss": 2.9328,
+      "step": 9162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016862288951768583,
+      "loss": 3.1348,
+      "step": 9163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001685251308556547,
+      "loss": 2.7044,
+      "step": 9164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001684273947956248,
+      "loss": 2.9041,
+      "step": 9165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001683296813442608,
+      "loss": 2.8325,
+      "step": 9166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016823199050822486,
+      "loss": 3.0493,
+      "step": 9167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016813432229417874,
+      "loss": 2.9736,
+      "step": 9168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016803667670878137,
+      "loss": 2.9426,
+      "step": 9169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016793905375869157,
+      "loss": 3.1278,
+      "step": 9170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016784145345056516,
+      "loss": 2.8233,
+      "step": 9171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016774387579105766,
+      "loss": 3.0239,
+      "step": 9172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016764632078682234,
+      "loss": 2.9828,
+      "step": 9173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001675487884445111,
+      "loss": 2.9009,
+      "step": 9174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016745127877077425,
+      "loss": 2.7931,
+      "step": 9175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016735379177226068,
+      "loss": 2.7989,
+      "step": 9176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001672563274556176,
+      "loss": 2.7198,
+      "step": 9177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016715888582749072,
+      "loss": 2.7847,
+      "step": 9178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001670614668945243,
+      "loss": 2.7144,
+      "step": 9179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016696407066336085,
+      "loss": 2.9064,
+      "step": 9180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016686669714064136,
+      "loss": 2.8685,
+      "step": 9181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016676934633300577,
+      "loss": 2.9052,
+      "step": 9182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016667201824709156,
+      "loss": 2.8777,
+      "step": 9183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016657471288953563,
+      "loss": 2.8545,
+      "step": 9184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016647743026697233,
+      "loss": 2.8936,
+      "step": 9185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016638017038603554,
+      "loss": 2.9479,
+      "step": 9186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016628293325335648,
+      "loss": 3.0648,
+      "step": 9187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016618571887556583,
+      "loss": 2.818,
+      "step": 9188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016608852725929208,
+      "loss": 2.8941,
+      "step": 9189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016599135841116242,
+      "loss": 2.687,
+      "step": 9190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016589421233780232,
+      "loss": 3.0544,
+      "step": 9191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016579708904583592,
+      "loss": 2.9921,
+      "step": 9192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016569998854188577,
+      "loss": 2.751,
+      "step": 9193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016560291083257228,
+      "loss": 2.8254,
+      "step": 9194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016550585592451533,
+      "loss": 2.9618,
+      "step": 9195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001654088238243326,
+      "loss": 2.8828,
+      "step": 9196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016531181453864025,
+      "loss": 2.8497,
+      "step": 9197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016521482807405302,
+      "loss": 2.8047,
+      "step": 9198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016511786443718402,
+      "loss": 2.7104,
+      "step": 9199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016502092363464488,
+      "loss": 2.9648,
+      "step": 9200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016492400567304555,
+      "loss": 3.0226,
+      "step": 9201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016482711055899456,
+      "loss": 2.9013,
+      "step": 9202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016473023829909877,
+      "loss": 2.7073,
+      "step": 9203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016463338889996344,
+      "loss": 3.2101,
+      "step": 9204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016453656236819274,
+      "loss": 2.7604,
+      "step": 9205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001644397587103883,
+      "loss": 2.9966,
+      "step": 9206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016434297793315144,
+      "loss": 2.7991,
+      "step": 9207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016424622004308065,
+      "loss": 2.9543,
+      "step": 9208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016414948504677414,
+      "loss": 2.8764,
+      "step": 9209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016405277295082716,
+      "loss": 2.9763,
+      "step": 9210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001639560837618347,
+      "loss": 2.7638,
+      "step": 9211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016385941748638938,
+      "loss": 3.0341,
+      "step": 9212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016376277413108258,
+      "loss": 2.9889,
+      "step": 9213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001636661537025041,
+      "loss": 3.0536,
+      "step": 9214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016356955620724196,
+      "loss": 2.8802,
+      "step": 9215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001634729816518829,
+      "loss": 3.0017,
+      "step": 9216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001633764300430119,
+      "loss": 2.8257,
+      "step": 9217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001632799013872125,
+      "loss": 2.8765,
+      "step": 9218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016318339569106654,
+      "loss": 2.853,
+      "step": 9219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016308691296115424,
+      "loss": 3.0729,
+      "step": 9220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016299045320405497,
+      "loss": 2.8654,
+      "step": 9221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016289401642634516,
+      "loss": 2.8368,
+      "step": 9222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001627976026346012,
+      "loss": 2.9719,
+      "step": 9223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001627012118353965,
+      "loss": 2.6219,
+      "step": 9224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016260484403530405,
+      "loss": 2.7231,
+      "step": 9225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016250849924089484,
+      "loss": 2.7755,
+      "step": 9226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016241217745873777,
+      "loss": 2.8715,
+      "step": 9227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016231587869540125,
+      "loss": 2.602,
+      "step": 9228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001622196029574509,
+      "loss": 2.9073,
+      "step": 9229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016212335025145207,
+      "loss": 2.7893,
+      "step": 9230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016202712058396724,
+      "loss": 2.9765,
+      "step": 9231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016193091396155828,
+      "loss": 2.8591,
+      "step": 9232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016183473039078518,
+      "loss": 2.8552,
+      "step": 9233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016173856987820618,
+      "loss": 2.8699,
+      "step": 9234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001616424324303782,
+      "loss": 2.9599,
+      "step": 9235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016154631805385638,
+      "loss": 2.6584,
+      "step": 9236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001614502267551945,
+      "loss": 2.9945,
+      "step": 9237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.0001613541585409446,
+      "loss": 2.945,
+      "step": 9238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016125811341765716,
+      "loss": 2.9276,
+      "step": 9239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 0.00016116209139188127,
+      "loss": 2.9342,
+      "step": 9240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016106609247016395,
+      "loss": 2.9237,
+      "step": 9241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016097011665905164,
+      "loss": 2.9802,
+      "step": 9242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001608741639650879,
+      "loss": 3.0623,
+      "step": 9243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000160778234394816,
+      "loss": 3.0781,
+      "step": 9244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016068232795477628,
+      "loss": 2.8464,
+      "step": 9245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016058644465150897,
+      "loss": 3.0469,
+      "step": 9246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001604905844915513,
+      "loss": 2.9868,
+      "step": 9247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016039474748144017,
+      "loss": 2.7168,
+      "step": 9248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016029893362771004,
+      "loss": 2.8326,
+      "step": 9249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016020314293689418,
+      "loss": 2.6816,
+      "step": 9250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016010737541552423,
+      "loss": 2.8629,
+      "step": 9251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00016001163107013018,
+      "loss": 2.8093,
+      "step": 9252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015991590990724038,
+      "loss": 2.9969,
+      "step": 9253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015982021193338176,
+      "loss": 2.9269,
+      "step": 9254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015972453715507962,
+      "loss": 2.7468,
+      "step": 9255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015962888557885757,
+      "loss": 3.0436,
+      "step": 9256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015953325721123786,
+      "loss": 2.8833,
+      "step": 9257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015943765205874094,
+      "loss": 2.8824,
+      "step": 9258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015934207012788576,
+      "loss": 2.8223,
+      "step": 9259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015924651142518965,
+      "loss": 2.9874,
+      "step": 9260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015915097595716848,
+      "loss": 3.052,
+      "step": 9261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015905546373033634,
+      "loss": 2.9245,
+      "step": 9262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000158959974751206,
+      "loss": 2.6786,
+      "step": 9263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015886450902628818,
+      "loss": 2.941,
+      "step": 9264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015876906656209284,
+      "loss": 2.9809,
+      "step": 9265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015867364736512723,
+      "loss": 2.9817,
+      "step": 9266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015857825144189824,
+      "loss": 2.9765,
+      "step": 9267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015848287879890994,
+      "loss": 2.8597,
+      "step": 9268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000158387529442666,
+      "loss": 2.7826,
+      "step": 9269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001582922033796673,
+      "loss": 2.7731,
+      "step": 9270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015819690061641433,
+      "loss": 3.0042,
+      "step": 9271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015810162115940512,
+      "loss": 2.8294,
+      "step": 9272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015800636501513656,
+      "loss": 2.4118,
+      "step": 9273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015791113219010367,
+      "loss": 2.8819,
+      "step": 9274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015781592269080007,
+      "loss": 2.964,
+      "step": 9275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015772073652371778,
+      "loss": 3.1714,
+      "step": 9276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015762557369534708,
+      "loss": 3.0067,
+      "step": 9277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015753043421217684,
+      "loss": 3.1568,
+      "step": 9278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015743531808069423,
+      "loss": 2.8868,
+      "step": 9279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015734022530738468,
+      "loss": 2.8528,
+      "step": 9280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015724515589873267,
+      "loss": 2.9977,
+      "step": 9281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015715010986122003,
+      "loss": 2.9042,
+      "step": 9282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015705508720132816,
+      "loss": 3.068,
+      "step": 9283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015696008792553568,
+      "loss": 2.8822,
+      "step": 9284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015686511204032072,
+      "loss": 3.0166,
+      "step": 9285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015677015955215917,
+      "loss": 3.1943,
+      "step": 9286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015667523046752552,
+      "loss": 2.7013,
+      "step": 9287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015658032479289247,
+      "loss": 2.8461,
+      "step": 9288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001564854425347314,
+      "loss": 2.8704,
+      "step": 9289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001563905836995121,
+      "loss": 2.7152,
+      "step": 9290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015629574829370213,
+      "loss": 3.0068,
+      "step": 9291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015620093632376847,
+      "loss": 2.8657,
+      "step": 9292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015610614779617576,
+      "loss": 2.967,
+      "step": 9293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015601138271738736,
+      "loss": 2.998,
+      "step": 9294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015591664109386487,
+      "loss": 3.0568,
+      "step": 9295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015582192293206836,
+      "loss": 2.6642,
+      "step": 9296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001557272282384563,
+      "loss": 2.7789,
+      "step": 9297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015563255701948558,
+      "loss": 2.7107,
+      "step": 9298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015553790928161148,
+      "loss": 2.987,
+      "step": 9299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015544328503128767,
+      "loss": 2.7746,
+      "step": 9300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000155348684274966,
+      "loss": 2.9209,
+      "step": 9301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015525410701909743,
+      "loss": 2.9688,
+      "step": 9302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015515955327013021,
+      "loss": 2.7715,
+      "step": 9303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001550650230345122,
+      "loss": 3.0612,
+      "step": 9304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015497051631868848,
+      "loss": 2.7542,
+      "step": 9305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001548760331291037,
+      "loss": 3.0512,
+      "step": 9306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015478157347219963,
+      "loss": 3.1225,
+      "step": 9307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015468713735441764,
+      "loss": 2.7975,
+      "step": 9308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015459272478219677,
+      "loss": 2.686,
+      "step": 9309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015449833576197468,
+      "loss": 2.8326,
+      "step": 9310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015440397030018743,
+      "loss": 2.9051,
+      "step": 9311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001543096284032694,
+      "loss": 2.7896,
+      "step": 9312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001542153100776534,
+      "loss": 3.2313,
+      "step": 9313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015412101532977064,
+      "loss": 2.9879,
+      "step": 9314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015402674416605067,
+      "loss": 2.9453,
+      "step": 9315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015393249659292159,
+      "loss": 2.6769,
+      "step": 9316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001538382726168095,
+      "loss": 3.0215,
+      "step": 9317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015374407224413972,
+      "loss": 2.6999,
+      "step": 9318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001536498954813348,
+      "loss": 2.8535,
+      "step": 9319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015355574233481684,
+      "loss": 3.0384,
+      "step": 9320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015346161281100523,
+      "loss": 2.9609,
+      "step": 9321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001533675069163189,
+      "loss": 2.7839,
+      "step": 9322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015327342465717408,
+      "loss": 2.8976,
+      "step": 9323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015317936603998595,
+      "loss": 2.9403,
+      "step": 9324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001530853310711684,
+      "loss": 2.8455,
+      "step": 9325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015299131975713266,
+      "loss": 3.1161,
+      "step": 9326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015289733210428974,
+      "loss": 2.9009,
+      "step": 9327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001528033681190476,
+      "loss": 3.0615,
+      "step": 9328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015270942780781395,
+      "loss": 2.9569,
+      "step": 9329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015261551117699357,
+      "loss": 2.9354,
+      "step": 9330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001525216182329907,
+      "loss": 2.7814,
+      "step": 9331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015242774898220745,
+      "loss": 2.7415,
+      "step": 9332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015233390343104435,
+      "loss": 2.9384,
+      "step": 9333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015224008158590046,
+      "loss": 2.6974,
+      "step": 9334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015214628345317305,
+      "loss": 3.0849,
+      "step": 9335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001520525090392579,
+      "loss": 2.7506,
+      "step": 9336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015195875835054902,
+      "loss": 2.5669,
+      "step": 9337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015186503139343903,
+      "loss": 2.9735,
+      "step": 9338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015177132817431884,
+      "loss": 2.9809,
+      "step": 9339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015167764869957744,
+      "loss": 2.7045,
+      "step": 9340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.000151583992975603,
+      "loss": 2.8652,
+      "step": 9341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015149036100878095,
+      "loss": 2.9467,
+      "step": 9342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015139675280549626,
+      "loss": 2.8647,
+      "step": 9343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015130316837213116,
+      "loss": 2.8145,
+      "step": 9344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001512096077150672,
+      "loss": 2.9804,
+      "step": 9345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001511160708406838,
+      "loss": 3.0914,
+      "step": 9346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015102255775535889,
+      "loss": 2.9684,
+      "step": 9347
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015092906846546879,
+      "loss": 3.0105,
+      "step": 9348
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015083560297738814,
+      "loss": 2.8892,
+      "step": 9349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015074216129749003,
+      "loss": 3.1078,
+      "step": 9350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015064874343214584,
+      "loss": 3.0495,
+      "step": 9351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015055534938772543,
+      "loss": 2.9644,
+      "step": 9352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001504619791705969,
+      "loss": 2.9219,
+      "step": 9353
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015036863278712693,
+      "loss": 3.004,
+      "step": 9354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015027531024368036,
+      "loss": 2.6115,
+      "step": 9355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001501820115466205,
+      "loss": 2.8043,
+      "step": 9356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00015008873670230904,
+      "loss": 2.8062,
+      "step": 9357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014999548571710615,
+      "loss": 2.8706,
+      "step": 9358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014990225859737007,
+      "loss": 3.071,
+      "step": 9359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014980905534945764,
+      "loss": 3.1385,
+      "step": 9360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014971587597972403,
+      "loss": 2.9611,
+      "step": 9361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001496227204945231,
+      "loss": 2.786,
+      "step": 9362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014952958890020619,
+      "loss": 3.1001,
+      "step": 9363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.0001494364812031242,
+      "loss": 2.873,
+      "step": 9364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 0.00014934339740962517,
+      "loss": 3.0056,
+      "step": 9365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014925033752605671,
+      "loss": 3.1892,
+      "step": 9366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014915730155876366,
+      "loss": 2.9917,
+      "step": 9367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014906428951409024,
+      "loss": 2.9789,
+      "step": 9368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014897130139837844,
+      "loss": 3.0813,
+      "step": 9369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014887833721796874,
+      "loss": 2.9362,
+      "step": 9370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014878539697919997,
+      "loss": 2.8087,
+      "step": 9371
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001486924806884094,
+      "loss": 3.0477,
+      "step": 9372
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014859958835193265,
+      "loss": 2.9449,
+      "step": 9373
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001485067199761037,
+      "loss": 2.8509,
+      "step": 9374
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001484138755672549,
+      "loss": 2.8792,
+      "step": 9375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014832105513171696,
+      "loss": 2.8404,
+      "step": 9376
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014822825867581867,
+      "loss": 2.9967,
+      "step": 9377
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014813548620588814,
+      "loss": 2.8415,
+      "step": 9378
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014804273772825034,
+      "loss": 2.6776,
+      "step": 9379
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014795001324923018,
+      "loss": 2.8601,
+      "step": 9380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014785731277514952,
+      "loss": 2.7671,
+      "step": 9381
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014776463631232988,
+      "loss": 2.7558,
+      "step": 9382
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014767198386708997,
+      "loss": 2.7539,
+      "step": 9383
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014757935544574774,
+      "loss": 2.8112,
+      "step": 9384
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014748675105461905,
+      "loss": 2.9604,
+      "step": 9385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014739417070001832,
+      "loss": 2.9058,
+      "step": 9386
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014730161438825817,
+      "loss": 2.8372,
+      "step": 9387
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014720908212564966,
+      "loss": 2.9555,
+      "step": 9388
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001471165739185024,
+      "loss": 2.8446,
+      "step": 9389
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014702408977312372,
+      "loss": 3.0129,
+      "step": 9390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014693162969582014,
+      "loss": 3.0375,
+      "step": 9391
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014683919369289612,
+      "loss": 2.9535,
+      "step": 9392
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014674678177065438,
+      "loss": 2.7711,
+      "step": 9393
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014665439393539626,
+      "loss": 2.775,
+      "step": 9394
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001465620301934213,
+      "loss": 2.9627,
+      "step": 9395
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014646969055102737,
+      "loss": 3.0589,
+      "step": 9396
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014637737501451082,
+      "loss": 2.982,
+      "step": 9397
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014628508359016624,
+      "loss": 3.0715,
+      "step": 9398
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001461928162842867,
+      "loss": 2.8254,
+      "step": 9399
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014610057310316332,
+      "loss": 3.0004,
+      "step": 9400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014600835405308637,
+      "loss": 2.9419,
+      "step": 9401
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014591615914034324,
+      "loss": 2.9365,
+      "step": 9402
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014582398837122092,
+      "loss": 2.7411,
+      "step": 9403
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001457318417520036,
+      "loss": 2.8198,
+      "step": 9404
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001456397192889749,
+      "loss": 2.9221,
+      "step": 9405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014554762098841612,
+      "loss": 2.6195,
+      "step": 9406
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014545554685660713,
+      "loss": 2.6836,
+      "step": 9407
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014536349689982597,
+      "loss": 2.8598,
+      "step": 9408
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014527147112434934,
+      "loss": 3.1627,
+      "step": 9409
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014517946953645206,
+      "loss": 2.9655,
+      "step": 9410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001450874921424074,
+      "loss": 2.8164,
+      "step": 9411
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001449955389484869,
+      "loss": 3.105,
+      "step": 9412
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014490360996096048,
+      "loss": 2.97,
+      "step": 9413
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014481170518609643,
+      "loss": 3.0008,
+      "step": 9414
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014471982463016143,
+      "loss": 2.9485,
+      "step": 9415
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001446279682994203,
+      "loss": 2.709,
+      "step": 9416
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014453613620013685,
+      "loss": 3.1209,
+      "step": 9417
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014444432833857207,
+      "loss": 2.8599,
+      "step": 9418
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014435254472098668,
+      "loss": 2.7911,
+      "step": 9419
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014426078535363835,
+      "loss": 2.8709,
+      "step": 9420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014416905024278436,
+      "loss": 2.8222,
+      "step": 9421
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014407733939467976,
+      "loss": 2.9554,
+      "step": 9422
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001439856528155774,
+      "loss": 2.9845,
+      "step": 9423
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014389399051172974,
+      "loss": 2.7322,
+      "step": 9424
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001438023524893863,
+      "loss": 2.9957,
+      "step": 9425
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014371073875479607,
+      "loss": 3.0928,
+      "step": 9426
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014361914931420523,
+      "loss": 3.0006,
+      "step": 9427
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014352758417385936,
+      "loss": 2.79,
+      "step": 9428
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014343604334000188,
+      "loss": 3.0219,
+      "step": 9429
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001433445268188745,
+      "loss": 2.826,
+      "step": 9430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001432530346167174,
+      "loss": 3.0727,
+      "step": 9431
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014316156673976915,
+      "loss": 3.0024,
+      "step": 9432
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001430701231942665,
+      "loss": 2.7319,
+      "step": 9433
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014297870398644475,
+      "loss": 2.8602,
+      "step": 9434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001428873091225373,
+      "loss": 3.1231,
+      "step": 9435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001427959386087761,
+      "loss": 2.9846,
+      "step": 9436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014270459245139118,
+      "loss": 2.8606,
+      "step": 9437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001426132706566115,
+      "loss": 2.6354,
+      "step": 9438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001425219732306634,
+      "loss": 2.6272,
+      "step": 9439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014243070017977267,
+      "loss": 2.988,
+      "step": 9440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001423394515101623,
+      "loss": 2.9784,
+      "step": 9441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014224822722805474,
+      "loss": 2.8422,
+      "step": 9442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001421570273396697,
+      "loss": 2.9801,
+      "step": 9443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014206585185122606,
+      "loss": 2.6062,
+      "step": 9444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014197470076894077,
+      "loss": 3.1074,
+      "step": 9445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001418835740990289,
+      "loss": 2.948,
+      "step": 9446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001417924718477041,
+      "loss": 2.9325,
+      "step": 9447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014170139402117833,
+      "loss": 3.0566,
+      "step": 9448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014161034062566176,
+      "loss": 2.9191,
+      "step": 9449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014151931166736303,
+      "loss": 2.8736,
+      "step": 9450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.000141428307152489,
+      "loss": 2.9496,
+      "step": 9451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014133732708724494,
+      "loss": 2.8137,
+      "step": 9452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014124637147783432,
+      "loss": 3.1256,
+      "step": 9453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014115544033045947,
+      "loss": 2.4976,
+      "step": 9454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014106453365132023,
+      "loss": 2.8948,
+      "step": 9455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014097365144661523,
+      "loss": 2.8131,
+      "step": 9456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001408827937225415,
+      "loss": 2.7585,
+      "step": 9457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014079196048529414,
+      "loss": 3.005,
+      "step": 9458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014070115174106684,
+      "loss": 3.0703,
+      "step": 9459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014061036749605127,
+      "loss": 2.7389,
+      "step": 9460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014051960775643825,
+      "loss": 2.9374,
+      "step": 9461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014042887252841552,
+      "loss": 2.8464,
+      "step": 9462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014033816181817078,
+      "loss": 3.036,
+      "step": 9463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001402474756318885,
+      "loss": 3.1041,
+      "step": 9464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00014015681397575274,
+      "loss": 3.2284,
+      "step": 9465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001400661768559452,
+      "loss": 2.9321,
+      "step": 9466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001399755642786461,
+      "loss": 2.9989,
+      "step": 9467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013988497625003394,
+      "loss": 3.0878,
+      "step": 9468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013979441277628556,
+      "loss": 2.9808,
+      "step": 9469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001397038738635762,
+      "loss": 2.5768,
+      "step": 9470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001396133595180793,
+      "loss": 2.7398,
+      "step": 9471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013952286974596673,
+      "loss": 2.9669,
+      "step": 9472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013943240455340856,
+      "loss": 2.9048,
+      "step": 9473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013934196394657333,
+      "loss": 2.7639,
+      "step": 9474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013925154793162787,
+      "loss": 2.8986,
+      "step": 9475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001391611565147371,
+      "loss": 3.0053,
+      "step": 9476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013907078970206487,
+      "loss": 2.8837,
+      "step": 9477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013898044749977245,
+      "loss": 3.1521,
+      "step": 9478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001388901299140205,
+      "loss": 3.224,
+      "step": 9479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013879983695096675,
+      "loss": 2.9304,
+      "step": 9480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001387095686167685,
+      "loss": 3.0646,
+      "step": 9481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013861932491758056,
+      "loss": 2.6823,
+      "step": 9482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013852910585955637,
+      "loss": 2.9355,
+      "step": 9483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013843891144884758,
+      "loss": 2.7934,
+      "step": 9484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013834874169160415,
+      "loss": 3.0041,
+      "step": 9485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.0001382585965939745,
+      "loss": 2.901,
+      "step": 9486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013816847616210525,
+      "loss": 3.0945,
+      "step": 9487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013807838040214137,
+      "loss": 3.0062,
+      "step": 9488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 0.00013798830932022617,
+      "loss": 3.02,
+      "step": 9489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001378982629225012,
+      "loss": 2.7384,
+      "step": 9490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013780824121510638,
+      "loss": 2.9344,
+      "step": 9491
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013771824420418,
+      "loss": 3.0194,
+      "step": 9492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013762827189585852,
+      "loss": 2.8671,
+      "step": 9493
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013753832429627694,
+      "loss": 3.142,
+      "step": 9494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013744840141156827,
+      "loss": 2.9825,
+      "step": 9495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001373585032478642,
+      "loss": 2.855,
+      "step": 9496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013726862981129422,
+      "loss": 2.8655,
+      "step": 9497
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013717878110798703,
+      "loss": 3.017,
+      "step": 9498
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013708895714406843,
+      "loss": 2.746,
+      "step": 9499
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001369991579256637,
+      "loss": 3.0457,
+      "step": 9500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013690938345889541,
+      "loss": 2.8507,
+      "step": 9501
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001368196337498855,
+      "loss": 2.987,
+      "step": 9502
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013672990880475306,
+      "loss": 3.0675,
+      "step": 9503
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013664020862961645,
+      "loss": 2.7316,
+      "step": 9504
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013655053323059196,
+      "loss": 2.7216,
+      "step": 9505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001364608826137942,
+      "loss": 3.1455,
+      "step": 9506
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000136371256785336,
+      "loss": 2.9481,
+      "step": 9507
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013628165575132873,
+      "loss": 2.7555,
+      "step": 9508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013619207951788188,
+      "loss": 3.137,
+      "step": 9509
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001361025280911033,
+      "loss": 2.8053,
+      "step": 9510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013601300147709916,
+      "loss": 2.7873,
+      "step": 9511
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013592349968197392,
+      "loss": 3.0145,
+      "step": 9512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013583402271183026,
+      "loss": 3.1805,
+      "step": 9513
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013574457057276973,
+      "loss": 3.212,
+      "step": 9514
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013565514327089107,
+      "loss": 2.8978,
+      "step": 9515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013556574081229262,
+      "loss": 3.0683,
+      "step": 9516
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013547636320306977,
+      "loss": 2.9289,
+      "step": 9517
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013538701044931751,
+      "loss": 2.7927,
+      "step": 9518
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013529768255712778,
+      "loss": 2.9761,
+      "step": 9519
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001352083795325919,
+      "loss": 2.9621,
+      "step": 9520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013511910138179928,
+      "loss": 2.6924,
+      "step": 9521
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001350298481108369,
+      "loss": 3.203,
+      "step": 9522
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001349406197257912,
+      "loss": 2.9537,
+      "step": 9523
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013485141623274566,
+      "loss": 2.914,
+      "step": 9524
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013476223763778333,
+      "loss": 3.1461,
+      "step": 9525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013467308394698467,
+      "loss": 2.6776,
+      "step": 9526
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001345839551664288,
+      "loss": 2.8471,
+      "step": 9527
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013449485130219308,
+      "loss": 3.0127,
+      "step": 9528
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013440577236035307,
+      "loss": 2.457,
+      "step": 9529
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013431671834698283,
+      "loss": 3.0434,
+      "step": 9530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013422768926815454,
+      "loss": 2.9299,
+      "step": 9531
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013413868512993877,
+      "loss": 2.9793,
+      "step": 9532
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013404970593840438,
+      "loss": 2.9618,
+      "step": 9533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001339607516996185,
+      "loss": 2.918,
+      "step": 9534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013387182241964663,
+      "loss": 2.8504,
+      "step": 9535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013378291810455228,
+      "loss": 2.6165,
+      "step": 9536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013369403876039803,
+      "loss": 2.9244,
+      "step": 9537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013360518439324364,
+      "loss": 2.9122,
+      "step": 9538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013351635500914822,
+      "loss": 3.0547,
+      "step": 9539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013342755061416823,
+      "loss": 2.7786,
+      "step": 9540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001333387712143593,
+      "loss": 2.8944,
+      "step": 9541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001332500168157748,
+      "loss": 2.879,
+      "step": 9542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013316128742446655,
+      "loss": 2.8503,
+      "step": 9543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013307258304648468,
+      "loss": 3.0665,
+      "step": 9544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013298390368787766,
+      "loss": 2.8164,
+      "step": 9545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013289524935469204,
+      "loss": 2.9759,
+      "step": 9546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000132806620052973,
+      "loss": 2.8528,
+      "step": 9547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013271801578876375,
+      "loss": 2.8117,
+      "step": 9548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013262943656810582,
+      "loss": 2.9531,
+      "step": 9549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001325408823970392,
+      "loss": 2.9712,
+      "step": 9550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000132452353281602,
+      "loss": 2.9603,
+      "step": 9551
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013236384922783061,
+      "loss": 2.8873,
+      "step": 9552
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013227537024175994,
+      "loss": 2.9863,
+      "step": 9553
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013218691632942291,
+      "loss": 2.7747,
+      "step": 9554
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001320984874968509,
+      "loss": 3.027,
+      "step": 9555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013201008375007356,
+      "loss": 2.9757,
+      "step": 9556
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001319217050951186,
+      "loss": 3.0088,
+      "step": 9557
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013183335153801263,
+      "loss": 2.9477,
+      "step": 9558
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013174502308477965,
+      "loss": 2.9333,
+      "step": 9559
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000131656719741443,
+      "loss": 2.6955,
+      "step": 9560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013156844151402313,
+      "loss": 2.899,
+      "step": 9561
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013148018840854003,
+      "loss": 2.8242,
+      "step": 9562
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013139196043101064,
+      "loss": 2.7803,
+      "step": 9563
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001313037575874515,
+      "loss": 2.9811,
+      "step": 9564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013121557988387655,
+      "loss": 2.9418,
+      "step": 9565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001311274273262983,
+      "loss": 2.9035,
+      "step": 9566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013103929992072755,
+      "loss": 2.9938,
+      "step": 9567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013095119767317343,
+      "loss": 2.7732,
+      "step": 9568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013086312058964322,
+      "loss": 2.9766,
+      "step": 9569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001307750686761426,
+      "loss": 2.9692,
+      "step": 9570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013068704193867542,
+      "loss": 2.8761,
+      "step": 9571
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000130599040383244,
+      "loss": 3.031,
+      "step": 9572
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013051106401584857,
+      "loss": 2.8531,
+      "step": 9573
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013042311284248843,
+      "loss": 2.8464,
+      "step": 9574
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013033518686916002,
+      "loss": 2.929,
+      "step": 9575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013024728610185927,
+      "loss": 3.0345,
+      "step": 9576
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00013015941054657915,
+      "loss": 2.9076,
+      "step": 9577
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001300715602093122,
+      "loss": 2.9198,
+      "step": 9578
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.000129983735096048,
+      "loss": 2.8029,
+      "step": 9579
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012989593521277543,
+      "loss": 3.1785,
+      "step": 9580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012980816056548116,
+      "loss": 3.149,
+      "step": 9581
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001297204111601501,
+      "loss": 2.9743,
+      "step": 9582
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001296326870027656,
+      "loss": 2.9821,
+      "step": 9583
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001295449880993092,
+      "loss": 2.8328,
+      "step": 9584
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001294573144557608,
+      "loss": 2.8078,
+      "step": 9585
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012936966607809852,
+      "loss": 3.0111,
+      "step": 9586
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012928204297229872,
+      "loss": 3.0617,
+      "step": 9587
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012919444514433614,
+      "loss": 2.881,
+      "step": 9588
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012910687260018377,
+      "loss": 3.0939,
+      "step": 9589
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012901932534581273,
+      "loss": 3.1935,
+      "step": 9590
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012893180338719262,
+      "loss": 2.7955,
+      "step": 9591
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012884430673029118,
+      "loss": 2.8723,
+      "step": 9592
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012875683538107454,
+      "loss": 2.8247,
+      "step": 9593
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012866938934550694,
+      "loss": 3.0234,
+      "step": 9594
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012858196862955106,
+      "loss": 2.9367,
+      "step": 9595
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012849457323916763,
+      "loss": 3.048,
+      "step": 9596
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012840720318031625,
+      "loss": 2.6499,
+      "step": 9597
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012831985845895372,
+      "loss": 3.0512,
+      "step": 9598
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001282325390810364,
+      "loss": 2.9458,
+      "step": 9599
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012814524505251756,
+      "loss": 2.8336,
+      "step": 9600
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012805797637935,
+      "loss": 3.2391,
+      "step": 9601
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012797073306748409,
+      "loss": 2.8349,
+      "step": 9602
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001278835151228685,
+      "loss": 3.0224,
+      "step": 9603
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001277963225514504,
+      "loss": 2.8795,
+      "step": 9604
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.0001277091553591751,
+      "loss": 3.124,
+      "step": 9605
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012762201355198617,
+      "loss": 2.8301,
+      "step": 9606
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012753489713582546,
+      "loss": 2.9056,
+      "step": 9607
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012744780611663315,
+      "loss": 2.7918,
+      "step": 9608
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012736074050034762,
+      "loss": 3.0854,
+      "step": 9609
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012727370029290552,
+      "loss": 2.88,
+      "step": 9610
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012718668550024188,
+      "loss": 2.8999,
+      "step": 9611
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012709969612828965,
+      "loss": 2.8102,
+      "step": 9612
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 0.00012701273218298077,
+      "loss": 2.7181,
+      "step": 9613
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012692579367024436,
+      "loss": 2.8232,
+      "step": 9614
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001268388805960091,
+      "loss": 2.7544,
+      "step": 9615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012675199296620067,
+      "loss": 2.8447,
+      "step": 9616
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012666513078674396,
+      "loss": 2.947,
+      "step": 9617
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001265782940635618,
+      "loss": 2.7687,
+      "step": 9618
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001264914828025749,
+      "loss": 3.1429,
+      "step": 9619
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012640469700970304,
+      "loss": 2.783,
+      "step": 9620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012631793669086322,
+      "loss": 2.9817,
+      "step": 9621
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012623120185197202,
+      "loss": 2.8875,
+      "step": 9622
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012614449249894288,
+      "loss": 2.789,
+      "step": 9623
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012605780863768863,
+      "loss": 3.0601,
+      "step": 9624
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012597115027411976,
+      "loss": 2.9295,
+      "step": 9625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012588451741414518,
+      "loss": 2.8133,
+      "step": 9626
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012579791006367207,
+      "loss": 3.0095,
+      "step": 9627
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001257113282286058,
+      "loss": 2.9499,
+      "step": 9628
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012562477191485012,
+      "loss": 2.7186,
+      "step": 9629
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012553824112830693,
+      "loss": 2.6905,
+      "step": 9630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012545173587487642,
+      "loss": 2.9028,
+      "step": 9631
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012536525616045708,
+      "loss": 2.9603,
+      "step": 9632
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012527880199094548,
+      "loss": 2.9191,
+      "step": 9633
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012519237337223704,
+      "loss": 3.0293,
+      "step": 9634
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012510597031022442,
+      "loss": 2.8811,
+      "step": 9635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012501959281079967,
+      "loss": 3.0726,
+      "step": 9636
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012493324087985199,
+      "loss": 2.9003,
+      "step": 9637
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001248469145232699,
+      "loss": 3.0668,
+      "step": 9638
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001247606137469392,
+      "loss": 2.8817,
+      "step": 9639
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012467433855674475,
+      "loss": 2.8585,
+      "step": 9640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001245880889585692,
+      "loss": 2.7994,
+      "step": 9641
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012450186495829363,
+      "loss": 2.9085,
+      "step": 9642
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012441566656179725,
+      "loss": 2.8397,
+      "step": 9643
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012432949377495767,
+      "loss": 2.7586,
+      "step": 9644
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001242433466036506,
+      "loss": 2.9222,
+      "step": 9645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012415722505375017,
+      "loss": 2.9482,
+      "step": 9646
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001240711291311286,
+      "loss": 2.9848,
+      "step": 9647
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001239850588416565,
+      "loss": 2.9679,
+      "step": 9648
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012389901419120248,
+      "loss": 2.9056,
+      "step": 9649
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001238129951856341,
+      "loss": 2.9194,
+      "step": 9650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012372700183081614,
+      "loss": 3.0564,
+      "step": 9651
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012364103413261242,
+      "loss": 2.8993,
+      "step": 9652
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012355509209688458,
+      "loss": 2.6679,
+      "step": 9653
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012346917572949285,
+      "loss": 3.1945,
+      "step": 9654
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012338328503629548,
+      "loss": 2.5942,
+      "step": 9655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001232974200231488,
+      "loss": 2.9852,
+      "step": 9656
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001232115806959082,
+      "loss": 2.9054,
+      "step": 9657
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.000123125767060426,
+      "loss": 3.0166,
+      "step": 9658
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012303997912255422,
+      "loss": 3.03,
+      "step": 9659
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012295421688814174,
+      "loss": 3.1486,
+      "step": 9660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012286848036303683,
+      "loss": 2.9626,
+      "step": 9661
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001227827695530853,
+      "loss": 2.9913,
+      "step": 9662
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012269708446413163,
+      "loss": 3.1045,
+      "step": 9663
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012261142510201824,
+      "loss": 3.0549,
+      "step": 9664
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012252579147258586,
+      "loss": 2.9932,
+      "step": 9665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012244018358167363,
+      "loss": 3.0374,
+      "step": 9666
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012235460143511878,
+      "loss": 2.6905,
+      "step": 9667
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012226904503875679,
+      "loss": 2.8619,
+      "step": 9668
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001221835143984215,
+      "loss": 2.9177,
+      "step": 9669
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012209800951994487,
+      "loss": 3.0239,
+      "step": 9670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012201253040915716,
+      "loss": 3.0352,
+      "step": 9671
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012192707707188671,
+      "loss": 2.8313,
+      "step": 9672
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012184164951396066,
+      "loss": 2.865,
+      "step": 9673
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012175624774120348,
+      "loss": 2.9631,
+      "step": 9674
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012167087175943892,
+      "loss": 2.8317,
+      "step": 9675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012158552157448787,
+      "loss": 2.7976,
+      "step": 9676
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012150019719217048,
+      "loss": 2.9927,
+      "step": 9677
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012141489861830451,
+      "loss": 2.8261,
+      "step": 9678
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012132962585870621,
+      "loss": 2.8687,
+      "step": 9679
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012124437891918994,
+      "loss": 3.0235,
+      "step": 9680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012115915780556841,
+      "loss": 3.0376,
+      "step": 9681
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001210739625236525,
+      "loss": 2.7418,
+      "step": 9682
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012098879307925132,
+      "loss": 2.8296,
+      "step": 9683
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012090364947817228,
+      "loss": 3.0303,
+      "step": 9684
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012081853172622098,
+      "loss": 2.9138,
+      "step": 9685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012073343982920132,
+      "loss": 3.0374,
+      "step": 9686
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012064837379291538,
+      "loss": 3.0129,
+      "step": 9687
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001205633336231634,
+      "loss": 3.1993,
+      "step": 9688
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012047831932574405,
+      "loss": 3.0185,
+      "step": 9689
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012039333090645405,
+      "loss": 2.8598,
+      "step": 9690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012030836837108849,
+      "loss": 2.9739,
+      "step": 9691
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012022343172544058,
+      "loss": 2.9639,
+      "step": 9692
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012013852097530175,
+      "loss": 3.0103,
+      "step": 9693
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00012005363612646214,
+      "loss": 3.0035,
+      "step": 9694
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011996877718470917,
+      "loss": 2.9948,
+      "step": 9695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001198839441558296,
+      "loss": 2.8757,
+      "step": 9696
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011979913704560724,
+      "loss": 2.8204,
+      "step": 9697
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011971435585982543,
+      "loss": 2.9045,
+      "step": 9698
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011962960060426447,
+      "loss": 2.8874,
+      "step": 9699
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011954487128470387,
+      "loss": 3.1714,
+      "step": 9700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011946016790692094,
+      "loss": 3.0845,
+      "step": 9701
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011937549047669122,
+      "loss": 2.9051,
+      "step": 9702
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.0001192908389997886,
+      "loss": 2.7837,
+      "step": 9703
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011920621348198507,
+      "loss": 2.8961,
+      "step": 9704
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011912161392905091,
+      "loss": 2.8978,
+      "step": 9705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011903704034675472,
+      "loss": 2.8812,
+      "step": 9706
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011895249274086323,
+      "loss": 2.8454,
+      "step": 9707
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011886797111714137,
+      "loss": 2.8817,
+      "step": 9708
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011878347548135221,
+      "loss": 2.6698,
+      "step": 9709
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011869900583925764,
+      "loss": 2.7338,
+      "step": 9710
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011861456219661676,
+      "loss": 2.9624,
+      "step": 9711
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011853014455918798,
+      "loss": 2.8275,
+      "step": 9712
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011844575293272692,
+      "loss": 2.881,
+      "step": 9713
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011836138732298841,
+      "loss": 2.6908,
+      "step": 9714
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011827704773572445,
+      "loss": 2.8432,
+      "step": 9715
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011819273417668636,
+      "loss": 2.8292,
+      "step": 9716
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011810844665162301,
+      "loss": 2.9194,
+      "step": 9717
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011802418516628139,
+      "loss": 2.7996,
+      "step": 9718
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011793994972640736,
+      "loss": 3.1379,
+      "step": 9719
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011785574033774421,
+      "loss": 3.1947,
+      "step": 9720
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011777155700603415,
+      "loss": 3.0029,
+      "step": 9721
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011768739973701725,
+      "loss": 2.8908,
+      "step": 9722
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011760326853643183,
+      "loss": 3.0672,
+      "step": 9723
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011751916341001451,
+      "loss": 2.9455,
+      "step": 9724
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011743508436350009,
+      "loss": 2.8674,
+      "step": 9725
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011735103140262155,
+      "loss": 2.8628,
+      "step": 9726
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011726700453311023,
+      "loss": 2.808,
+      "step": 9727
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011718300376069551,
+      "loss": 2.7495,
+      "step": 9728
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011709902909110515,
+      "loss": 3.2853,
+      "step": 9729
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011701508053006499,
+      "loss": 2.7587,
+      "step": 9730
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011693115808329919,
+      "loss": 2.8367,
+      "step": 9731
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011684726175652999,
+      "loss": 2.7568,
+      "step": 9732
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011676339155547838,
+      "loss": 3.0104,
+      "step": 9733
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011667954748586257,
+      "loss": 2.8674,
+      "step": 9734
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011659572955340009,
+      "loss": 2.8634,
+      "step": 9735
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011651193776380564,
+      "loss": 2.8484,
+      "step": 9736
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 0.00011642817212279311,
+      "loss": 2.929,
+      "step": 9737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011634443263607398,
+      "loss": 3.0073,
+      "step": 9738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001162607193093581,
+      "loss": 2.9784,
+      "step": 9739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011617703214835362,
+      "loss": 3.0083,
+      "step": 9740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011609337115876678,
+      "loss": 2.7299,
+      "step": 9741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011600973634630219,
+      "loss": 2.7941,
+      "step": 9742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001159261277166625,
+      "loss": 2.7087,
+      "step": 9743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011584254527554872,
+      "loss": 2.7316,
+      "step": 9744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011575898902866,
+      "loss": 2.8768,
+      "step": 9745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011567545898169374,
+      "loss": 2.9022,
+      "step": 9746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011559195514034549,
+      "loss": 3.0381,
+      "step": 9747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001155084775103089,
+      "loss": 2.9354,
+      "step": 9748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011542502609727651,
+      "loss": 3.2188,
+      "step": 9749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011534160090693801,
+      "loss": 3.1016,
+      "step": 9750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001152582019449821,
+      "loss": 2.8366,
+      "step": 9751
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011517482921709538,
+      "loss": 2.6292,
+      "step": 9752
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011509148272896253,
+      "loss": 3.0877,
+      "step": 9753
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011500816248626711,
+      "loss": 2.9579,
+      "step": 9754
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011492486849468986,
+      "loss": 2.8906,
+      "step": 9755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011484160075991079,
+      "loss": 2.8448,
+      "step": 9756
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011475835928760709,
+      "loss": 2.7927,
+      "step": 9757
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011467514408345514,
+      "loss": 2.8745,
+      "step": 9758
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011459195515312864,
+      "loss": 2.8845,
+      "step": 9759
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011450879250230028,
+      "loss": 2.7637,
+      "step": 9760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011442565613664052,
+      "loss": 2.7559,
+      "step": 9761
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.000114342546061818,
+      "loss": 2.8187,
+      "step": 9762
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011425946228349976,
+      "loss": 2.9436,
+      "step": 9763
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011417640480735091,
+      "loss": 3.0723,
+      "step": 9764
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011409337363903488,
+      "loss": 3.2635,
+      "step": 9765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011401036878421322,
+      "loss": 2.8923,
+      "step": 9766
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001139273902485457,
+      "loss": 2.977,
+      "step": 9767
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011384443803769035,
+      "loss": 2.9993,
+      "step": 9768
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011376151215730312,
+      "loss": 3.22,
+      "step": 9769
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011367861261303897,
+      "loss": 3.0224,
+      "step": 9770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011359573941054985,
+      "loss": 3.0089,
+      "step": 9771
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001135128925554872,
+      "loss": 2.9364,
+      "step": 9772
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011343007205349937,
+      "loss": 2.7877,
+      "step": 9773
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001133472779102342,
+      "loss": 2.6415,
+      "step": 9774
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011326451013133648,
+      "loss": 3.1688,
+      "step": 9775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011318176872245035,
+      "loss": 2.7969,
+      "step": 9776
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011309905368921741,
+      "loss": 2.8226,
+      "step": 9777
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011301636503727775,
+      "loss": 2.901,
+      "step": 9778
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001129337027722695,
+      "loss": 3.1849,
+      "step": 9779
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011285106689982916,
+      "loss": 2.8279,
+      "step": 9780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011276845742559133,
+      "loss": 2.9486,
+      "step": 9781
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011268587435518885,
+      "loss": 2.8917,
+      "step": 9782
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011260331769425269,
+      "loss": 2.7473,
+      "step": 9783
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001125207874484121,
+      "loss": 2.8889,
+      "step": 9784
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001124382836232945,
+      "loss": 2.7921,
+      "step": 9785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011235580622452557,
+      "loss": 2.7918,
+      "step": 9786
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011227335525772902,
+      "loss": 2.8877,
+      "step": 9787
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011219093072852693,
+      "loss": 2.8149,
+      "step": 9788
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011210853264253951,
+      "loss": 2.8426,
+      "step": 9789
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011202616100538515,
+      "loss": 3.0154,
+      "step": 9790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011194381582268054,
+      "loss": 3.0147,
+      "step": 9791
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011186149710004017,
+      "loss": 2.9748,
+      "step": 9792
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011177920484307763,
+      "loss": 2.9411,
+      "step": 9793
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011169693905740347,
+      "loss": 2.6866,
+      "step": 9794
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011161469974862765,
+      "loss": 2.8711,
+      "step": 9795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011153248692235717,
+      "loss": 2.7309,
+      "step": 9796
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011145030058419825,
+      "loss": 2.8502,
+      "step": 9797
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011136814073975477,
+      "loss": 2.8226,
+      "step": 9798
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011128600739462885,
+      "loss": 2.7523,
+      "step": 9799
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011120390055442082,
+      "loss": 2.9597,
+      "step": 9800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011112182022472928,
+      "loss": 2.8896,
+      "step": 9801
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011103976641115094,
+      "loss": 2.7691,
+      "step": 9802
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001109577391192807,
+      "loss": 2.911,
+      "step": 9803
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011087573835471181,
+      "loss": 2.9039,
+      "step": 9804
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011079376412303548,
+      "loss": 2.8125,
+      "step": 9805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011071181642984124,
+      "loss": 2.8273,
+      "step": 9806
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011062989528071682,
+      "loss": 2.6572,
+      "step": 9807
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011054800068124793,
+      "loss": 3.0595,
+      "step": 9808
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011046613263701911,
+      "loss": 3.0188,
+      "step": 9809
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011038429115361204,
+      "loss": 2.9818,
+      "step": 9810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011030247623660771,
+      "loss": 2.9228,
+      "step": 9811
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001102206878915843,
+      "loss": 2.9555,
+      "step": 9812
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011013892612411897,
+      "loss": 2.8131,
+      "step": 9813
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00011005719093978667,
+      "loss": 3.059,
+      "step": 9814
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010997548234416055,
+      "loss": 2.8518,
+      "step": 9815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010989380034281215,
+      "loss": 2.8008,
+      "step": 9816
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010981214494131065,
+      "loss": 2.6355,
+      "step": 9817
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010973051614522439,
+      "loss": 2.8911,
+      "step": 9818
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010964891396011877,
+      "loss": 3.0218,
+      "step": 9819
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010956733839155836,
+      "loss": 2.8292,
+      "step": 9820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010948578944510529,
+      "loss": 2.7042,
+      "step": 9821
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010940426712632013,
+      "loss": 3.2165,
+      "step": 9822
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010932277144076153,
+      "loss": 2.8208,
+      "step": 9823
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010924130239398639,
+      "loss": 2.8046,
+      "step": 9824
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010915985999154976,
+      "loss": 2.8635,
+      "step": 9825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010907844423900486,
+      "loss": 3.0734,
+      "step": 9826
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010899705514190322,
+      "loss": 2.8584,
+      "step": 9827
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010891569270579437,
+      "loss": 3.1004,
+      "step": 9828
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010883435693622596,
+      "loss": 2.8988,
+      "step": 9829
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010875304783874441,
+      "loss": 2.8027,
+      "step": 9830
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010867176541889328,
+      "loss": 3.0176,
+      "step": 9831
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010859050968221552,
+      "loss": 2.9504,
+      "step": 9832
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010850928063425103,
+      "loss": 2.9051,
+      "step": 9833
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010842807828053913,
+      "loss": 2.7445,
+      "step": 9834
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010834690262661607,
+      "loss": 2.9829,
+      "step": 9835
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010826575367801738,
+      "loss": 2.9015,
+      "step": 9836
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010818463144027619,
+      "loss": 2.9673,
+      "step": 9837
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001081035359189238,
+      "loss": 2.8027,
+      "step": 9838
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010802246711948987,
+      "loss": 2.752,
+      "step": 9839
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010794142504750215,
+      "loss": 2.8105,
+      "step": 9840
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010786040970848665,
+      "loss": 2.8084,
+      "step": 9841
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010777942110796745,
+      "loss": 2.7936,
+      "step": 9842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010769845925146682,
+      "loss": 2.926,
+      "step": 9843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010761752414450531,
+      "loss": 2.9789,
+      "step": 9844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010753661579260138,
+      "loss": 2.832,
+      "step": 9845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010745573420127226,
+      "loss": 3.0172,
+      "step": 9846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010737487937603246,
+      "loss": 2.7955,
+      "step": 9847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010729405132239572,
+      "loss": 3.164,
+      "step": 9848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.0001072132500458729,
+      "loss": 3.1105,
+      "step": 9849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010713247555197375,
+      "loss": 2.7379,
+      "step": 9850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010705172784620593,
+      "loss": 2.9985,
+      "step": 9851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010697100693407518,
+      "loss": 2.8442,
+      "step": 9852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010689031282108603,
+      "loss": 2.9039,
+      "step": 9853
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010680964551274002,
+      "loss": 2.5587,
+      "step": 9854
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010672900501453825,
+      "loss": 2.9352,
+      "step": 9855
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010664839133197857,
+      "loss": 2.8525,
+      "step": 9856
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010656780447055825,
+      "loss": 3.0296,
+      "step": 9857
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010648724443577206,
+      "loss": 3.0934,
+      "step": 9858
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010640671123311302,
+      "loss": 2.7895,
+      "step": 9859
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010632620486807243,
+      "loss": 2.8907,
+      "step": 9860
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 0.00010624572534613969,
+      "loss": 2.9778,
+      "step": 9861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010616527267280234,
+      "loss": 3.1043,
+      "step": 9862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010608484685354624,
+      "loss": 2.9538,
+      "step": 9863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010600444789385527,
+      "loss": 2.8054,
+      "step": 9864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010592407579921154,
+      "loss": 2.7838,
+      "step": 9865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010584373057509528,
+      "loss": 3.0565,
+      "step": 9866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010576341222698494,
+      "loss": 2.8753,
+      "step": 9867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010568312076035697,
+      "loss": 2.8338,
+      "step": 9868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001056028561806866,
+      "loss": 2.8583,
+      "step": 9869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010552261849344613,
+      "loss": 3.0008,
+      "step": 9870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010544240770410729,
+      "loss": 2.9258,
+      "step": 9871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010536222381813881,
+      "loss": 3.1528,
+      "step": 9872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010528206684100844,
+      "loss": 3.2132,
+      "step": 9873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010520193677818169,
+      "loss": 2.7089,
+      "step": 9874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010512183363512234,
+      "loss": 3.1614,
+      "step": 9875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010504175741729232,
+      "loss": 2.9475,
+      "step": 9876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010496170813015171,
+      "loss": 2.6981,
+      "step": 9877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010488168577915875,
+      "loss": 2.8174,
+      "step": 9878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010480169036976983,
+      "loss": 2.866,
+      "step": 9879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010472172190743956,
+      "loss": 3.021,
+      "step": 9880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010464178039762073,
+      "loss": 2.8248,
+      "step": 9881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010456186584576416,
+      "loss": 3.0544,
+      "step": 9882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010448197825731897,
+      "loss": 2.8057,
+      "step": 9883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010440211763773238,
+      "loss": 2.8878,
+      "step": 9884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001043222839924498,
+      "loss": 3.0224,
+      "step": 9885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010424247732691472,
+      "loss": 2.9154,
+      "step": 9886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010416269764656894,
+      "loss": 3.0929,
+      "step": 9887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010408294495685223,
+      "loss": 2.7426,
+      "step": 9888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010400321926320255,
+      "loss": 2.917,
+      "step": 9889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010392352057105648,
+      "loss": 2.7936,
+      "step": 9890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001038438488858478,
+      "loss": 2.9514,
+      "step": 9891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010376420421300964,
+      "loss": 2.888,
+      "step": 9892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010368458655797203,
+      "loss": 2.8209,
+      "step": 9893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001036049959261644,
+      "loss": 2.7086,
+      "step": 9894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010352543232301315,
+      "loss": 2.9478,
+      "step": 9895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010344589575394381,
+      "loss": 2.8986,
+      "step": 9896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001033663862243796,
+      "loss": 2.7655,
+      "step": 9897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010328690373974192,
+      "loss": 2.892,
+      "step": 9898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010320744830545037,
+      "loss": 2.6969,
+      "step": 9899
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010312801992692278,
+      "loss": 2.9216,
+      "step": 9900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010304861860957499,
+      "loss": 2.8282,
+      "step": 9901
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001029692443588211,
+      "loss": 3.0159,
+      "step": 9902
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001028898971800733,
+      "loss": 2.9584,
+      "step": 9903
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010281057707874208,
+      "loss": 2.7896,
+      "step": 9904
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010273128406023574,
+      "loss": 3.0597,
+      "step": 9905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010265201812996139,
+      "loss": 2.8764,
+      "step": 9906
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010257277929332332,
+      "loss": 2.9297,
+      "step": 9907
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010249356755572508,
+      "loss": 2.7244,
+      "step": 9908
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010241438292256732,
+      "loss": 2.9498,
+      "step": 9909
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010233522539924978,
+      "loss": 2.621,
+      "step": 9910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010225609499116945,
+      "loss": 2.9761,
+      "step": 9911
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001021769917037223,
+      "loss": 3.0097,
+      "step": 9912
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010209791554230208,
+      "loss": 2.7973,
+      "step": 9913
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001020188665123003,
+      "loss": 2.7689,
+      "step": 9914
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010193984461910755,
+      "loss": 2.733,
+      "step": 9915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010186084986811145,
+      "loss": 3.1458,
+      "step": 9916
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001017818822646988,
+      "loss": 2.8564,
+      "step": 9917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010170294181425394,
+      "loss": 3.0627,
+      "step": 9918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010162402852215957,
+      "loss": 2.7435,
+      "step": 9919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010154514239379636,
+      "loss": 2.8604,
+      "step": 9920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001014662834345434,
+      "loss": 2.9775,
+      "step": 9921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001013874516497777,
+      "loss": 2.9853,
+      "step": 9922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010130864704487452,
+      "loss": 2.8697,
+      "step": 9923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010122986962520719,
+      "loss": 2.8645,
+      "step": 9924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010115111939614735,
+      "loss": 2.9485,
+      "step": 9925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010107239636306459,
+      "loss": 3.0014,
+      "step": 9926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010099370053132673,
+      "loss": 2.991,
+      "step": 9927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010091503190629964,
+      "loss": 3.1064,
+      "step": 9928
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.0001008363904933478,
+      "loss": 2.8625,
+      "step": 9929
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010075777629783301,
+      "loss": 2.9379,
+      "step": 9930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010067918932511611,
+      "loss": 2.9561,
+      "step": 9931
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010060062958055511,
+      "loss": 2.734,
+      "step": 9932
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010052209706950721,
+      "loss": 2.8747,
+      "step": 9933
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010044359179732698,
+      "loss": 2.8828,
+      "step": 9934
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010036511376936747,
+      "loss": 2.9571,
+      "step": 9935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010028666299097977,
+      "loss": 2.8258,
+      "step": 9936
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010020823946751317,
+      "loss": 3.0911,
+      "step": 9937
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010012984320431501,
+      "loss": 2.8523,
+      "step": 9938
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 0.00010005147420673088,
+      "loss": 2.8488,
+      "step": 9939
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.997313248010443e-05,
+      "loss": 2.6935,
+      "step": 9940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.989481802977752e-05,
+      "loss": 2.8377,
+      "step": 9941
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.981653086108994e-05,
+      "loss": 2.9269,
+      "step": 9942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.973827097938021e-05,
+      "loss": 2.9142,
+      "step": 9943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.966003838998405e-05,
+      "loss": 2.9388,
+      "step": 9944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.958183309823632e-05,
+      "loss": 2.8967,
+      "step": 9945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.950365510946918e-05,
+      "loss": 2.8921,
+      "step": 9946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.94255044290135e-05,
+      "loss": 3.0027,
+      "step": 9947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.934738106219793e-05,
+      "loss": 3.0652,
+      "step": 9948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.926928501434935e-05,
+      "loss": 2.7773,
+      "step": 9949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.91912162907932e-05,
+      "loss": 3.0089,
+      "step": 9950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.911317489685218e-05,
+      "loss": 2.9491,
+      "step": 9951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.903516083784814e-05,
+      "loss": 3.0055,
+      "step": 9952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.895717411910005e-05,
+      "loss": 2.9289,
+      "step": 9953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.887921474592609e-05,
+      "loss": 2.9532,
+      "step": 9954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.880128272364142e-05,
+      "loss": 2.6571,
+      "step": 9955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.872337805756032e-05,
+      "loss": 2.9883,
+      "step": 9956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.864550075299472e-05,
+      "loss": 2.8441,
+      "step": 9957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.856765081525476e-05,
+      "loss": 2.9906,
+      "step": 9958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.84898282496487e-05,
+      "loss": 3.1263,
+      "step": 9959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.841203306148305e-05,
+      "loss": 2.9258,
+      "step": 9960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.833426525606226e-05,
+      "loss": 3.0161,
+      "step": 9961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.825652483868907e-05,
+      "loss": 2.6537,
+      "step": 9962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.817881181466431e-05,
+      "loss": 3.0625,
+      "step": 9963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.81011261892869e-05,
+      "loss": 3.1176,
+      "step": 9964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.802346796785383e-05,
+      "loss": 2.8504,
+      "step": 9965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.794583715566068e-05,
+      "loss": 2.9452,
+      "step": 9966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.786823375800025e-05,
+      "loss": 2.7071,
+      "step": 9967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.779065778016466e-05,
+      "loss": 2.8512,
+      "step": 9968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.771310922744286e-05,
+      "loss": 2.9043,
+      "step": 9969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.763558810512319e-05,
+      "loss": 2.7701,
+      "step": 9970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.755809441849095e-05,
+      "loss": 2.8336,
+      "step": 9971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.748062817283055e-05,
+      "loss": 2.6482,
+      "step": 9972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.740318937342397e-05,
+      "loss": 2.9285,
+      "step": 9973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.732577802555153e-05,
+      "loss": 3.1216,
+      "step": 9974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.724839413449155e-05,
+      "loss": 3.0537,
+      "step": 9975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.717103770552055e-05,
+      "loss": 3.0048,
+      "step": 9976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.709370874391315e-05,
+      "loss": 2.8499,
+      "step": 9977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.701640725494215e-05,
+      "loss": 2.8711,
+      "step": 9978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.693913324387843e-05,
+      "loss": 2.6832,
+      "step": 9979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.686188671599106e-05,
+      "loss": 2.7052,
+      "step": 9980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.67846676765471e-05,
+      "loss": 2.9995,
+      "step": 9981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.670747613081183e-05,
+      "loss": 2.8354,
+      "step": 9982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.663031208404871e-05,
+      "loss": 3.1274,
+      "step": 9983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.655317554151921e-05,
+      "loss": 2.8688,
+      "step": 9984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.647606650848295e-05,
+      "loss": 3.1454,
+      "step": 9985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.639898499019778e-05,
+      "loss": 3.1197,
+      "step": 9986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.632193099191954e-05,
+      "loss": 2.9042,
+      "step": 9987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.624490451890217e-05,
+      "loss": 2.7706,
+      "step": 9988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.616790557639809e-05,
+      "loss": 2.8743,
+      "step": 9989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.609093416965714e-05,
+      "loss": 2.8761,
+      "step": 9990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.601399030392821e-05,
+      "loss": 3.0131,
+      "step": 9991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.593707398445727e-05,
+      "loss": 2.86,
+      "step": 9992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.586018521648932e-05,
+      "loss": 2.78,
+      "step": 9993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.578332400526701e-05,
+      "loss": 2.6179,
+      "step": 9994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.570649035603123e-05,
+      "loss": 2.7099,
+      "step": 9995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.562968427402097e-05,
+      "loss": 2.7352,
+      "step": 9996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.555290576447323e-05,
+      "loss": 3.0057,
+      "step": 9997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.54761548326234e-05,
+      "loss": 2.6881,
+      "step": 9998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.539943148370472e-05,
+      "loss": 2.8157,
+      "step": 9999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.532273572294876e-05,
+      "loss": 2.9539,
+      "step": 10000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.524606755558502e-05,
+      "loss": 2.9734,
+      "step": 10001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.516942698684112e-05,
+      "loss": 2.921,
+      "step": 10002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.509281402194325e-05,
+      "loss": 2.9344,
+      "step": 10003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.501622866611492e-05,
+      "loss": 2.8227,
+      "step": 10004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.493967092457856e-05,
+      "loss": 2.7333,
+      "step": 10005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.486314080255399e-05,
+      "loss": 2.9695,
+      "step": 10006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.478663830525991e-05,
+      "loss": 2.997,
+      "step": 10007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.471016343791227e-05,
+      "loss": 2.8833,
+      "step": 10008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.463371620572598e-05,
+      "loss": 2.6535,
+      "step": 10009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.455729661391355e-05,
+      "loss": 3.0599,
+      "step": 10010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.448090466768566e-05,
+      "loss": 2.8883,
+      "step": 10011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.440454037225149e-05,
+      "loss": 2.791,
+      "step": 10012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.432820373281748e-05,
+      "loss": 2.9382,
+      "step": 10013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.425189475458934e-05,
+      "loss": 2.9246,
+      "step": 10014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.41756134427697e-05,
+      "loss": 2.9917,
+      "step": 10015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.409935980256035e-05,
+      "loss": 2.7274,
+      "step": 10016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.402313383916056e-05,
+      "loss": 2.9179,
+      "step": 10017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.394693555776796e-05,
+      "loss": 2.6194,
+      "step": 10018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.387076496357804e-05,
+      "loss": 2.9364,
+      "step": 10019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.379462206178479e-05,
+      "loss": 2.7029,
+      "step": 10020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.371850685757998e-05,
+      "loss": 2.9611,
+      "step": 10021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.364241935615364e-05,
+      "loss": 2.8661,
+      "step": 10022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.356635956269394e-05,
+      "loss": 2.8539,
+      "step": 10023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.349032748238701e-05,
+      "loss": 2.8933,
+      "step": 10024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.341432312041708e-05,
+      "loss": 2.7876,
+      "step": 10025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.333834648196704e-05,
+      "loss": 2.8738,
+      "step": 10026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.326239757221689e-05,
+      "loss": 2.9085,
+      "step": 10027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.31864763963458e-05,
+      "loss": 3.1984,
+      "step": 10028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.311058295953012e-05,
+      "loss": 3.1396,
+      "step": 10029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.303471726694513e-05,
+      "loss": 2.9966,
+      "step": 10030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.295887932376329e-05,
+      "loss": 2.8032,
+      "step": 10031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.28830691351562e-05,
+      "loss": 2.7916,
+      "step": 10032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.280728670629291e-05,
+      "loss": 2.5453,
+      "step": 10033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.273153204234069e-05,
+      "loss": 3.2978,
+      "step": 10034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.265580514846495e-05,
+      "loss": 3.0397,
+      "step": 10035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.258010602982924e-05,
+      "loss": 2.8964,
+      "step": 10036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.250443469159525e-05,
+      "loss": 3.0641,
+      "step": 10037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.24287911389226e-05,
+      "loss": 2.8563,
+      "step": 10038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.235317537696924e-05,
+      "loss": 2.703,
+      "step": 10039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.22775874108911e-05,
+      "loss": 2.8141,
+      "step": 10040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.220202724584204e-05,
+      "loss": 2.8309,
+      "step": 10041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.212649488697461e-05,
+      "loss": 2.7443,
+      "step": 10042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.205099033943864e-05,
+      "loss": 3.0524,
+      "step": 10043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.197551360838302e-05,
+      "loss": 3.0619,
+      "step": 10044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.190006469895368e-05,
+      "loss": 2.8084,
+      "step": 10045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.18246436162955e-05,
+      "loss": 2.913,
+      "step": 10046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.174925036555104e-05,
+      "loss": 2.9361,
+      "step": 10047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.167388495186097e-05,
+      "loss": 2.9217,
+      "step": 10048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.15985473803646e-05,
+      "loss": 3.0792,
+      "step": 10049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.152323765619836e-05,
+      "loss": 2.8191,
+      "step": 10050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.144795578449783e-05,
+      "loss": 2.7922,
+      "step": 10051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.137270177039569e-05,
+      "loss": 2.7265,
+      "step": 10052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.129747561902363e-05,
+      "loss": 2.9236,
+      "step": 10053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.122227733551087e-05,
+      "loss": 2.6965,
+      "step": 10054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.11471069249849e-05,
+      "loss": 2.6438,
+      "step": 10055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.107196439257137e-05,
+      "loss": 3.1641,
+      "step": 10056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.099684974339389e-05,
+      "loss": 2.8521,
+      "step": 10057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.092176298257421e-05,
+      "loss": 2.7036,
+      "step": 10058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.084670411523232e-05,
+      "loss": 2.797,
+      "step": 10059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.07716731464861e-05,
+      "loss": 2.8692,
+      "step": 10060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.069667008145166e-05,
+      "loss": 2.9183,
+      "step": 10061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.062169492524303e-05,
+      "loss": 2.741,
+      "step": 10062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.054674768297289e-05,
+      "loss": 2.8971,
+      "step": 10063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.047182835975104e-05,
+      "loss": 2.9207,
+      "step": 10064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.039693696068657e-05,
+      "loss": 2.9392,
+      "step": 10065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.03220734908854e-05,
+      "loss": 2.697,
+      "step": 10066
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.02472379554527e-05,
+      "loss": 2.8031,
+      "step": 10067
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.01724303594908e-05,
+      "loss": 2.984,
+      "step": 10068
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.009765070810083e-05,
+      "loss": 2.9187,
+      "step": 10069
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.002289900638167e-05,
+      "loss": 2.8356,
+      "step": 10070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.994817525943033e-05,
+      "loss": 2.7758,
+      "step": 10071
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.987347947234193e-05,
+      "loss": 2.6847,
+      "step": 10072
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.97988116502097e-05,
+      "loss": 3.059,
+      "step": 10073
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.972417179812492e-05,
+      "loss": 2.7623,
+      "step": 10074
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.964955992117702e-05,
+      "loss": 3.1197,
+      "step": 10075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.957497602445353e-05,
+      "loss": 3.0557,
+      "step": 10076
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.950042011304e-05,
+      "loss": 2.8492,
+      "step": 10077
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.942589219202007e-05,
+      "loss": 3.0329,
+      "step": 10078
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.935139226647565e-05,
+      "loss": 3.0724,
+      "step": 10079
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.927692034148643e-05,
+      "loss": 2.8432,
+      "step": 10080
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.920247642213052e-05,
+      "loss": 2.8349,
+      "step": 10081
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.912806051348388e-05,
+      "loss": 2.9466,
+      "step": 10082
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.905367262062063e-05,
+      "loss": 3.0631,
+      "step": 10083
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.897931274861309e-05,
+      "loss": 3.0683,
+      "step": 10084
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.890498090253135e-05,
+      "loss": 2.7609,
+      "step": 10085
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.883067708744425e-05,
+      "loss": 2.8025,
+      "step": 10086
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.875640130841772e-05,
+      "loss": 2.8538,
+      "step": 10087
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.86821535705169e-05,
+      "loss": 2.7265,
+      "step": 10088
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.860793387880395e-05,
+      "loss": 2.9313,
+      "step": 10089
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.853374223834015e-05,
+      "loss": 2.9526,
+      "step": 10090
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.845957865418375e-05,
+      "loss": 2.9833,
+      "step": 10091
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.838544313139219e-05,
+      "loss": 3.0091,
+      "step": 10092
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.831133567502025e-05,
+      "loss": 3.0568,
+      "step": 10093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.823725629012103e-05,
+      "loss": 2.8078,
+      "step": 10094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.816320498174579e-05,
+      "loss": 2.8146,
+      "step": 10095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.80891817549438e-05,
+      "loss": 2.7071,
+      "step": 10096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.801518661476238e-05,
+      "loss": 3.0347,
+      "step": 10097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.794121956624701e-05,
+      "loss": 2.9708,
+      "step": 10098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.786728061444122e-05,
+      "loss": 3.0495,
+      "step": 10099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.779336976438657e-05,
+      "loss": 2.781,
+      "step": 10100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.771948702112264e-05,
+      "loss": 2.9656,
+      "step": 10101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.764563238968765e-05,
+      "loss": 2.67,
+      "step": 10102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.757180587511693e-05,
+      "loss": 2.8556,
+      "step": 10103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.74980074824449e-05,
+      "loss": 2.9078,
+      "step": 10104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.742423721670317e-05,
+      "loss": 2.9597,
+      "step": 10105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.735049508292226e-05,
+      "loss": 2.9964,
+      "step": 10106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.727678108612991e-05,
+      "loss": 2.7443,
+      "step": 10107
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.720309523135272e-05,
+      "loss": 2.7827,
+      "step": 10108
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.712943752361502e-05,
+      "loss": 2.898,
+      "step": 10109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.705580796793916e-05,
+      "loss": 2.9881,
+      "step": 10110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.698220656934586e-05,
+      "loss": 3.0449,
+      "step": 10111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.690863333285326e-05,
+      "loss": 3.1224,
+      "step": 10112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.683508826347841e-05,
+      "loss": 2.7687,
+      "step": 10113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.676157136623603e-05,
+      "loss": 2.7458,
+      "step": 10114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.668808264613887e-05,
+      "loss": 3.0343,
+      "step": 10115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.661462210819788e-05,
+      "loss": 2.7159,
+      "step": 10116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.65411897574221e-05,
+      "loss": 3.1517,
+      "step": 10117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.646778559881846e-05,
+      "loss": 3.1428,
+      "step": 10118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.639440963739226e-05,
+      "loss": 2.9162,
+      "step": 10119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.632106187814664e-05,
+      "loss": 2.7706,
+      "step": 10120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.624774232608296e-05,
+      "loss": 2.8796,
+      "step": 10121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.617445098620042e-05,
+      "loss": 2.7694,
+      "step": 10122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.610118786349692e-05,
+      "loss": 3.0291,
+      "step": 10123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.602795296296739e-05,
+      "loss": 2.7012,
+      "step": 10124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.595474628960598e-05,
+      "loss": 2.9849,
+      "step": 10125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.588156784840389e-05,
+      "loss": 3.1077,
+      "step": 10126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.580841764435144e-05,
+      "loss": 2.8515,
+      "step": 10127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.573529568243582e-05,
+      "loss": 3.0022,
+      "step": 10128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.566220196764341e-05,
+      "loss": 2.773,
+      "step": 10129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.558913650495804e-05,
+      "loss": 2.9225,
+      "step": 10130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.551609929936177e-05,
+      "loss": 3.0222,
+      "step": 10131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.54430903558347e-05,
+      "loss": 2.4797,
+      "step": 10132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.537010967935504e-05,
+      "loss": 2.837,
+      "step": 10133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.529715727489911e-05,
+      "loss": 2.8191,
+      "step": 10134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.522423314744121e-05,
+      "loss": 2.9547,
+      "step": 10135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.51513373019538e-05,
+      "loss": 2.688,
+      "step": 10136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.507846974340727e-05,
+      "loss": 3.0678,
+      "step": 10137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.500563047677017e-05,
+      "loss": 2.9572,
+      "step": 10138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.493281950700943e-05,
+      "loss": 2.6548,
+      "step": 10139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.486003683908927e-05,
+      "loss": 2.9114,
+      "step": 10140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.478728247797302e-05,
+      "loss": 2.8687,
+      "step": 10141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.471455642862091e-05,
+      "loss": 2.8796,
+      "step": 10142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.464185869599245e-05,
+      "loss": 2.7867,
+      "step": 10143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.456918928504415e-05,
+      "loss": 2.859,
+      "step": 10144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.449654820073116e-05,
+      "loss": 3.0329,
+      "step": 10145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.442393544800691e-05,
+      "loss": 3.144,
+      "step": 10146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.435135103182212e-05,
+      "loss": 2.8425,
+      "step": 10147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.427879495712648e-05,
+      "loss": 2.882,
+      "step": 10148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.420626722886687e-05,
+      "loss": 2.741,
+      "step": 10149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.413376785198917e-05,
+      "loss": 3.2978,
+      "step": 10150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.406129683143638e-05,
+      "loss": 2.8841,
+      "step": 10151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.398885417215024e-05,
+      "loss": 3.0751,
+      "step": 10152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.391643987907038e-05,
+      "loss": 2.9697,
+      "step": 10153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.384405395713441e-05,
+      "loss": 3.0608,
+      "step": 10154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.377169641127796e-05,
+      "loss": 2.9012,
+      "step": 10155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.369936724643495e-05,
+      "loss": 3.267,
+      "step": 10156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.36270664675371e-05,
+      "loss": 3.1416,
+      "step": 10157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.355479407951444e-05,
+      "loss": 2.6555,
+      "step": 10158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.348255008729483e-05,
+      "loss": 2.7475,
+      "step": 10159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.341033449580443e-05,
+      "loss": 2.6594,
+      "step": 10160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.333814730996714e-05,
+      "loss": 2.9111,
+      "step": 10161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.326598853470552e-05,
+      "loss": 2.9605,
+      "step": 10162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.31938581749393e-05,
+      "loss": 2.6662,
+      "step": 10163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.312175623558726e-05,
+      "loss": 2.8215,
+      "step": 10164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.304968272156527e-05,
+      "loss": 2.9784,
+      "step": 10165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.297763763778832e-05,
+      "loss": 3.0274,
+      "step": 10166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.290562098916826e-05,
+      "loss": 3.0385,
+      "step": 10167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.283363278061606e-05,
+      "loss": 2.9089,
+      "step": 10168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.276167301704019e-05,
+      "loss": 3.157,
+      "step": 10169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.268974170334736e-05,
+      "loss": 3.0274,
+      "step": 10170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.261783884444219e-05,
+      "loss": 2.8744,
+      "step": 10171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.254596444522755e-05,
+      "loss": 2.983,
+      "step": 10172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.247411851060427e-05,
+      "loss": 2.9411,
+      "step": 10173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.24023010454712e-05,
+      "loss": 2.8669,
+      "step": 10174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.23305120547253e-05,
+      "loss": 2.6972,
+      "step": 10175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.225875154326168e-05,
+      "loss": 2.9346,
+      "step": 10176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.218701951597329e-05,
+      "loss": 2.6328,
+      "step": 10177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.211531597775135e-05,
+      "loss": 2.9256,
+      "step": 10178
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.204364093348498e-05,
+      "loss": 2.9268,
+      "step": 10179
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.197199438806152e-05,
+      "loss": 2.7143,
+      "step": 10180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.190037634636615e-05,
+      "loss": 3.0744,
+      "step": 10181
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.18287868132822e-05,
+      "loss": 3.1597,
+      "step": 10182
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.17572257936915e-05,
+      "loss": 2.7527,
+      "step": 10183
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.168569329247288e-05,
+      "loss": 2.8986,
+      "step": 10184
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.161418931450443e-05,
+      "loss": 2.895,
+      "step": 10185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.154271386466127e-05,
+      "loss": 2.9746,
+      "step": 10186
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.147126694781754e-05,
+      "loss": 3.0822,
+      "step": 10187
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.139984856884435e-05,
+      "loss": 2.9484,
+      "step": 10188
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.13284587326119e-05,
+      "loss": 3.0287,
+      "step": 10189
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.125709744398785e-05,
+      "loss": 2.9264,
+      "step": 10190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.118576470783806e-05,
+      "loss": 2.7732,
+      "step": 10191
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.111446052902638e-05,
+      "loss": 2.5605,
+      "step": 10192
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.10431849124148e-05,
+      "loss": 2.9861,
+      "step": 10193
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.097193786286339e-05,
+      "loss": 3.1463,
+      "step": 10194
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.090071938523008e-05,
+      "loss": 3.0539,
+      "step": 10195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.082952948437111e-05,
+      "loss": 2.7988,
+      "step": 10196
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.07583681651406e-05,
+      "loss": 2.7622,
+      "step": 10197
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.068723543239065e-05,
+      "loss": 2.5503,
+      "step": 10198
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.06161312909719e-05,
+      "loss": 2.7779,
+      "step": 10199
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.054505574573218e-05,
+      "loss": 3.0435,
+      "step": 10200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.047400880151834e-05,
+      "loss": 2.7753,
+      "step": 10201
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.040299046317428e-05,
+      "loss": 2.9694,
+      "step": 10202
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.033200073554298e-05,
+      "loss": 3.1096,
+      "step": 10203
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.026103962346454e-05,
+      "loss": 2.8976,
+      "step": 10204
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.019010713177776e-05,
+      "loss": 2.6901,
+      "step": 10205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.011920326531925e-05,
+      "loss": 2.8694,
+      "step": 10206
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.004832802892365e-05,
+      "loss": 2.9869,
+      "step": 10207
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.997748142742372e-05,
+      "loss": 2.9932,
+      "step": 10208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.99066634656499e-05,
+      "loss": 2.8041,
+      "step": 10209
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.983587414843147e-05,
+      "loss": 2.8287,
+      "step": 10210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.976511348059484e-05,
+      "loss": 2.6677,
+      "step": 10211
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.969438146696529e-05,
+      "loss": 2.8463,
+      "step": 10212
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.962367811236549e-05,
+      "loss": 2.8817,
+      "step": 10213
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.955300342161659e-05,
+      "loss": 3.1361,
+      "step": 10214
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.948235739953757e-05,
+      "loss": 2.9162,
+      "step": 10215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.941174005094554e-05,
+      "loss": 2.8199,
+      "step": 10216
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.934115138065562e-05,
+      "loss": 2.9287,
+      "step": 10217
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.927059139348092e-05,
+      "loss": 2.8044,
+      "step": 10218
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.920006009423269e-05,
+      "loss": 2.8429,
+      "step": 10219
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.912955748772022e-05,
+      "loss": 2.8487,
+      "step": 10220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.905908357875064e-05,
+      "loss": 2.8728,
+      "step": 10221
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.898863837212972e-05,
+      "loss": 2.8193,
+      "step": 10222
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.891822187266035e-05,
+      "loss": 2.6562,
+      "step": 10223
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.88478340851444e-05,
+      "loss": 3.1138,
+      "step": 10224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.877747501438088e-05,
+      "loss": 2.8874,
+      "step": 10225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.870714466516782e-05,
+      "loss": 3.0401,
+      "step": 10226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.863684304230023e-05,
+      "loss": 2.757,
+      "step": 10227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.856657015057212e-05,
+      "loss": 2.8318,
+      "step": 10228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.849632599477501e-05,
+      "loss": 3.0334,
+      "step": 10229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.842611057969862e-05,
+      "loss": 3.1883,
+      "step": 10230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.835592391013052e-05,
+      "loss": 2.8264,
+      "step": 10231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.828576599085663e-05,
+      "loss": 2.9068,
+      "step": 10232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.821563682666072e-05,
+      "loss": 3.1544,
+      "step": 10233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.814553642232452e-05,
+      "loss": 3.1593,
+      "step": 10234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.807546478262806e-05,
+      "loss": 3.0419,
+      "step": 10235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.800542191234916e-05,
+      "loss": 2.7467,
+      "step": 10236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.793540781626362e-05,
+      "loss": 2.5895,
+      "step": 10237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.786542249914591e-05,
+      "loss": 2.7815,
+      "step": 10238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.779546596576753e-05,
+      "loss": 3.1464,
+      "step": 10239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.772553822089901e-05,
+      "loss": 3.0942,
+      "step": 10240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.765563926930808e-05,
+      "loss": 2.8818,
+      "step": 10241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.758576911576093e-05,
+      "loss": 2.9969,
+      "step": 10242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.75159277650221e-05,
+      "loss": 3.0347,
+      "step": 10243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.744611522185325e-05,
+      "loss": 2.6518,
+      "step": 10244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.737633149101514e-05,
+      "loss": 2.6253,
+      "step": 10245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.730657657726564e-05,
+      "loss": 3.0373,
+      "step": 10246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.72368504853615e-05,
+      "loss": 2.8892,
+      "step": 10247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.716715322005657e-05,
+      "loss": 2.8682,
+      "step": 10248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.709748478610367e-05,
+      "loss": 2.8847,
+      "step": 10249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.702784518825301e-05,
+      "loss": 3.0059,
+      "step": 10250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.695823443125315e-05,
+      "loss": 2.8897,
+      "step": 10251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.68886525198505e-05,
+      "loss": 3.0106,
+      "step": 10252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.681909945878962e-05,
+      "loss": 2.9521,
+      "step": 10253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.674957525281307e-05,
+      "loss": 2.8239,
+      "step": 10254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.668007990666142e-05,
+      "loss": 2.8513,
+      "step": 10255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.661061342507331e-05,
+      "loss": 2.7462,
+      "step": 10256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.654117581278542e-05,
+      "loss": 3.0765,
+      "step": 10257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.647176707453224e-05,
+      "loss": 2.6044,
+      "step": 10258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.640238721504688e-05,
+      "loss": 3.0125,
+      "step": 10259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.633303623905963e-05,
+      "loss": 2.6765,
+      "step": 10260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.626371415129967e-05,
+      "loss": 2.7789,
+      "step": 10261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.619442095649348e-05,
+      "loss": 2.7432,
+      "step": 10262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.612515665936626e-05,
+      "loss": 2.9479,
+      "step": 10263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.60559212646404e-05,
+      "loss": 3.0309,
+      "step": 10264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.598671477703717e-05,
+      "loss": 2.6982,
+      "step": 10265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.591753720127536e-05,
+      "loss": 2.9703,
+      "step": 10266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5848388542072e-05,
+      "loss": 3.1314,
+      "step": 10267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.577926880414198e-05,
+      "loss": 3.0064,
+      "step": 10268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.571017799219843e-05,
+      "loss": 3.0208,
+      "step": 10269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.56411161109522e-05,
+      "loss": 2.8852,
+      "step": 10270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.557208316511255e-05,
+      "loss": 2.9355,
+      "step": 10271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.550307915938643e-05,
+      "loss": 3.1077,
+      "step": 10272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.543410409847906e-05,
+      "loss": 3.2326,
+      "step": 10273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.536515798709348e-05,
+      "loss": 2.8012,
+      "step": 10274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.529624082993097e-05,
+      "loss": 2.8681,
+      "step": 10275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.522735263169062e-05,
+      "loss": 2.9803,
+      "step": 10276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.515849339706976e-05,
+      "loss": 2.5812,
+      "step": 10277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.50896631307636e-05,
+      "loss": 3.1208,
+      "step": 10278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.502086183746538e-05,
+      "loss": 2.851,
+      "step": 10279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.495208952186638e-05,
+      "loss": 2.8826,
+      "step": 10280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.488334618865583e-05,
+      "loss": 2.6629,
+      "step": 10281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.481463184252147e-05,
+      "loss": 2.8052,
+      "step": 10282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.47459464881482e-05,
+      "loss": 2.916,
+      "step": 10283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.467729013021979e-05,
+      "loss": 2.8714,
+      "step": 10284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.460866277341727e-05,
+      "loss": 2.6124,
+      "step": 10285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.454006442242051e-05,
+      "loss": 2.9317,
+      "step": 10286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.447149508190649e-05,
+      "loss": 3.1104,
+      "step": 10287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.440295475655101e-05,
+      "loss": 2.7537,
+      "step": 10288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.433444345102758e-05,
+      "loss": 2.8705,
+      "step": 10289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.426596117000761e-05,
+      "loss": 2.6607,
+      "step": 10290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.419750791816072e-05,
+      "loss": 2.8487,
+      "step": 10291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.412908370015436e-05,
+      "loss": 2.8774,
+      "step": 10292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.406068852065423e-05,
+      "loss": 2.9536,
+      "step": 10293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.399232238432391e-05,
+      "loss": 2.7598,
+      "step": 10294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.392398529582501e-05,
+      "loss": 2.7509,
+      "step": 10295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.385567725981712e-05,
+      "loss": 2.9519,
+      "step": 10296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.378739828095793e-05,
+      "loss": 2.8395,
+      "step": 10297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.37191483639033e-05,
+      "loss": 2.774,
+      "step": 10298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.36509275133066e-05,
+      "loss": 2.9638,
+      "step": 10299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.358273573381991e-05,
+      "loss": 2.9435,
+      "step": 10300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.351457303009256e-05,
+      "loss": 2.7589,
+      "step": 10301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.34464394067727e-05,
+      "loss": 2.9546,
+      "step": 10302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.337833486850592e-05,
+      "loss": 2.6707,
+      "step": 10303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.331025941993596e-05,
+      "loss": 2.8912,
+      "step": 10304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.324221306570472e-05,
+      "loss": 2.8717,
+      "step": 10305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.317419581045199e-05,
+      "loss": 2.8738,
+      "step": 10306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.310620765881565e-05,
+      "loss": 2.9576,
+      "step": 10307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.303824861543134e-05,
+      "loss": 2.8239,
+      "step": 10308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.297031868493314e-05,
+      "loss": 2.7753,
+      "step": 10309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.290241787195296e-05,
+      "loss": 2.9914,
+      "step": 10310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.283454618112057e-05,
+      "loss": 2.8647,
+      "step": 10311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.276670361706389e-05,
+      "loss": 3.1629,
+      "step": 10312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.26988901844089e-05,
+      "loss": 2.982,
+      "step": 10313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.263110588777949e-05,
+      "loss": 2.9963,
+      "step": 10314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.25633507317977e-05,
+      "loss": 2.9025,
+      "step": 10315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.24956247210834e-05,
+      "loss": 3.0156,
+      "step": 10316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.242792786025459e-05,
+      "loss": 2.807,
+      "step": 10317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.236026015392716e-05,
+      "loss": 2.6028,
+      "step": 10318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.229262160671546e-05,
+      "loss": 3.0361,
+      "step": 10319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.222501222323102e-05,
+      "loss": 2.9641,
+      "step": 10320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.215743200808433e-05,
+      "loss": 2.9365,
+      "step": 10321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.208988096588299e-05,
+      "loss": 2.7384,
+      "step": 10322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.20223591012335e-05,
+      "loss": 2.6333,
+      "step": 10323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.195486641873944e-05,
+      "loss": 3.16,
+      "step": 10324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.188740292300321e-05,
+      "loss": 3.0142,
+      "step": 10325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.181996861862483e-05,
+      "loss": 3.0599,
+      "step": 10326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.175256351020237e-05,
+      "loss": 2.914,
+      "step": 10327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.168518760233184e-05,
+      "loss": 2.8713,
+      "step": 10328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.161784089960744e-05,
+      "loss": 2.8769,
+      "step": 10329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.155052340662127e-05,
+      "loss": 2.8077,
+      "step": 10330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.148323512796345e-05,
+      "loss": 3.0367,
+      "step": 10331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.141597606822204e-05,
+      "loss": 2.9653,
+      "step": 10332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.134874623198329e-05,
+      "loss": 2.7449,
+      "step": 10333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.128154562383116e-05,
+      "loss": 2.8356,
+      "step": 10334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.121437424834815e-05,
+      "loss": 2.9383,
+      "step": 10335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.114723211011398e-05,
+      "loss": 2.9374,
+      "step": 10336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.108011921370727e-05,
+      "loss": 2.8678,
+      "step": 10337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.10130355637037e-05,
+      "loss": 3.0549,
+      "step": 10338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.094598116467793e-05,
+      "loss": 3.0,
+      "step": 10339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.087895602120181e-05,
+      "loss": 2.8431,
+      "step": 10340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.081196013784558e-05,
+      "loss": 2.6829,
+      "step": 10341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.074499351917763e-05,
+      "loss": 2.7236,
+      "step": 10342
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.067805616976386e-05,
+      "loss": 2.9906,
+      "step": 10343
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.06111480941688e-05,
+      "loss": 2.5618,
+      "step": 10344
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.054426929695429e-05,
+      "loss": 2.9146,
+      "step": 10345
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.04774197826809e-05,
+      "loss": 2.9416,
+      "step": 10346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.041059955590651e-05,
+      "loss": 2.9119,
+      "step": 10347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.034380862118756e-05,
+      "loss": 2.881,
+      "step": 10348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.027704698307824e-05,
+      "loss": 3.0629,
+      "step": 10349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.02103146461307e-05,
+      "loss": 2.7816,
+      "step": 10350
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.014361161489524e-05,
+      "loss": 2.9762,
+      "step": 10351
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.007693789392005e-05,
+      "loss": 2.807,
+      "step": 10352
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.001029348775129e-05,
+      "loss": 2.9488,
+      "step": 10353
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.99436784009333e-05,
+      "loss": 2.957,
+      "step": 10354
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.987709263800824e-05,
+      "loss": 3.0574,
+      "step": 10355
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.981053620351629e-05,
+      "loss": 2.8563,
+      "step": 10356
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.974400910199569e-05,
+      "loss": 2.9659,
+      "step": 10357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.967751133798295e-05,
+      "loss": 2.9313,
+      "step": 10358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.961104291601178e-05,
+      "loss": 2.7962,
+      "step": 10359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.954460384061494e-05,
+      "loss": 2.8709,
+      "step": 10360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.947819411632222e-05,
+      "loss": 2.8983,
+      "step": 10361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.941181374766214e-05,
+      "loss": 2.9879,
+      "step": 10362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.934546273916076e-05,
+      "loss": 2.8844,
+      "step": 10363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.927914109534239e-05,
+      "loss": 3.1615,
+      "step": 10364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.92128488207292e-05,
+      "loss": 2.8667,
+      "step": 10365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.914658591984146e-05,
+      "loss": 2.8817,
+      "step": 10366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.908035239719734e-05,
+      "loss": 3.0714,
+      "step": 10367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9014148257313e-05,
+      "loss": 2.8556,
+      "step": 10368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.894797350470272e-05,
+      "loss": 3.0257,
+      "step": 10369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.888182814387872e-05,
+      "loss": 2.8367,
+      "step": 10370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.881571217935117e-05,
+      "loss": 3.1946,
+      "step": 10371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.874962561562819e-05,
+      "loss": 3.0213,
+      "step": 10372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.868356845721607e-05,
+      "loss": 3.008,
+      "step": 10373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.861754070861898e-05,
+      "loss": 2.9939,
+      "step": 10374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.855154237433909e-05,
+      "loss": 2.9017,
+      "step": 10375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.848557345887651e-05,
+      "loss": 2.9053,
+      "step": 10376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.84196339667295e-05,
+      "loss": 2.876,
+      "step": 10377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.835372390239408e-05,
+      "loss": 2.6537,
+      "step": 10378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.828784327036469e-05,
+      "loss": 2.8039,
+      "step": 10379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.822199207513313e-05,
+      "loss": 2.9458,
+      "step": 10380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.815617032118987e-05,
+      "loss": 2.9515,
+      "step": 10381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.80903780130227e-05,
+      "loss": 2.8889,
+      "step": 10382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.802461515511815e-05,
+      "loss": 3.0072,
+      "step": 10383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.795888175195986e-05,
+      "loss": 3.0747,
+      "step": 10384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.789317780803034e-05,
+      "loss": 2.9961,
+      "step": 10385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.782750332780951e-05,
+      "loss": 2.5878,
+      "step": 10386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.776185831577547e-05,
+      "loss": 2.6199,
+      "step": 10387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.769624277640435e-05,
+      "loss": 3.0169,
+      "step": 10388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.763065671417024e-05,
+      "loss": 2.9389,
+      "step": 10389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.756510013354511e-05,
+      "loss": 3.192,
+      "step": 10390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.749957303899906e-05,
+      "loss": 2.881,
+      "step": 10391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.743407543500013e-05,
+      "loss": 3.0378,
+      "step": 10392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.736860732601436e-05,
+      "loss": 3.0099,
+      "step": 10393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.730316871650565e-05,
+      "loss": 2.9429,
+      "step": 10394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.723775961093636e-05,
+      "loss": 2.7165,
+      "step": 10395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.717238001376602e-05,
+      "loss": 2.9772,
+      "step": 10396
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.710702992945306e-05,
+      "loss": 2.786,
+      "step": 10397
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.704170936245302e-05,
+      "loss": 2.9273,
+      "step": 10398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.697641831722029e-05,
+      "loss": 2.9728,
+      "step": 10399
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.691115679820636e-05,
+      "loss": 2.8457,
+      "step": 10400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.684592480986157e-05,
+      "loss": 2.7582,
+      "step": 10401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.678072235663362e-05,
+      "loss": 2.9979,
+      "step": 10402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.671554944296849e-05,
+      "loss": 2.7764,
+      "step": 10403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.665040607331002e-05,
+      "loss": 2.7798,
+      "step": 10404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.658529225210008e-05,
+      "loss": 3.0537,
+      "step": 10405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.652020798377873e-05,
+      "loss": 3.0607,
+      "step": 10406
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.645515327278345e-05,
+      "loss": 2.98,
+      "step": 10407
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.639012812355039e-05,
+      "loss": 2.7085,
+      "step": 10408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.63251325405132e-05,
+      "loss": 3.016,
+      "step": 10409
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.626016652810374e-05,
+      "loss": 3.0715,
+      "step": 10410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.619523009075179e-05,
+      "loss": 3.0113,
+      "step": 10411
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.613032323288515e-05,
+      "loss": 2.85,
+      "step": 10412
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.606544595892949e-05,
+      "loss": 2.6555,
+      "step": 10413
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.600059827330863e-05,
+      "loss": 2.741,
+      "step": 10414
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.593578018044416e-05,
+      "loss": 2.7387,
+      "step": 10415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.587099168475591e-05,
+      "loss": 2.9644,
+      "step": 10416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.58062327906614e-05,
+      "loss": 2.8785,
+      "step": 10417
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.574150350257663e-05,
+      "loss": 2.8876,
+      "step": 10418
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.567680382491475e-05,
+      "loss": 2.9866,
+      "step": 10419
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.561213376208785e-05,
+      "loss": 2.7654,
+      "step": 10420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.554749331850512e-05,
+      "loss": 3.0866,
+      "step": 10421
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.548288249857448e-05,
+      "loss": 2.833,
+      "step": 10422
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.541830130670134e-05,
+      "loss": 2.9658,
+      "step": 10423
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.53537497472893e-05,
+      "loss": 2.9343,
+      "step": 10424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.528922782473985e-05,
+      "loss": 2.8613,
+      "step": 10425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.522473554345248e-05,
+      "loss": 3.0797,
+      "step": 10426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.516027290782467e-05,
+      "loss": 2.9636,
+      "step": 10427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.509583992225193e-05,
+      "loss": 3.2775,
+      "step": 10428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.503143659112765e-05,
+      "loss": 2.7319,
+      "step": 10429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.496706291884325e-05,
+      "loss": 2.6451,
+      "step": 10430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.490271890978816e-05,
+      "loss": 2.8915,
+      "step": 10431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.483840456834972e-05,
+      "loss": 3.2095,
+      "step": 10432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.477411989891313e-05,
+      "loss": 2.914,
+      "step": 10433
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.470986490586217e-05,
+      "loss": 3.1224,
+      "step": 10434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.464563959357755e-05,
+      "loss": 3.0007,
+      "step": 10435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.458144396643912e-05,
+      "loss": 2.9553,
+      "step": 10436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.451727802882356e-05,
+      "loss": 2.9519,
+      "step": 10437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.445314178510658e-05,
+      "loss": 2.8066,
+      "step": 10438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.438903523966127e-05,
+      "loss": 3.2764,
+      "step": 10439
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.432495839685854e-05,
+      "loss": 2.7099,
+      "step": 10440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.426091126106792e-05,
+      "loss": 2.679,
+      "step": 10441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.419689383665617e-05,
+      "loss": 3.0158,
+      "step": 10442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.413290612798883e-05,
+      "loss": 2.829,
+      "step": 10443
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.406894813942848e-05,
+      "loss": 2.7353,
+      "step": 10444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.400501987533652e-05,
+      "loss": 3.0227,
+      "step": 10445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.39411213400719e-05,
+      "loss": 3.0263,
+      "step": 10446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.387725253799159e-05,
+      "loss": 2.9253,
+      "step": 10447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.381341347345054e-05,
+      "loss": 2.9124,
+      "step": 10448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.374960415080172e-05,
+      "loss": 2.836,
+      "step": 10449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.368582457439604e-05,
+      "loss": 2.9461,
+      "step": 10450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.362207474858239e-05,
+      "loss": 2.8784,
+      "step": 10451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.35583546777076e-05,
+      "loss": 3.0287,
+      "step": 10452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.349466436611651e-05,
+      "loss": 2.9457,
+      "step": 10453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.34310038181518e-05,
+      "loss": 2.8921,
+      "step": 10454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.33673730381546e-05,
+      "loss": 2.5986,
+      "step": 10455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.330377203046323e-05,
+      "loss": 3.2178,
+      "step": 10456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.324020079941473e-05,
+      "loss": 2.8911,
+      "step": 10457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.317665934934341e-05,
+      "loss": 2.9965,
+      "step": 10458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.311314768458242e-05,
+      "loss": 2.7566,
+      "step": 10459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.30496658094618e-05,
+      "loss": 3.1418,
+      "step": 10460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.29862137283106e-05,
+      "loss": 2.7124,
+      "step": 10461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.292279144545516e-05,
+      "loss": 3.1447,
+      "step": 10462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.285939896522002e-05,
+      "loss": 3.0506,
+      "step": 10463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.279603629192776e-05,
+      "loss": 2.828,
+      "step": 10464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.273270342989873e-05,
+      "loss": 2.9537,
+      "step": 10465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.266940038345143e-05,
+      "loss": 3.0074,
+      "step": 10466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.260612715690223e-05,
+      "loss": 2.6522,
+      "step": 10467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.254288375456546e-05,
+      "loss": 3.1047,
+      "step": 10468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.247967018075346e-05,
+      "loss": 2.9863,
+      "step": 10469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.241648643977649e-05,
+      "loss": 3.0958,
+      "step": 10470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.235333253594289e-05,
+      "loss": 2.863,
+      "step": 10471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.229020847355881e-05,
+      "loss": 2.9902,
+      "step": 10472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.22271142569285e-05,
+      "loss": 2.9062,
+      "step": 10473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.216404989035406e-05,
+      "loss": 3.121,
+      "step": 10474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.210101537813562e-05,
+      "loss": 2.8348,
+      "step": 10475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.203801072457127e-05,
+      "loss": 2.6199,
+      "step": 10476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.197503593395693e-05,
+      "loss": 3.1137,
+      "step": 10477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.191209101058698e-05,
+      "loss": 3.204,
+      "step": 10478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.18491759587529e-05,
+      "loss": 3.0684,
+      "step": 10479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.178629078274511e-05,
+      "loss": 2.819,
+      "step": 10480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.172343548685105e-05,
+      "loss": 2.7782,
+      "step": 10481
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.16606100753569e-05,
+      "loss": 2.8569,
+      "step": 10482
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.159781455254643e-05,
+      "loss": 2.891,
+      "step": 10483
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.153504892270134e-05,
+      "loss": 2.9112,
+      "step": 10484
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.147231319010139e-05,
+      "loss": 3.1222,
+      "step": 10485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.14096073590244e-05,
+      "loss": 2.9886,
+      "step": 10486
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.134693143374587e-05,
+      "loss": 3.021,
+      "step": 10487
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.128428541853958e-05,
+      "loss": 2.682,
+      "step": 10488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.122166931767709e-05,
+      "loss": 2.8479,
+      "step": 10489
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.115908313542784e-05,
+      "loss": 3.1148,
+      "step": 10490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.109652687605949e-05,
+      "loss": 2.8729,
+      "step": 10491
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1034000543837455e-05,
+      "loss": 2.9943,
+      "step": 10492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0971504143025045e-05,
+      "loss": 2.7776,
+      "step": 10493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0909037677883975e-05,
+      "loss": 2.7634,
+      "step": 10494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.084660115267315e-05,
+      "loss": 2.942,
+      "step": 10495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0784194571650356e-05,
+      "loss": 3.0881,
+      "step": 10496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.072181793907039e-05,
+      "loss": 2.7603,
+      "step": 10497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.065947125918681e-05,
+      "loss": 2.7946,
+      "step": 10498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.059715453625064e-05,
+      "loss": 2.7931,
+      "step": 10499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0534867774511116e-05,
+      "loss": 2.9752,
+      "step": 10500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.047261097821527e-05,
+      "loss": 2.6379,
+      "step": 10501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0410384151608154e-05,
+      "loss": 3.018,
+      "step": 10502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0348187298932934e-05,
+      "loss": 2.9252,
+      "step": 10503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.028602042443021e-05,
+      "loss": 2.6501,
+      "step": 10504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0223883532339206e-05,
+      "loss": 2.7395,
+      "step": 10505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.016177662689676e-05,
+      "loss": 3.0881,
+      "step": 10506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.009969971233764e-05,
+      "loss": 2.699,
+      "step": 10507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.003765279289464e-05,
+      "loss": 2.9535,
+      "step": 10508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.997563587279853e-05,
+      "loss": 2.8164,
+      "step": 10509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.991364895627799e-05,
+      "loss": 2.7862,
+      "step": 10510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.985169204755964e-05,
+      "loss": 2.8814,
+      "step": 10511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.978976515086815e-05,
+      "loss": 2.8364,
+      "step": 10512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9727868270426036e-05,
+      "loss": 2.6828,
+      "step": 10513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.966600141045375e-05,
+      "loss": 3.009,
+      "step": 10514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.960416457516998e-05,
+      "loss": 2.8757,
+      "step": 10515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.95423577687908e-05,
+      "loss": 2.9912,
+      "step": 10516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9480580995530996e-05,
+      "loss": 3.0239,
+      "step": 10517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9418834259602426e-05,
+      "loss": 3.0289,
+      "step": 10518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.935711756521584e-05,
+      "loss": 2.7951,
+      "step": 10519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9295430916578964e-05,
+      "loss": 2.7729,
+      "step": 10520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9233774317898334e-05,
+      "loss": 2.9847,
+      "step": 10521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9172147773378036e-05,
+      "loss": 2.834,
+      "step": 10522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.911055128722004e-05,
+      "loss": 2.8767,
+      "step": 10523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9048984863624425e-05,
+      "loss": 2.7221,
+      "step": 10524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8987448506789165e-05,
+      "loss": 2.849,
+      "step": 10525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.892594222091019e-05,
+      "loss": 2.9335,
+      "step": 10526
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.886446601018136e-05,
+      "loss": 3.0607,
+      "step": 10527
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.880301987879449e-05,
+      "loss": 3.2644,
+      "step": 10528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.874160383093946e-05,
+      "loss": 2.7288,
+      "step": 10529
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8680217870803796e-05,
+      "loss": 2.9405,
+      "step": 10530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8618862002573494e-05,
+      "loss": 2.7586,
+      "step": 10531
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.855753623043175e-05,
+      "loss": 2.9396,
+      "step": 10532
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.849624055856056e-05,
+      "loss": 2.8466,
+      "step": 10533
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.843497499113909e-05,
+      "loss": 2.958,
+      "step": 10534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8373739532345096e-05,
+      "loss": 2.8772,
+      "step": 10535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8312534186353795e-05,
+      "loss": 2.865,
+      "step": 10536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8251358957338464e-05,
+      "loss": 2.9087,
+      "step": 10537
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.819021384947076e-05,
+      "loss": 2.8144,
+      "step": 10538
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8129098866919526e-05,
+      "loss": 2.7814,
+      "step": 10539
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.806801401385231e-05,
+      "loss": 2.8889,
+      "step": 10540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.800695929443395e-05,
+      "loss": 3.0814,
+      "step": 10541
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.794593471282772e-05,
+      "loss": 2.8305,
+      "step": 10542
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.788494027319463e-05,
+      "loss": 2.9719,
+      "step": 10543
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7823975979693634e-05,
+      "loss": 3.2775,
+      "step": 10544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7763041836481676e-05,
+      "loss": 2.8118,
+      "step": 10545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.770213784771361e-05,
+      "loss": 3.0453,
+      "step": 10546
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7641264017542216e-05,
+      "loss": 3.0849,
+      "step": 10547
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.758042035011823e-05,
+      "loss": 2.9032,
+      "step": 10548
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.751960684959046e-05,
+      "loss": 2.9423,
+      "step": 10549
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.74588235201054e-05,
+      "loss": 3.062,
+      "step": 10550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.739807036580774e-05,
+      "loss": 2.8121,
+      "step": 10551
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7337347390839954e-05,
+      "loss": 2.8319,
+      "step": 10552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.727665459934245e-05,
+      "loss": 2.7688,
+      "step": 10553
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.721599199545391e-05,
+      "loss": 2.87,
+      "step": 10554
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.715535958331025e-05,
+      "loss": 2.8237,
+      "step": 10555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.709475736704623e-05,
+      "loss": 2.8238,
+      "step": 10556
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7034185350793636e-05,
+      "loss": 2.953,
+      "step": 10557
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.697364353868295e-05,
+      "loss": 2.9458,
+      "step": 10558
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.691313193484221e-05,
+      "loss": 2.8592,
+      "step": 10559
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.685265054339744e-05,
+      "loss": 2.9646,
+      "step": 10560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.679219936847269e-05,
+      "loss": 2.9151,
+      "step": 10561
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6731778414189805e-05,
+      "loss": 2.6379,
+      "step": 10562
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.667138768466873e-05,
+      "loss": 3.0149,
+      "step": 10563
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.661102718402722e-05,
+      "loss": 2.8779,
+      "step": 10564
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6550696916381096e-05,
+      "loss": 2.877,
+      "step": 10565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6490396885843956e-05,
+      "loss": 3.1622,
+      "step": 10566
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6430127096527564e-05,
+      "loss": 2.6902,
+      "step": 10567
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.636988755254135e-05,
+      "loss": 2.896,
+      "step": 10568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.630967825799288e-05,
+      "loss": 2.924,
+      "step": 10569
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6249499216987576e-05,
+      "loss": 2.9365,
+      "step": 10570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6189350433628825e-05,
+      "loss": 3.0532,
+      "step": 10571
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.612923191201796e-05,
+      "loss": 2.6472,
+      "step": 10572
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6069143656254205e-05,
+      "loss": 2.8764,
+      "step": 10573
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6009085670434666e-05,
+      "loss": 3.0102,
+      "step": 10574
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.594905795865479e-05,
+      "loss": 3.0185,
+      "step": 10575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.588906052500719e-05,
+      "loss": 2.8516,
+      "step": 10576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.582909337358327e-05,
+      "loss": 2.811,
+      "step": 10577
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.576915650847159e-05,
+      "loss": 2.9938,
+      "step": 10578
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.570924993375936e-05,
+      "loss": 2.8227,
+      "step": 10579
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.564937365353101e-05,
+      "loss": 2.9268,
+      "step": 10580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.558952767186959e-05,
+      "loss": 2.8163,
+      "step": 10581
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.552971199285567e-05,
+      "loss": 2.8012,
+      "step": 10582
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5469926620567827e-05,
+      "loss": 2.8446,
+      "step": 10583
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.541017155908262e-05,
+      "loss": 2.7933,
+      "step": 10584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.535044681247453e-05,
+      "loss": 2.8918,
+      "step": 10585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.52907523848159e-05,
+      "loss": 2.6028,
+      "step": 10586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.523108828017709e-05,
+      "loss": 2.7334,
+      "step": 10587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5171454502626396e-05,
+      "loss": 2.9507,
+      "step": 10588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5111851056230014e-05,
+      "loss": 2.8172,
+      "step": 10589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.505227794505191e-05,
+      "loss": 2.8161,
+      "step": 10590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.499273517315451e-05,
+      "loss": 3.0073,
+      "step": 10591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.493322274459739e-05,
+      "loss": 2.8413,
+      "step": 10592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.487374066343892e-05,
+      "loss": 3.1069,
+      "step": 10593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.481428893373447e-05,
+      "loss": 2.9558,
+      "step": 10594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.475486755953824e-05,
+      "loss": 2.9002,
+      "step": 10595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.469547654490165e-05,
+      "loss": 2.8364,
+      "step": 10596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.463611589387452e-05,
+      "loss": 3.0881,
+      "step": 10597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.457678561050433e-05,
+      "loss": 2.8456,
+      "step": 10598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4517485698836686e-05,
+      "loss": 2.861,
+      "step": 10599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4458216162914966e-05,
+      "loss": 2.8964,
+      "step": 10600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.439897700678048e-05,
+      "loss": 2.8733,
+      "step": 10601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.433976823447262e-05,
+      "loss": 2.908,
+      "step": 10602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.428058985002854e-05,
+      "loss": 3.0129,
+      "step": 10603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.42214418574834e-05,
+      "loss": 2.7607,
+      "step": 10604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.416232426087026e-05,
+      "loss": 2.9379,
+      "step": 10605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.4103237064220156e-05,
+      "loss": 3.1619,
+      "step": 10606
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.4044180271561994e-05,
+      "loss": 2.7553,
+      "step": 10607
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3985153886922654e-05,
+      "loss": 2.7068,
+      "step": 10608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.392615791432687e-05,
+      "loss": 2.7512,
+      "step": 10609
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.386719235779741e-05,
+      "loss": 2.9038,
+      "step": 10610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3808257221354903e-05,
+      "loss": 2.7183,
+      "step": 10611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3749352509017856e-05,
+      "loss": 2.8475,
+      "step": 10612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.369047822480272e-05,
+      "loss": 3.0518,
+      "step": 10613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.363163437272411e-05,
+      "loss": 3.0919,
+      "step": 10614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.357282095679411e-05,
+      "loss": 2.7129,
+      "step": 10615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.351403798102328e-05,
+      "loss": 3.0256,
+      "step": 10616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3455285449419414e-05,
+      "loss": 2.8428,
+      "step": 10617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339656336598892e-05,
+      "loss": 2.8977,
+      "step": 10618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3337871734735756e-05,
+      "loss": 2.8518,
+      "step": 10619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.327921055966189e-05,
+      "loss": 3.0013,
+      "step": 10620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3220579844767194e-05,
+      "loss": 2.6794,
+      "step": 10621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.316197959404939e-05,
+      "loss": 2.9859,
+      "step": 10622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3103409811504355e-05,
+      "loss": 2.892,
+      "step": 10623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3044870501125605e-05,
+      "loss": 2.671,
+      "step": 10624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.298636166690479e-05,
+      "loss": 2.8544,
+      "step": 10625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2927883312831384e-05,
+      "loss": 2.9812,
+      "step": 10626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.28694354428928e-05,
+      "loss": 2.9669,
+      "step": 10627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.281101806107436e-05,
+      "loss": 2.8205,
+      "step": 10628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.27526311713592e-05,
+      "loss": 2.7909,
+      "step": 10629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.269427477772887e-05,
+      "loss": 2.7705,
+      "step": 10630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.263594888416201e-05,
+      "loss": 2.8473,
+      "step": 10631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.257765349463611e-05,
+      "loss": 2.9965,
+      "step": 10632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.251938861312561e-05,
+      "loss": 2.8719,
+      "step": 10633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.246115424360371e-05,
+      "loss": 2.9718,
+      "step": 10634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2402950390041226e-05,
+      "loss": 3.0758,
+      "step": 10635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2344777056406503e-05,
+      "loss": 3.1493,
+      "step": 10636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2286634246666573e-05,
+      "loss": 3.1169,
+      "step": 10637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.222852196478561e-05,
+      "loss": 2.9777,
+      "step": 10638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.217044021472639e-05,
+      "loss": 2.9633,
+      "step": 10639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.211238900044896e-05,
+      "loss": 2.683,
+      "step": 10640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.205436832591187e-05,
+      "loss": 2.7309,
+      "step": 10641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.199637819507125e-05,
+      "loss": 3.0513,
+      "step": 10642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.19384186118812e-05,
+      "loss": 3.0113,
+      "step": 10643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.188048958029379e-05,
+      "loss": 2.9519,
+      "step": 10644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1822591104258964e-05,
+      "loss": 2.8851,
+      "step": 10645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1764723187724624e-05,
+      "loss": 2.8467,
+      "step": 10646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.170688583463651e-05,
+      "loss": 2.9985,
+      "step": 10647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.164907904893834e-05,
+      "loss": 3.0652,
+      "step": 10648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.159130283457175e-05,
+      "loss": 2.9181,
+      "step": 10649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1533557195476197e-05,
+      "loss": 2.9257,
+      "step": 10650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.147584213558942e-05,
+      "loss": 2.8303,
+      "step": 10651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1418157658846334e-05,
+      "loss": 2.7967,
+      "step": 10652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1360503769180725e-05,
+      "loss": 2.996,
+      "step": 10653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.130288047052334e-05,
+      "loss": 2.9041,
+      "step": 10654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.124528776680371e-05,
+      "loss": 2.8671,
+      "step": 10655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.11877256619484e-05,
+      "loss": 2.9307,
+      "step": 10656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.113019415988274e-05,
+      "loss": 2.7609,
+      "step": 10657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.10726932645294e-05,
+      "loss": 2.9758,
+      "step": 10658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.101522297980915e-05,
+      "loss": 2.7015,
+      "step": 10659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0957783309640736e-05,
+      "loss": 2.7283,
+      "step": 10660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0900374257940696e-05,
+      "loss": 2.8784,
+      "step": 10661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.084299582862356e-05,
+      "loss": 2.862,
+      "step": 10662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.078564802560176e-05,
+      "loss": 2.73,
+      "step": 10663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.072833085278556e-05,
+      "loss": 2.7373,
+      "step": 10664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0671044314083256e-05,
+      "loss": 2.7309,
+      "step": 10665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.061378841340086e-05,
+      "loss": 2.9376,
+      "step": 10666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0556563154642746e-05,
+      "loss": 2.8287,
+      "step": 10667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0499368541710555e-05,
+      "loss": 2.7808,
+      "step": 10668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.044220457850435e-05,
+      "loss": 2.6868,
+      "step": 10669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.038507126892189e-05,
+      "loss": 2.932,
+      "step": 10670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.032796861685879e-05,
+      "loss": 2.9647,
+      "step": 10671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0270896626208763e-05,
+      "loss": 2.912,
+      "step": 10672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.021385530086325e-05,
+      "loss": 2.7983,
+      "step": 10673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.015684464471187e-05,
+      "loss": 2.9246,
+      "step": 10674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0099864661641626e-05,
+      "loss": 2.7895,
+      "step": 10675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.004291535553812e-05,
+      "loss": 2.8511,
+      "step": 10676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.99859967302842e-05,
+      "loss": 2.8309,
+      "step": 10677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9929108789761144e-05,
+      "loss": 2.7085,
+      "step": 10678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.987225153784791e-05,
+      "loss": 3.0969,
+      "step": 10679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.981542497842123e-05,
+      "loss": 3.1323,
+      "step": 10680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9758629115356056e-05,
+      "loss": 2.6982,
+      "step": 10681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.970186395252496e-05,
+      "loss": 3.0427,
+      "step": 10682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.964512949379857e-05,
+      "loss": 3.0918,
+      "step": 10683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.958842574304545e-05,
+      "loss": 2.8256,
+      "step": 10684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.953175270413196e-05,
+      "loss": 3.0957,
+      "step": 10685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9475110380922396e-05,
+      "loss": 2.9555,
+      "step": 10686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9418498777279e-05,
+      "loss": 2.8273,
+      "step": 10687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9361917897061904e-05,
+      "loss": 2.8961,
+      "step": 10688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.930536774412908e-05,
+      "loss": 3.0127,
+      "step": 10689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.924884832233673e-05,
+      "loss": 2.911,
+      "step": 10690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.919235963553831e-05,
+      "loss": 2.9441,
+      "step": 10691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9135901687585915e-05,
+      "loss": 2.8569,
+      "step": 10692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.907947448232886e-05,
+      "loss": 3.1003,
+      "step": 10693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9023078023614995e-05,
+      "loss": 2.95,
+      "step": 10694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8966712315289596e-05,
+      "loss": 2.6651,
+      "step": 10695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8910377361196135e-05,
+      "loss": 2.9971,
+      "step": 10696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.885407316517587e-05,
+      "loss": 2.7621,
+      "step": 10697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.879779973106785e-05,
+      "loss": 2.8892,
+      "step": 10698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8741557062709275e-05,
+      "loss": 3.067,
+      "step": 10699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.868534516393508e-05,
+      "loss": 3.0022,
+      "step": 10700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.862916403857809e-05,
+      "loss": 2.9022,
+      "step": 10701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8573013690469125e-05,
+      "loss": 2.9811,
+      "step": 10702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.851689412343685e-05,
+      "loss": 2.7784,
+      "step": 10703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.846080534130787e-05,
+      "loss": 3.0682,
+      "step": 10704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.840474734790662e-05,
+      "loss": 3.1437,
+      "step": 10705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.834872014705549e-05,
+      "loss": 2.8318,
+      "step": 10706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.829272374257482e-05,
+      "loss": 2.9502,
+      "step": 10707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.823675813828271e-05,
+      "loss": 3.0777,
+      "step": 10708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.818082333799523e-05,
+      "loss": 2.869,
+      "step": 10709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.812491934552637e-05,
+      "loss": 2.8651,
+      "step": 10710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.806904616468821e-05,
+      "loss": 2.8081,
+      "step": 10711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8013203799290136e-05,
+      "loss": 2.7302,
+      "step": 10712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7957392253140276e-05,
+      "loss": 3.1066,
+      "step": 10713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.790161153004374e-05,
+      "loss": 2.9645,
+      "step": 10714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.78458616338045e-05,
+      "loss": 2.8487,
+      "step": 10715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.779014256822345e-05,
+      "loss": 3.0916,
+      "step": 10716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.773445433710016e-05,
+      "loss": 2.8413,
+      "step": 10717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.767879694423172e-05,
+      "loss": 3.0338,
+      "step": 10718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7623170393413187e-05,
+      "loss": 2.8699,
+      "step": 10719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.756757468843748e-05,
+      "loss": 3.103,
+      "step": 10720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.751200983309556e-05,
+      "loss": 2.9233,
+      "step": 10721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.745647583117607e-05,
+      "loss": 2.8228,
+      "step": 10722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.740097268646576e-05,
+      "loss": 2.8912,
+      "step": 10723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.73455004027491e-05,
+      "loss": 3.1737,
+      "step": 10724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7290058983808556e-05,
+      "loss": 2.8954,
+      "step": 10725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7234648433424396e-05,
+      "loss": 2.677,
+      "step": 10726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.717926875537515e-05,
+      "loss": 2.9816,
+      "step": 10727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.712391995343646e-05,
+      "loss": 2.7705,
+      "step": 10728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.706860203138286e-05,
+      "loss": 2.7081,
+      "step": 10729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.7013314992985836e-05,
+      "loss": 2.9119,
+      "step": 10730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6958058842015536e-05,
+      "loss": 2.7858,
+      "step": 10731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6902833582239346e-05,
+      "loss": 3.1475,
+      "step": 10732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6847639217423134e-05,
+      "loss": 2.6957,
+      "step": 10733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6792475751330335e-05,
+      "loss": 2.8674,
+      "step": 10734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.673734318772216e-05,
+      "loss": 2.746,
+      "step": 10735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.668224153035816e-05,
+      "loss": 2.7646,
+      "step": 10736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.662717078299522e-05,
+      "loss": 3.0502,
+      "step": 10737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6572130949388615e-05,
+      "loss": 3.0933,
+      "step": 10738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.651712203329122e-05,
+      "loss": 2.9639,
+      "step": 10739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6462144038453945e-05,
+      "loss": 2.7586,
+      "step": 10740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.640719696862544e-05,
+      "loss": 2.9002,
+      "step": 10741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.635228082755239e-05,
+      "loss": 2.8931,
+      "step": 10742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6297395618979275e-05,
+      "loss": 2.9754,
+      "step": 10743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6242541346648514e-05,
+      "loss": 3.1199,
+      "step": 10744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.618771801430049e-05,
+      "loss": 3.1804,
+      "step": 10745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.613292562567328e-05,
+      "loss": 3.1196,
+      "step": 10746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.607816418450306e-05,
+      "loss": 3.1941,
+      "step": 10747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.602343369452372e-05,
+      "loss": 2.7422,
+      "step": 10748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.596873415946712e-05,
+      "loss": 2.9739,
+      "step": 10749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.591406558306321e-05,
+      "loss": 2.9675,
+      "step": 10750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.58594279690393e-05,
+      "loss": 2.7654,
+      "step": 10751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.580482132112129e-05,
+      "loss": 3.0525,
+      "step": 10752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.575024564303221e-05,
+      "loss": 3.0201,
+      "step": 10753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5695700938493624e-05,
+      "loss": 2.805,
+      "step": 10754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.564118721122473e-05,
+      "loss": 2.9541,
+      "step": 10755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.558670446494251e-05,
+      "loss": 2.9445,
+      "step": 10756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5532252703361976e-05,
+      "loss": 3.0241,
+      "step": 10757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.547783193019595e-05,
+      "loss": 2.944,
+      "step": 10758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.542344214915522e-05,
+      "loss": 2.8844,
+      "step": 10759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5369083363948406e-05,
+      "loss": 2.9044,
+      "step": 10760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5314755578282016e-05,
+      "loss": 2.8032,
+      "step": 10761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.526045879586049e-05,
+      "loss": 2.8143,
+      "step": 10762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.520619302038603e-05,
+      "loss": 3.2346,
+      "step": 10763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.51519582555589e-05,
+      "loss": 3.0012,
+      "step": 10764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5097754505077135e-05,
+      "loss": 2.7748,
+      "step": 10765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.504358177263668e-05,
+      "loss": 2.7238,
+      "step": 10766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4989440061931284e-05,
+      "loss": 2.8717,
+      "step": 10767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4935329376652786e-05,
+      "loss": 3.0014,
+      "step": 10768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.488124972049079e-05,
+      "loss": 2.6527,
+      "step": 10769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.482720109713256e-05,
+      "loss": 2.9802,
+      "step": 10770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.477318351026383e-05,
+      "loss": 2.8395,
+      "step": 10771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.471919696356752e-05,
+      "loss": 2.983,
+      "step": 10772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.466524146072504e-05,
+      "loss": 3.0005,
+      "step": 10773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.46113170054151e-05,
+      "loss": 2.7487,
+      "step": 10774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.455742360131498e-05,
+      "loss": 2.9029,
+      "step": 10775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.450356125209903e-05,
+      "loss": 2.6682,
+      "step": 10776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.444972996144025e-05,
+      "loss": 3.0787,
+      "step": 10777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.43959297330091e-05,
+      "loss": 2.8892,
+      "step": 10778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.434216057047397e-05,
+      "loss": 2.7524,
+      "step": 10779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.42884224775012e-05,
+      "loss": 2.9084,
+      "step": 10780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4234715457755036e-05,
+      "loss": 2.8265,
+      "step": 10781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.418103951489743e-05,
+      "loss": 3.026,
+      "step": 10782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.412739465258847e-05,
+      "loss": 2.7942,
+      "step": 10783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.407378087448588e-05,
+      "loss": 3.2001,
+      "step": 10784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4020198184245465e-05,
+      "loss": 3.0929,
+      "step": 10785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3966646585520686e-05,
+      "loss": 2.8063,
+      "step": 10786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3913126081963295e-05,
+      "loss": 2.814,
+      "step": 10787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.385963667722232e-05,
+      "loss": 2.971,
+      "step": 10788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.380617837494533e-05,
+      "loss": 2.9001,
+      "step": 10789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.375275117877714e-05,
+      "loss": 2.912,
+      "step": 10790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3699355092361e-05,
+      "loss": 2.8997,
+      "step": 10791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.364599011933751e-05,
+      "loss": 2.9774,
+      "step": 10792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.359265626334563e-05,
+      "loss": 2.7441,
+      "step": 10793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3539353528021965e-05,
+      "loss": 2.9865,
+      "step": 10794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3486081917000984e-05,
+      "loss": 2.906,
+      "step": 10795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.343284143391507e-05,
+      "loss": 2.8796,
+      "step": 10796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3379632082394535e-05,
+      "loss": 2.6612,
+      "step": 10797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.332645386606748e-05,
+      "loss": 3.0964,
+      "step": 10798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.32733067885599e-05,
+      "loss": 2.7873,
+      "step": 10799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3220190853495764e-05,
+      "loss": 2.9956,
+      "step": 10800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.316710606449675e-05,
+      "loss": 2.7769,
+      "step": 10801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3114052425182624e-05,
+      "loss": 2.8229,
+      "step": 10802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3061029939170826e-05,
+      "loss": 2.9748,
+      "step": 10803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.300803861007674e-05,
+      "loss": 2.7155,
+      "step": 10804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.295507844151375e-05,
+      "loss": 3.0418,
+      "step": 10805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2902149437092865e-05,
+      "loss": 2.938,
+      "step": 10806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.284925160042319e-05,
+      "loss": 3.0817,
+      "step": 10807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2796384935111674e-05,
+      "loss": 2.8287,
+      "step": 10808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.274354944476289e-05,
+      "loss": 2.9598,
+      "step": 10809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2690745132979827e-05,
+      "loss": 2.5604,
+      "step": 10810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.263797200336261e-05,
+      "loss": 2.9618,
+      "step": 10811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.258523005951004e-05,
+      "loss": 2.8416,
+      "step": 10812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.253251930501806e-05,
+      "loss": 3.1688,
+      "step": 10813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2479839743480967e-05,
+      "loss": 3.1342,
+      "step": 10814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.242719137849077e-05,
+      "loss": 2.8419,
+      "step": 10815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.237457421363733e-05,
+      "loss": 2.9199,
+      "step": 10816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2321988252508436e-05,
+      "loss": 2.9507,
+      "step": 10817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.226943349868972e-05,
+      "loss": 2.9985,
+      "step": 10818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2216909955764714e-05,
+      "loss": 2.6563,
+      "step": 10819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2164417627314714e-05,
+      "loss": 2.832,
+      "step": 10820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.211195651691901e-05,
+      "loss": 3.1202,
+      "step": 10821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2059526628154765e-05,
+      "loss": 2.7122,
+      "step": 10822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.200712796459693e-05,
+      "loss": 3.0612,
+      "step": 10823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1954760529818394e-05,
+      "loss": 2.9919,
+      "step": 10824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1902424327389774e-05,
+      "loss": 2.9228,
+      "step": 10825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.185011936088001e-05,
+      "loss": 2.9607,
+      "step": 10826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.179784563385508e-05,
+      "loss": 2.7887,
+      "step": 10827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1745603149879855e-05,
+      "loss": 2.8296,
+      "step": 10828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.169339191251609e-05,
+      "loss": 2.8002,
+      "step": 10829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.164121192532411e-05,
+      "loss": 2.8423,
+      "step": 10830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.158906319186195e-05,
+      "loss": 3.1763,
+      "step": 10831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.153694571568517e-05,
+      "loss": 2.8894,
+      "step": 10832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.148485950034775e-05,
+      "loss": 2.8794,
+      "step": 10833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.14328045494009e-05,
+      "loss": 2.6578,
+      "step": 10834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.138078086639446e-05,
+      "loss": 2.8632,
+      "step": 10835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.132878845487531e-05,
+      "loss": 2.9272,
+      "step": 10836
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.127682731838894e-05,
+      "loss": 2.9315,
+      "step": 10837
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.122489746047825e-05,
+      "loss": 3.1072,
+      "step": 10838
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.117299888468412e-05,
+      "loss": 2.8302,
+      "step": 10839
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.112113159454539e-05,
+      "loss": 2.8956,
+      "step": 10840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.106929559359862e-05,
+      "loss": 2.7765,
+      "step": 10841
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1017490885378316e-05,
+      "loss": 2.8243,
+      "step": 10842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.096571747341687e-05,
+      "loss": 2.9498,
+      "step": 10843
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.091397536124447e-05,
+      "loss": 2.8449,
+      "step": 10844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.086226455238928e-05,
+      "loss": 2.8,
+      "step": 10845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.081058505037716e-05,
+      "loss": 2.8064,
+      "step": 10846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0758936858732165e-05,
+      "loss": 3.0083,
+      "step": 10847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0707319980975596e-05,
+      "loss": 2.6062,
+      "step": 10848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.065573442062748e-05,
+      "loss": 2.7211,
+      "step": 10849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.060418018120482e-05,
+      "loss": 2.9029,
+      "step": 10850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0552657266223205e-05,
+      "loss": 2.7714,
+      "step": 10851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.050116567919554e-05,
+      "loss": 2.7779,
+      "step": 10852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.044970542363302e-05,
+      "loss": 3.0736,
+      "step": 10853
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.039827650304445e-05,
+      "loss": 2.8932,
+      "step": 10854
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.034687892093664e-05,
+      "loss": 2.9848,
+      "step": 10855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.029551268081405e-05,
+      "loss": 2.7277,
+      "step": 10856
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.024417778617928e-05,
+      "loss": 2.9092,
+      "step": 10857
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.019287424053264e-05,
+      "loss": 2.9341,
+      "step": 10858
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.014160204737227e-05,
+      "loss": 3.1731,
+      "step": 10859
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.009036121019427e-05,
+      "loss": 3.027,
+      "step": 10860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.0039151732492504e-05,
+      "loss": 2.7375,
+      "step": 10861
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.998797361775869e-05,
+      "loss": 2.814,
+      "step": 10862
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.99368268694828e-05,
+      "loss": 3.1142,
+      "step": 10863
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9885711491151946e-05,
+      "loss": 2.9629,
+      "step": 10864
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9834627486251664e-05,
+      "loss": 2.811,
+      "step": 10865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.978357485826512e-05,
+      "loss": 2.9462,
+      "step": 10866
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.973255361067346e-05,
+      "loss": 3.0753,
+      "step": 10867
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.968156374695564e-05,
+      "loss": 3.1123,
+      "step": 10868
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.963060527058826e-05,
+      "loss": 2.917,
+      "step": 10869
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.957967818504638e-05,
+      "loss": 3.179,
+      "step": 10870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.952878249380204e-05,
+      "loss": 2.9204,
+      "step": 10871
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.947791820032609e-05,
+      "loss": 2.9434,
+      "step": 10872
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.942708530808636e-05,
+      "loss": 2.6654,
+      "step": 10873
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9376283820549254e-05,
+      "loss": 3.1722,
+      "step": 10874
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.93255137411786e-05,
+      "loss": 3.0007,
+      "step": 10875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.927477507343619e-05,
+      "loss": 3.0216,
+      "step": 10876
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.92240678207818e-05,
+      "loss": 2.8765,
+      "step": 10877
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.917339198667291e-05,
+      "loss": 3.0489,
+      "step": 10878
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9122747574564856e-05,
+      "loss": 2.7251,
+      "step": 10879
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.907213458791098e-05,
+      "loss": 2.871,
+      "step": 10880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9021553030162305e-05,
+      "loss": 2.8746,
+      "step": 10881
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.897100290476785e-05,
+      "loss": 3.1598,
+      "step": 10882
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.892048421517441e-05,
+      "loss": 2.6627,
+      "step": 10883
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.886999696482668e-05,
+      "loss": 2.9212,
+      "step": 10884
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.881954115716713e-05,
+      "loss": 2.9955,
+      "step": 10885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.876911679563633e-05,
+      "loss": 2.8033,
+      "step": 10886
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.87187238836722e-05,
+      "loss": 3.2334,
+      "step": 10887
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8668362424711224e-05,
+      "loss": 2.793,
+      "step": 10888
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.861803242218698e-05,
+      "loss": 2.9723,
+      "step": 10889
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.856773387953155e-05,
+      "loss": 2.6722,
+      "step": 10890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.851746680017454e-05,
+      "loss": 2.8206,
+      "step": 10891
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8467231187543416e-05,
+      "loss": 3.1761,
+      "step": 10892
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.841702704506361e-05,
+      "loss": 3.001,
+      "step": 10893
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.836685437615828e-05,
+      "loss": 3.1077,
+      "step": 10894
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8316713184248566e-05,
+      "loss": 2.7801,
+      "step": 10895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8266603472753346e-05,
+      "loss": 3.0056,
+      "step": 10896
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.821652524508951e-05,
+      "loss": 3.1079,
+      "step": 10897
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.816647850467159e-05,
+      "loss": 3.2002,
+      "step": 10898
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.811646325491214e-05,
+      "loss": 2.9378,
+      "step": 10899
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.806647949922149e-05,
+      "loss": 2.8543,
+      "step": 10900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.801652724100785e-05,
+      "loss": 2.8847,
+      "step": 10901
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.796660648367728e-05,
+      "loss": 2.8157,
+      "step": 10902
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.791671723063367e-05,
+      "loss": 3.0236,
+      "step": 10903
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.786685948527879e-05,
+      "loss": 2.6746,
+      "step": 10904
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.781703325101227e-05,
+      "loss": 2.9715,
+      "step": 10905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7767238531231374e-05,
+      "loss": 2.7687,
+      "step": 10906
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.771747532933184e-05,
+      "loss": 2.6537,
+      "step": 10907
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7667743648706345e-05,
+      "loss": 3.0541,
+      "step": 10908
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7618043492746335e-05,
+      "loss": 3.1037,
+      "step": 10909
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.756837486484027e-05,
+      "loss": 3.027,
+      "step": 10910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.751873776837522e-05,
+      "loss": 2.5645,
+      "step": 10911
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7469132206735424e-05,
+      "loss": 2.9336,
+      "step": 10912
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7419558183303556e-05,
+      "loss": 3.0481,
+      "step": 10913
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.737001570145976e-05,
+      "loss": 2.7206,
+      "step": 10914
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7320504764582205e-05,
+      "loss": 3.0024,
+      "step": 10915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7271025376046766e-05,
+      "loss": 2.9313,
+      "step": 10916
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7221577539227334e-05,
+      "loss": 2.6209,
+      "step": 10917
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.717216125749556e-05,
+      "loss": 2.7213,
+      "step": 10918
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.712277653422091e-05,
+      "loss": 2.9352,
+      "step": 10919
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7073423372770756e-05,
+      "loss": 2.6461,
+      "step": 10920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.702410177651028e-05,
+      "loss": 2.7976,
+      "step": 10921
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.697481174880246e-05,
+      "loss": 2.8902,
+      "step": 10922
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6925553293008506e-05,
+      "loss": 2.7687,
+      "step": 10923
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6876326412486786e-05,
+      "loss": 2.8322,
+      "step": 10924
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.682713111059416e-05,
+      "loss": 3.0242,
+      "step": 10925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.677796739068484e-05,
+      "loss": 2.764,
+      "step": 10926
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.672883525611137e-05,
+      "loss": 2.8015,
+      "step": 10927
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.667973471022362e-05,
+      "loss": 2.5492,
+      "step": 10928
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.66306657563697e-05,
+      "loss": 2.8625,
+      "step": 10929
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.658162839789558e-05,
+      "loss": 3.0323,
+      "step": 10930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.653262263814455e-05,
+      "loss": 2.9212,
+      "step": 10931
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.64836484804586e-05,
+      "loss": 2.8062,
+      "step": 10932
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.64347059281766e-05,
+      "loss": 2.9595,
+      "step": 10933
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.638579498463612e-05,
+      "loss": 2.953,
+      "step": 10934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6336915653172044e-05,
+      "loss": 2.8936,
+      "step": 10935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6288067937117367e-05,
+      "loss": 2.8191,
+      "step": 10936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.62392518398027e-05,
+      "loss": 3.0298,
+      "step": 10937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.619046736455672e-05,
+      "loss": 2.869,
+      "step": 10938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.614171451470582e-05,
+      "loss": 2.8425,
+      "step": 10939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.60929932935743e-05,
+      "loss": 2.8307,
+      "step": 10940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6044303704484246e-05,
+      "loss": 2.8706,
+      "step": 10941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5995645750755646e-05,
+      "loss": 3.1343,
+      "step": 10942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.594701943570622e-05,
+      "loss": 2.9097,
+      "step": 10943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.58984247626517e-05,
+      "loss": 2.8887,
+      "step": 10944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5849861734905454e-05,
+      "loss": 3.0399,
+      "step": 10945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.580133035577904e-05,
+      "loss": 2.7908,
+      "step": 10946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.575283062858131e-05,
+      "loss": 2.9666,
+      "step": 10947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5704362556619576e-05,
+      "loss": 2.8793,
+      "step": 10948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5655926143198404e-05,
+      "loss": 2.9119,
+      "step": 10949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.560752139162077e-05,
+      "loss": 3.0551,
+      "step": 10950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.555914830518703e-05,
+      "loss": 2.98,
+      "step": 10951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.55108068871956e-05,
+      "loss": 2.9482,
+      "step": 10952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.546249714094274e-05,
+      "loss": 2.8866,
+      "step": 10953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541421906972242e-05,
+      "loss": 3.0585,
+      "step": 10954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536597267682662e-05,
+      "loss": 2.7825,
+      "step": 10955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.531775796554504e-05,
+      "loss": 3.0266,
+      "step": 10956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5269574939165216e-05,
+      "loss": 3.0886,
+      "step": 10957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.522142360097264e-05,
+      "loss": 3.1041,
+      "step": 10958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.517330395425056e-05,
+      "loss": 2.8572,
+      "step": 10959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5125216002279966e-05,
+      "loss": 2.9269,
+      "step": 10960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.507715974833986e-05,
+      "loss": 3.0501,
+      "step": 10961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.502913519570716e-05,
+      "loss": 2.9864,
+      "step": 10962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.498114234765626e-05,
+      "loss": 2.8036,
+      "step": 10963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4933181207459695e-05,
+      "loss": 2.9318,
+      "step": 10964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.488525177838775e-05,
+      "loss": 2.7579,
+      "step": 10965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.483735406370842e-05,
+      "loss": 2.9851,
+      "step": 10966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.478948806668797e-05,
+      "loss": 2.9668,
+      "step": 10967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4741653790589854e-05,
+      "loss": 2.9533,
+      "step": 10968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.469385123867608e-05,
+      "loss": 2.83,
+      "step": 10969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4646080414205697e-05,
+      "loss": 2.7828,
+      "step": 10970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4598341320436436e-05,
+      "loss": 2.685,
+      "step": 10971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4550633960623024e-05,
+      "loss": 3.0491,
+      "step": 10972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4502958338018753e-05,
+      "loss": 2.7803,
+      "step": 10973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.445531445587435e-05,
+      "loss": 2.9181,
+      "step": 10974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4407702317438504e-05,
+      "loss": 2.9867,
+      "step": 10975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.436012192595761e-05,
+      "loss": 2.8398,
+      "step": 10976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.431257328467607e-05,
+      "loss": 2.9321,
+      "step": 10977
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4265056396836025e-05,
+      "loss": 2.9287,
+      "step": 10978
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.421757126567743e-05,
+      "loss": 2.9902,
+      "step": 10979
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.417011789443819e-05,
+      "loss": 2.8583,
+      "step": 10980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.412269628635395e-05,
+      "loss": 2.4774,
+      "step": 10981
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.407530644465806e-05,
+      "loss": 3.155,
+      "step": 10982
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.402794837258211e-05,
+      "loss": 3.0263,
+      "step": 10983
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3980622073355e-05,
+      "loss": 2.7584,
+      "step": 10984
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3933327550204064e-05,
+      "loss": 2.8417,
+      "step": 10985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.388606480635376e-05,
+      "loss": 3.1206,
+      "step": 10986
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.383883384502706e-05,
+      "loss": 2.9588,
+      "step": 10987
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3791634669444124e-05,
+      "loss": 2.6815,
+      "step": 10988
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3744467282823656e-05,
+      "loss": 3.0646,
+      "step": 10989
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3697331688381574e-05,
+      "loss": 3.0313,
+      "step": 10990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3650227889331976e-05,
+      "loss": 2.8749,
+      "step": 10991
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.360315588888668e-05,
+      "loss": 2.8479,
+      "step": 10992
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.355611569025535e-05,
+      "loss": 3.0175,
+      "step": 10993
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.350910729664541e-05,
+      "loss": 2.8309,
+      "step": 10994
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.346213071126225e-05,
+      "loss": 2.8767,
+      "step": 10995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.341518593730897e-05,
+      "loss": 3.0193,
+      "step": 10996
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.336827297798661e-05,
+      "loss": 2.7412,
+      "step": 10997
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3321391836493965e-05,
+      "loss": 2.7959,
+      "step": 10998
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.327454251602769e-05,
+      "loss": 3.0385,
+      "step": 10999
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.322772501978222e-05,
+      "loss": 2.9283,
+      "step": 11000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.31809393509499e-05,
+      "loss": 2.7226,
+      "step": 11001
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.313418551272085e-05,
+      "loss": 2.7279,
+      "step": 11002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.308746350828301e-05,
+      "loss": 3.0817,
+      "step": 11003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.304077334082223e-05,
+      "loss": 2.9219,
+      "step": 11004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.299411501352201e-05,
+      "loss": 2.6637,
+      "step": 11005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.294748852956409e-05,
+      "loss": 3.0416,
+      "step": 11006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.290089389212736e-05,
+      "loss": 3.1034,
+      "step": 11007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.285433110438934e-05,
+      "loss": 2.7236,
+      "step": 11008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.280780016952456e-05,
+      "loss": 2.8334,
+      "step": 11009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.276130109070607e-05,
+      "loss": 2.8108,
+      "step": 11010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.271483387110441e-05,
+      "loss": 2.9511,
+      "step": 11011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.266839851388792e-05,
+      "loss": 2.8585,
+      "step": 11012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2621995022222906e-05,
+      "loss": 2.7425,
+      "step": 11013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2575623399273445e-05,
+      "loss": 2.7875,
+      "step": 11014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.252928364820146e-05,
+      "loss": 3.0819,
+      "step": 11015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.248297577216663e-05,
+      "loss": 3.0207,
+      "step": 11016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.24366997743265e-05,
+      "loss": 2.8952,
+      "step": 11017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.239045565783655e-05,
+      "loss": 2.9149,
+      "step": 11018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2344243425849906e-05,
+      "loss": 2.7414,
+      "step": 11019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.229806308151762e-05,
+      "loss": 2.8718,
+      "step": 11020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2251914627988486e-05,
+      "loss": 3.0053,
+      "step": 11021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.22057980684094e-05,
+      "loss": 2.662,
+      "step": 11022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.215971340592461e-05,
+      "loss": 2.9529,
+      "step": 11023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.211366064367671e-05,
+      "loss": 2.8378,
+      "step": 11024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2067639784805534e-05,
+      "loss": 2.7014,
+      "step": 11025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.20216508324494e-05,
+      "loss": 2.7548,
+      "step": 11026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.197569378974396e-05,
+      "loss": 2.8344,
+      "step": 11027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.192976865982289e-05,
+      "loss": 2.9691,
+      "step": 11028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1883875445817665e-05,
+      "loss": 2.8204,
+      "step": 11029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.183801415085741e-05,
+      "loss": 3.0592,
+      "step": 11030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1792184778069496e-05,
+      "loss": 2.9398,
+      "step": 11031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.174638733057855e-05,
+      "loss": 2.941,
+      "step": 11032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.170062181150757e-05,
+      "loss": 2.9507,
+      "step": 11033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1654888223977054e-05,
+      "loss": 2.6304,
+      "step": 11034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.160918657110534e-05,
+      "loss": 3.0268,
+      "step": 11035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.156351685600878e-05,
+      "loss": 2.8829,
+      "step": 11036
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.151787908180126e-05,
+      "loss": 2.7493,
+      "step": 11037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.147227325159474e-05,
+      "loss": 2.952,
+      "step": 11038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.14266993684989e-05,
+      "loss": 2.9969,
+      "step": 11039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.138115743562125e-05,
+      "loss": 2.9421,
+      "step": 11040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1335647456067097e-05,
+      "loss": 2.4629,
+      "step": 11041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1290169432939555e-05,
+      "loss": 2.7987,
+      "step": 11042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.124472336933976e-05,
+      "loss": 2.9526,
+      "step": 11043
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.119930926836623e-05,
+      "loss": 2.7071,
+      "step": 11044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.115392713311588e-05,
+      "loss": 2.9572,
+      "step": 11045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1108576966682846e-05,
+      "loss": 2.7885,
+      "step": 11046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.106325877215971e-05,
+      "loss": 3.0095,
+      "step": 11047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.101797255263622e-05,
+      "loss": 2.7841,
+      "step": 11048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0972718311200466e-05,
+      "loss": 2.9578,
+      "step": 11049
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.092749605093814e-05,
+      "loss": 2.9918,
+      "step": 11050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0882305774932786e-05,
+      "loss": 2.827,
+      "step": 11051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.08371474862657e-05,
+      "loss": 2.9137,
+      "step": 11052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.079202118801605e-05,
+      "loss": 2.9227,
+      "step": 11053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0746926883260865e-05,
+      "loss": 2.9852,
+      "step": 11054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0701864575074954e-05,
+      "loss": 3.0068,
+      "step": 11055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.065683426653088e-05,
+      "loss": 2.833,
+      "step": 11056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0611835960699164e-05,
+      "loss": 2.9451,
+      "step": 11057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.056686966064798e-05,
+      "loss": 3.0909,
+      "step": 11058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.052193536944359e-05,
+      "loss": 2.9312,
+      "step": 11059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.047703309014971e-05,
+      "loss": 2.9757,
+      "step": 11060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0432162825828114e-05,
+      "loss": 2.9502,
+      "step": 11061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.03873245795383e-05,
+      "loss": 2.8887,
+      "step": 11062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0342518354337655e-05,
+      "loss": 2.7468,
+      "step": 11063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.029774415328135e-05,
+      "loss": 2.7231,
+      "step": 11064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.025300197942227e-05,
+      "loss": 2.5513,
+      "step": 11065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0208291835811474e-05,
+      "loss": 2.9906,
+      "step": 11066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0163613725497197e-05,
+      "loss": 2.6836,
+      "step": 11067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0118967651526275e-05,
+      "loss": 2.8228,
+      "step": 11068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0074353616942552e-05,
+      "loss": 2.8489,
+      "step": 11069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.002977162478837e-05,
+      "loss": 2.8729,
+      "step": 11070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9985221678103525e-05,
+      "loss": 2.9842,
+      "step": 11071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9940703779925692e-05,
+      "loss": 2.8878,
+      "step": 11072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9896217933290337e-05,
+      "loss": 2.9578,
+      "step": 11073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9851764141230864e-05,
+      "loss": 3.1318,
+      "step": 11074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9807342406778347e-05,
+      "loss": 2.9628,
+      "step": 11075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9762952732961812e-05,
+      "loss": 2.803,
+      "step": 11076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.971859512280789e-05,
+      "loss": 2.9197,
+      "step": 11077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9674269579341273e-05,
+      "loss": 2.9442,
+      "step": 11078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9629976105584266e-05,
+      "loss": 2.8024,
+      "step": 11079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9585714704557176e-05,
+      "loss": 2.8016,
+      "step": 11080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.954148537927781e-05,
+      "loss": 2.7603,
+      "step": 11081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.949728813276231e-05,
+      "loss": 2.8985,
+      "step": 11082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9453122968023992e-05,
+      "loss": 2.8324,
+      "step": 11083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9408989888074667e-05,
+      "loss": 2.9328,
+      "step": 11084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9364888895923202e-05,
+      "loss": 3.0288,
+      "step": 11085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.932081999457692e-05,
+      "loss": 2.7683,
+      "step": 11086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.927678318704069e-05,
+      "loss": 2.9288,
+      "step": 11087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9232778476317223e-05,
+      "loss": 2.8272,
+      "step": 11088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9188805865406953e-05,
+      "loss": 2.7888,
+      "step": 11089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9144865357308205e-05,
+      "loss": 2.9147,
+      "step": 11090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9100956955017188e-05,
+      "loss": 3.0632,
+      "step": 11091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9057080661527845e-05,
+      "loss": 2.9287,
+      "step": 11092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.901323647983184e-05,
+      "loss": 2.7981,
+      "step": 11093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.896942441291883e-05,
+      "loss": 3.11,
+      "step": 11094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8925644463776156e-05,
+      "loss": 2.9383,
+      "step": 11095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8881896635389037e-05,
+      "loss": 3.1047,
+      "step": 11096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8838180930740422e-05,
+      "loss": 3.0091,
+      "step": 11097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8794497352811146e-05,
+      "loss": 2.8694,
+      "step": 11098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8750845904579826e-05,
+      "loss": 3.1545,
+      "step": 11099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8707226589022862e-05,
+      "loss": 2.8377,
+      "step": 11100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8663639409114482e-05,
+      "loss": 2.7675,
+      "step": 11101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8620084367826704e-05,
+      "loss": 2.7632,
+      "step": 11102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8576561468129536e-05,
+      "loss": 2.9735,
+      "step": 11103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8533070712990438e-05,
+      "loss": 2.69,
+      "step": 11104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8489612105375095e-05,
+      "loss": 2.8488,
+      "step": 11105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8446185648246526e-05,
+      "loss": 2.8627,
+      "step": 11106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8402791344566027e-05,
+      "loss": 2.859,
+      "step": 11107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.835942919729234e-05,
+      "loss": 2.9697,
+      "step": 11108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8316099209382274e-05,
+      "loss": 2.6162,
+      "step": 11109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8272801383790236e-05,
+      "loss": 2.9342,
+      "step": 11110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8229535723468645e-05,
+      "loss": 2.7902,
+      "step": 11111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.818630223136759e-05,
+      "loss": 2.5421,
+      "step": 11112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.814310091043498e-05,
+      "loss": 3.0372,
+      "step": 11113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8099931763616526e-05,
+      "loss": 2.6968,
+      "step": 11114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8056794793855757e-05,
+      "loss": 2.8231,
+      "step": 11115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8013690004094095e-05,
+      "loss": 2.7278,
+      "step": 11116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.797061739727058e-05,
+      "loss": 2.9638,
+      "step": 11117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.79275769763222e-05,
+      "loss": 2.9879,
+      "step": 11118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.788456874418388e-05,
+      "loss": 2.8555,
+      "step": 11119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7841592703787942e-05,
+      "loss": 2.8831,
+      "step": 11120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.779864885806499e-05,
+      "loss": 2.777,
+      "step": 11121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7755737209942954e-05,
+      "loss": 2.9264,
+      "step": 11122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7712857762348053e-05,
+      "loss": 2.9557,
+      "step": 11123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.767001051820378e-05,
+      "loss": 2.9785,
+      "step": 11124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.762719548043202e-05,
+      "loss": 2.8463,
+      "step": 11125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.758441265195216e-05,
+      "loss": 2.9802,
+      "step": 11126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.754166203568109e-05,
+      "loss": 2.862,
+      "step": 11127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7498943634534146e-05,
+      "loss": 2.9504,
+      "step": 11128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7456257451423884e-05,
+      "loss": 2.8531,
+      "step": 11129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7413603489261087e-05,
+      "loss": 2.7438,
+      "step": 11130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7370981750954092e-05,
+      "loss": 2.8527,
+      "step": 11131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.732839223940914e-05,
+      "loss": 2.7832,
+      "step": 11132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7285834957530176e-05,
+      "loss": 2.8395,
+      "step": 11133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7243309908219106e-05,
+      "loss": 2.9276,
+      "step": 11134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7200817094375495e-05,
+      "loss": 2.8997,
+      "step": 11135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.715835651889681e-05,
+      "loss": 3.0566,
+      "step": 11136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.711592818467823e-05,
+      "loss": 2.9819,
+      "step": 11137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7073532094612772e-05,
+      "loss": 2.7783,
+      "step": 11138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7031168251591298e-05,
+      "loss": 3.079,
+      "step": 11139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6988836658502438e-05,
+      "loss": 3.0197,
+      "step": 11140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6946537318232543e-05,
+      "loss": 3.0354,
+      "step": 11141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.690427023366604e-05,
+      "loss": 2.942,
+      "step": 11142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6862035407684727e-05,
+      "loss": 2.9504,
+      "step": 11143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6819832843168635e-05,
+      "loss": 2.7531,
+      "step": 11144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6777662542995186e-05,
+      "loss": 2.6863,
+      "step": 11145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6735524510039965e-05,
+      "loss": 2.904,
+      "step": 11146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.669341874717618e-05,
+      "loss": 2.7341,
+      "step": 11147
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.665134525727486e-05,
+      "loss": 2.5307,
+      "step": 11148
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6609304043204773e-05,
+      "loss": 2.9238,
+      "step": 11149
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6567295107832622e-05,
+      "loss": 2.7927,
+      "step": 11150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6525318454022783e-05,
+      "loss": 2.7064,
+      "step": 11151
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.648337408463758e-05,
+      "loss": 2.8829,
+      "step": 11152
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6441462002536888e-05,
+      "loss": 2.894,
+      "step": 11153
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6399582210578643e-05,
+      "loss": 2.907,
+      "step": 11154
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6357734711618342e-05,
+      "loss": 2.7071,
+      "step": 11155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6315919508509645e-05,
+      "loss": 2.7875,
+      "step": 11156
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6274136604103493e-05,
+      "loss": 2.7945,
+      "step": 11157
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6232386001249165e-05,
+      "loss": 3.0736,
+      "step": 11158
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6190667702793268e-05,
+      "loss": 3.12,
+      "step": 11159
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6148981711580478e-05,
+      "loss": 2.5758,
+      "step": 11160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6107328030453236e-05,
+      "loss": 3.0616,
+      "step": 11161
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.606570666225161e-05,
+      "loss": 2.8729,
+      "step": 11162
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.602411760981388e-05,
+      "loss": 2.6727,
+      "step": 11163
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.598256087597556e-05,
+      "loss": 2.9709,
+      "step": 11164
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.594103646357049e-05,
+      "loss": 2.9816,
+      "step": 11165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5899544375429795e-05,
+      "loss": 2.6403,
+      "step": 11166
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5858084614382882e-05,
+      "loss": 3.0081,
+      "step": 11167
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5816657183256597e-05,
+      "loss": 2.813,
+      "step": 11168
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5775262084875794e-05,
+      "loss": 3.177,
+      "step": 11169
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5733899322063046e-05,
+      "loss": 3.0727,
+      "step": 11170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5692568897638657e-05,
+      "loss": 2.9556,
+      "step": 11171
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.565127081442087e-05,
+      "loss": 2.9412,
+      "step": 11172
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5610005075225597e-05,
+      "loss": 2.8332,
+      "step": 11173
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5568771682866587e-05,
+      "loss": 3.0112,
+      "step": 11174
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.552757064015543e-05,
+      "loss": 2.8483,
+      "step": 11175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.548640194990143e-05,
+      "loss": 3.0855,
+      "step": 11176
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.544526561491173e-05,
+      "loss": 3.0701,
+      "step": 11177
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.540416163799114e-05,
+      "loss": 2.6873,
+      "step": 11178
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5363090021942704e-05,
+      "loss": 2.9675,
+      "step": 11179
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5322050769566562e-05,
+      "loss": 3.1045,
+      "step": 11180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5281043883661313e-05,
+      "loss": 3.0879,
+      "step": 11181
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5240069367022833e-05,
+      "loss": 2.8228,
+      "step": 11182
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5199127222445273e-05,
+      "loss": 2.9331,
+      "step": 11183
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5158217452720068e-05,
+      "loss": 2.7716,
+      "step": 11184
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.511734006063682e-05,
+      "loss": 3.0583,
+      "step": 11185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5076495048982796e-05,
+      "loss": 2.8177,
+      "step": 11186
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5035682420543048e-05,
+      "loss": 2.9652,
+      "step": 11187
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4994902178100463e-05,
+      "loss": 2.929,
+      "step": 11188
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4954154324435652e-05,
+      "loss": 3.0359,
+      "step": 11189
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4913438862327054e-05,
+      "loss": 3.0835,
+      "step": 11190
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4872755794550895e-05,
+      "loss": 3.005,
+      "step": 11191
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.483210512388123e-05,
+      "loss": 2.783,
+      "step": 11192
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4791486853089907e-05,
+      "loss": 2.8388,
+      "step": 11193
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4750900984946422e-05,
+      "loss": 2.7818,
+      "step": 11194
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4710347522218236e-05,
+      "loss": 2.7706,
+      "step": 11195
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4669826467670577e-05,
+      "loss": 2.8921,
+      "step": 11196
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.462933782406629e-05,
+      "loss": 3.0624,
+      "step": 11197
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.458888159416628e-05,
+      "loss": 2.8505,
+      "step": 11198
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.454845778072906e-05,
+      "loss": 2.7762,
+      "step": 11199
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.450806638651093e-05,
+      "loss": 3.0084,
+      "step": 11200
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.446770741426596e-05,
+      "loss": 2.6517,
+      "step": 11201
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.442738086674634e-05,
+      "loss": 2.7863,
+      "step": 11202
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4387086746701424e-05,
+      "loss": 2.8704,
+      "step": 11203
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4346825056879073e-05,
+      "loss": 3.1117,
+      "step": 11204
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4306595800024257e-05,
+      "loss": 2.6509,
+      "step": 11205
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4266398978880276e-05,
+      "loss": 2.862,
+      "step": 11206
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4226234596187946e-05,
+      "loss": 3.0205,
+      "step": 11207
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.418610265468585e-05,
+      "loss": 2.7318,
+      "step": 11208
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4146003157110518e-05,
+      "loss": 2.8761,
+      "step": 11209
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4105936106196103e-05,
+      "loss": 2.7322,
+      "step": 11210
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4065901504674694e-05,
+      "loss": 2.7914,
+      "step": 11211
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4025899355276047e-05,
+      "loss": 2.7889,
+      "step": 11212
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3985929660727823e-05,
+      "loss": 2.9689,
+      "step": 11213
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.394599242375528e-05,
+      "loss": 2.7544,
+      "step": 11214
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.390608764708163e-05,
+      "loss": 2.8332,
+      "step": 11215
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.386621533342803e-05,
+      "loss": 2.8406,
+      "step": 11216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3826375485512863e-05,
+      "loss": 2.9321,
+      "step": 11217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3786568106053007e-05,
+      "loss": 2.8871,
+      "step": 11218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3746793197762463e-05,
+      "loss": 2.949,
+      "step": 11219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3707050763353555e-05,
+      "loss": 2.7232,
+      "step": 11220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.366734080553601e-05,
+      "loss": 2.9586,
+      "step": 11221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3627663327017655e-05,
+      "loss": 2.86,
+      "step": 11222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3588018330503834e-05,
+      "loss": 2.8992,
+      "step": 11223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.354840581869783e-05,
+      "loss": 3.1351,
+      "step": 11224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3508825794300694e-05,
+      "loss": 2.8715,
+      "step": 11225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3469278260011107e-05,
+      "loss": 2.9009,
+      "step": 11226
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3429763218525857e-05,
+      "loss": 3.0046,
+      "step": 11227
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3390280672539065e-05,
+      "loss": 3.0633,
+      "step": 11228
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3350830624743193e-05,
+      "loss": 3.1947,
+      "step": 11229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.331141307782797e-05,
+      "loss": 2.8419,
+      "step": 11230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.327202803448125e-05,
+      "loss": 3.055,
+      "step": 11231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.32326754973885e-05,
+      "loss": 2.8043,
+      "step": 11232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3193355469233014e-05,
+      "loss": 3.1147,
+      "step": 11233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3154067952695868e-05,
+      "loss": 2.862,
+      "step": 11234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.311481295045592e-05,
+      "loss": 2.9657,
+      "step": 11235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.307559046518992e-05,
+      "loss": 2.7291,
+      "step": 11236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3036400499572174e-05,
+      "loss": 2.8518,
+      "step": 11237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.299724305627482e-05,
+      "loss": 2.8984,
+      "step": 11238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2958118137968164e-05,
+      "loss": 2.9165,
+      "step": 11239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2919025747319633e-05,
+      "loss": 2.9567,
+      "step": 11240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2879965886995147e-05,
+      "loss": 2.9116,
+      "step": 11241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.284093855965763e-05,
+      "loss": 2.9659,
+      "step": 11242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2801943767968626e-05,
+      "loss": 3.1007,
+      "step": 11243
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2762981514586668e-05,
+      "loss": 2.7958,
+      "step": 11244
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.272405180216869e-05,
+      "loss": 3.1367,
+      "step": 11245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2685154633369064e-05,
+      "loss": 3.2362,
+      "step": 11246
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2646290010840066e-05,
+      "loss": 3.0012,
+      "step": 11247
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2607457937231733e-05,
+      "loss": 2.8764,
+      "step": 11248
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.25686584151919e-05,
+      "loss": 2.7721,
+      "step": 11249
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2529891447366056e-05,
+      "loss": 2.8952,
+      "step": 11250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2491157036397648e-05,
+      "loss": 3.0756,
+      "step": 11251
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2452455184927834e-05,
+      "loss": 2.9511,
+      "step": 11252
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2413785895595505e-05,
+      "loss": 3.0422,
+      "step": 11253
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2375149171037334e-05,
+      "loss": 2.978,
+      "step": 11254
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2336545013888042e-05,
+      "loss": 2.6165,
+      "step": 11255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2297973426779528e-05,
+      "loss": 2.8736,
+      "step": 11256
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2259434412342183e-05,
+      "loss": 2.8703,
+      "step": 11257
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.222092797320363e-05,
+      "loss": 2.8794,
+      "step": 11258
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2182454111989547e-05,
+      "loss": 2.8713,
+      "step": 11259
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2144012831323335e-05,
+      "loss": 3.0746,
+      "step": 11260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2105604133826006e-05,
+      "loss": 2.9512,
+      "step": 11261
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2067228022116804e-05,
+      "loss": 2.8892,
+      "step": 11262
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.202888449881213e-05,
+      "loss": 2.9373,
+      "step": 11263
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1990573566526727e-05,
+      "loss": 2.9686,
+      "step": 11264
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1952295227872676e-05,
+      "loss": 2.9178,
+      "step": 11265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1914049485460164e-05,
+      "loss": 2.7788,
+      "step": 11266
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1875836341896992e-05,
+      "loss": 2.8758,
+      "step": 11267
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1837655799788746e-05,
+      "loss": 2.8579,
+      "step": 11268
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1799507861738787e-05,
+      "loss": 2.8606,
+      "step": 11269
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1761392530348368e-05,
+      "loss": 3.0765,
+      "step": 11270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1723309808216352e-05,
+      "loss": 2.9223,
+      "step": 11271
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.168525969793944e-05,
+      "loss": 2.8439,
+      "step": 11272
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.164724220211217e-05,
+      "loss": 3.012,
+      "step": 11273
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1609257323326803e-05,
+      "loss": 2.9177,
+      "step": 11274
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.157130506417332e-05,
+      "loss": 2.8658,
+      "step": 11275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1533385427239704e-05,
+      "loss": 2.9699,
+      "step": 11276
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.149549841511128e-05,
+      "loss": 2.9722,
+      "step": 11277
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.145764403037176e-05,
+      "loss": 2.8904,
+      "step": 11278
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1419822275601964e-05,
+      "loss": 2.6658,
+      "step": 11279
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1382033153381055e-05,
+      "loss": 2.8314,
+      "step": 11280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.134427666628552e-05,
+      "loss": 2.8991,
+      "step": 11281
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1306552816889978e-05,
+      "loss": 2.8543,
+      "step": 11282
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1268861607766698e-05,
+      "loss": 3.161,
+      "step": 11283
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.123120304148557e-05,
+      "loss": 2.9138,
+      "step": 11284
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1193577120614483e-05,
+      "loss": 2.9761,
+      "step": 11285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.115598384771894e-05,
+      "loss": 2.7962,
+      "step": 11286
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1118423225362393e-05,
+      "loss": 2.9565,
+      "step": 11287
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.108089525610579e-05,
+      "loss": 2.8038,
+      "step": 11288
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1043399942508146e-05,
+      "loss": 2.9326,
+      "step": 11289
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1005937287126076e-05,
+      "loss": 2.8985,
+      "step": 11290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.096850729251404e-05,
+      "loss": 2.9168,
+      "step": 11291
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.093110996122427e-05,
+      "loss": 2.7511,
+      "step": 11292
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0893745295806622e-05,
+      "loss": 2.7241,
+      "step": 11293
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0856413298808997e-05,
+      "loss": 2.9558,
+      "step": 11294
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0819113972776805e-05,
+      "loss": 2.8492,
+      "step": 11295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0781847320253454e-05,
+      "loss": 2.9438,
+      "step": 11296
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0744613343779916e-05,
+      "loss": 3.0406,
+      "step": 11297
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0707412045895048e-05,
+      "loss": 2.9312,
+      "step": 11298
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0670243429135594e-05,
+      "loss": 2.8403,
+      "step": 11299
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.063310749603575e-05,
+      "loss": 2.9705,
+      "step": 11300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.059600424912783e-05,
+      "loss": 3.0478,
+      "step": 11301
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0558933690941585e-05,
+      "loss": 3.0036,
+      "step": 11302
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0521895824004934e-05,
+      "loss": 3.0499,
+      "step": 11303
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0484890650843136e-05,
+      "loss": 2.9973,
+      "step": 11304
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.044791817397956e-05,
+      "loss": 2.7636,
+      "step": 11305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0410978395935253e-05,
+      "loss": 3.1432,
+      "step": 11306
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0374071319228858e-05,
+      "loss": 3.1215,
+      "step": 11307
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0337196946377035e-05,
+      "loss": 2.7698,
+      "step": 11308
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.03003552798941e-05,
+      "loss": 2.8809,
+      "step": 11309
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0263546322292048e-05,
+      "loss": 2.9476,
+      "step": 11310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.022677007608087e-05,
+      "loss": 2.9661,
+      "step": 11311
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0190026543768114e-05,
+      "loss": 2.9652,
+      "step": 11312
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0153315727859223e-05,
+      "loss": 2.671,
+      "step": 11313
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.011663763085725e-05,
+      "loss": 2.8261,
+      "step": 11314
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0079992255263413e-05,
+      "loss": 2.9589,
+      "step": 11315
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0043379603576052e-05,
+      "loss": 2.8335,
+      "step": 11316
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0006799678291997e-05,
+      "loss": 2.8463,
+      "step": 11317
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9970252481905205e-05,
+      "loss": 2.7321,
+      "step": 11318
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9933738016907898e-05,
+      "loss": 3.0649,
+      "step": 11319
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9897256285789645e-05,
+      "loss": 2.6746,
+      "step": 11320
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.986080729103823e-05,
+      "loss": 3.0114,
+      "step": 11321
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9824391035138835e-05,
+      "loss": 3.0139,
+      "step": 11322
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.978800752057458e-05,
+      "loss": 2.8409,
+      "step": 11323
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.975165674982632e-05,
+      "loss": 2.6692,
+      "step": 11324
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9715338725372566e-05,
+      "loss": 2.7917,
+      "step": 11325
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9679053449689898e-05,
+      "loss": 2.7233,
+      "step": 11326
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9642800925252335e-05,
+      "loss": 2.7518,
+      "step": 11327
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.96065811545319e-05,
+      "loss": 2.9951,
+      "step": 11328
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.957039413999817e-05,
+      "loss": 2.9679,
+      "step": 11329
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.953423988411862e-05,
+      "loss": 2.51,
+      "step": 11330
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9498118389358554e-05,
+      "loss": 2.7536,
+      "step": 11331
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9462029658180836e-05,
+      "loss": 2.7548,
+      "step": 11332
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9425973693046328e-05,
+      "loss": 2.7609,
+      "step": 11333
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.938995049641351e-05,
+      "loss": 2.8808,
+      "step": 11334
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9353960070738586e-05,
+      "loss": 2.9772,
+      "step": 11335
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9318002418475756e-05,
+      "loss": 2.8396,
+      "step": 11336
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.928207754207667e-05,
+      "loss": 2.9404,
+      "step": 11337
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9246185443991148e-05,
+      "loss": 2.9916,
+      "step": 11338
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9210326126666234e-05,
+      "loss": 2.8765,
+      "step": 11339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.91744995925473e-05,
+      "loss": 2.7524,
+      "step": 11340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9138705844076952e-05,
+      "loss": 2.9599,
+      "step": 11341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9102944883696128e-05,
+      "loss": 2.7115,
+      "step": 11342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9067216713842983e-05,
+      "loss": 2.8415,
+      "step": 11343
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.903152133695385e-05,
+      "loss": 2.8427,
+      "step": 11344
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8995858755462557e-05,
+      "loss": 3.0428,
+      "step": 11345
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8960228971800885e-05,
+      "loss": 2.9334,
+      "step": 11346
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8924631988398223e-05,
+      "loss": 2.9046,
+      "step": 11347
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8889067807681794e-05,
+      "loss": 2.9155,
+      "step": 11348
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8853536432076602e-05,
+      "loss": 2.7578,
+      "step": 11349
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8818037864005434e-05,
+      "loss": 3.0151,
+      "step": 11350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8782572105888685e-05,
+      "loss": 2.9613,
+      "step": 11351
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.874713916014481e-05,
+      "loss": 3.1168,
+      "step": 11352
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8711739029189656e-05,
+      "loss": 2.8517,
+      "step": 11353
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8676371715437235e-05,
+      "loss": 3.0228,
+      "step": 11354
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8641037221298895e-05,
+      "loss": 2.9946,
+      "step": 11355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8605735549184043e-05,
+      "loss": 3.014,
+      "step": 11356
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.857046670149981e-05,
+      "loss": 2.9909,
+      "step": 11357
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8535230680650882e-05,
+      "loss": 2.8425,
+      "step": 11358
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.850002748904017e-05,
+      "loss": 2.7169,
+      "step": 11359
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.84648571290677e-05,
+      "loss": 2.8063,
+      "step": 11360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8429719603131944e-05,
+      "loss": 3.0441,
+      "step": 11361
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8394614913628426e-05,
+      "loss": 2.7825,
+      "step": 11362
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8359543062951177e-05,
+      "loss": 2.8767,
+      "step": 11363
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8324504053491287e-05,
+      "loss": 2.953,
+      "step": 11364
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8289497887638118e-05,
+      "loss": 2.8611,
+      "step": 11365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8254524567778597e-05,
+      "loss": 2.9461,
+      "step": 11366
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.821958409629737e-05,
+      "loss": 2.9861,
+      "step": 11367
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.818467647557692e-05,
+      "loss": 2.9054,
+      "step": 11368
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.814980170799746e-05,
+      "loss": 3.0115,
+      "step": 11369
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8114959795936914e-05,
+      "loss": 2.9621,
+      "step": 11370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8080150741771106e-05,
+      "loss": 2.8571,
+      "step": 11371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8045374547873528e-05,
+      "loss": 2.9898,
+      "step": 11372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8010631216615335e-05,
+      "loss": 2.9684,
+      "step": 11373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7975920750365583e-05,
+      "loss": 2.9032,
+      "step": 11374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.794124315149115e-05,
+      "loss": 3.0837,
+      "step": 11375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7906598422356425e-05,
+      "loss": 3.0123,
+      "step": 11376
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.787198656532385e-05,
+      "loss": 3.0468,
+      "step": 11377
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7837407582753262e-05,
+      "loss": 2.7153,
+      "step": 11378
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.780286147700272e-05,
+      "loss": 2.8745,
+      "step": 11379
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7768348250427556e-05,
+      "loss": 3.0476,
+      "step": 11380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7733867905381228e-05,
+      "loss": 2.8419,
+      "step": 11381
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7699420444214798e-05,
+      "loss": 2.9445,
+      "step": 11382
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7665005869277108e-05,
+      "loss": 3.1099,
+      "step": 11383
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.763062418291478e-05,
+      "loss": 2.9208,
+      "step": 11384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7596275387472106e-05,
+      "loss": 2.9738,
+      "step": 11385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7561959485291268e-05,
+      "loss": 2.7844,
+      "step": 11386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7527676478712064e-05,
+      "loss": 2.8258,
+      "step": 11387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7493426370072173e-05,
+      "loss": 3.0709,
+      "step": 11388
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.745920916170701e-05,
+      "loss": 2.8509,
+      "step": 11389
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.74250248559496e-05,
+      "loss": 2.9187,
+      "step": 11390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7390873455130907e-05,
+      "loss": 2.9551,
+      "step": 11391
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7356754961579625e-05,
+      "loss": 2.93,
+      "step": 11392
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7322669377622112e-05,
+      "loss": 2.8816,
+      "step": 11393
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.728861670558257e-05,
+      "loss": 3.021,
+      "step": 11394
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.725459694778281e-05,
+      "loss": 3.2405,
+      "step": 11395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.722061010654269e-05,
+      "loss": 2.701,
+      "step": 11396
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7186656184179473e-05,
+      "loss": 3.0169,
+      "step": 11397
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7152735183008473e-05,
+      "loss": 2.9222,
+      "step": 11398
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7118847105342504e-05,
+      "loss": 2.8894,
+      "step": 11399
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7084991953492444e-05,
+      "loss": 2.9099,
+      "step": 11400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7051169729766557e-05,
+      "loss": 2.7765,
+      "step": 11401
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.701738043647122e-05,
+      "loss": 2.909,
+      "step": 11402
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6983624075910253e-05,
+      "loss": 2.8203,
+      "step": 11403
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.694990065038543e-05,
+      "loss": 2.8569,
+      "step": 11404
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6916210162196245e-05,
+      "loss": 2.9179,
+      "step": 11405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6882552613639912e-05,
+      "loss": 2.8873,
+      "step": 11406
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6848928007011377e-05,
+      "loss": 2.902,
+      "step": 11407
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6815336344603415e-05,
+      "loss": 2.8554,
+      "step": 11408
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.678177762870653e-05,
+      "loss": 3.0378,
+      "step": 11409
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6748251861608886e-05,
+      "loss": 2.9165,
+      "step": 11410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.671475904559644e-05,
+      "loss": 3.0002,
+      "step": 11411
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.668129918295319e-05,
+      "loss": 2.7271,
+      "step": 11412
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.664787227596026e-05,
+      "loss": 2.8592,
+      "step": 11413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.661447832689733e-05,
+      "loss": 2.9443,
+      "step": 11414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6581117338040962e-05,
+      "loss": 3.1471,
+      "step": 11415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.654778931166634e-05,
+      "loss": 3.0822,
+      "step": 11416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6514494250045586e-05,
+      "loss": 2.8741,
+      "step": 11417
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6481232155449223e-05,
+      "loss": 2.8687,
+      "step": 11418
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.644800303014521e-05,
+      "loss": 2.8921,
+      "step": 11419
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6414806876399234e-05,
+      "loss": 3.0208,
+      "step": 11420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.638164369647499e-05,
+      "loss": 2.7877,
+      "step": 11421
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6348513492633444e-05,
+      "loss": 2.8527,
+      "step": 11422
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6315416267133954e-05,
+      "loss": 3.0015,
+      "step": 11423
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.628235202223305e-05,
+      "loss": 2.9211,
+      "step": 11424
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624932076018537e-05,
+      "loss": 2.7367,
+      "step": 11425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6216322483243172e-05,
+      "loss": 2.9607,
+      "step": 11426
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.618335719365649e-05,
+      "loss": 2.8169,
+      "step": 11427
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6150424893673078e-05,
+      "loss": 2.8787,
+      "step": 11428
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6117525585538472e-05,
+      "loss": 2.6897,
+      "step": 11429
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.608465927149594e-05,
+      "loss": 2.8599,
+      "step": 11430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.605182595378657e-05,
+      "loss": 2.9614,
+      "step": 11431
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.601902563464902e-05,
+      "loss": 2.8344,
+      "step": 11432
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5986258316319946e-05,
+      "loss": 3.0257,
+      "step": 11433
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.595352400103345e-05,
+      "loss": 2.5684,
+      "step": 11434
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5920822691021854e-05,
+      "loss": 2.9451,
+      "step": 11435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5888154388514598e-05,
+      "loss": 2.9782,
+      "step": 11436
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5855519095739514e-05,
+      "loss": 3.0351,
+      "step": 11437
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5822916814921594e-05,
+      "loss": 2.9413,
+      "step": 11438
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5790347548284178e-05,
+      "loss": 2.9324,
+      "step": 11439
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5757811298047652e-05,
+      "loss": 2.8401,
+      "step": 11440
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5725308066430855e-05,
+      "loss": 3.0184,
+      "step": 11441
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5692837855649953e-05,
+      "loss": 2.7355,
+      "step": 11442
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5660400667919018e-05,
+      "loss": 2.9503,
+      "step": 11443
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.562799650544966e-05,
+      "loss": 2.9396,
+      "step": 11444
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.559562537045156e-05,
+      "loss": 2.885,
+      "step": 11445
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.55632872651319e-05,
+      "loss": 2.8319,
+      "step": 11446
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.553098219169574e-05,
+      "loss": 2.8513,
+      "step": 11447
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5498710152345776e-05,
+      "loss": 2.8787,
+      "step": 11448
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.546647114928257e-05,
+      "loss": 2.8468,
+      "step": 11449
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.543426518470431e-05,
+      "loss": 3.0361,
+      "step": 11450
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.540209226080713e-05,
+      "loss": 2.9133,
+      "step": 11451
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5369952379784612e-05,
+      "loss": 2.8771,
+      "step": 11452
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5337845543828387e-05,
+      "loss": 2.6907,
+      "step": 11453
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5305771755127652e-05,
+      "loss": 3.1747,
+      "step": 11454
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.527373101586932e-05,
+      "loss": 2.9396,
+      "step": 11455
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5241723328238255e-05,
+      "loss": 3.1003,
+      "step": 11456
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5209748694416769e-05,
+      "loss": 2.6986,
+      "step": 11457
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.517780711658534e-05,
+      "loss": 2.6486,
+      "step": 11458
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.514589859692167e-05,
+      "loss": 2.9213,
+      "step": 11459
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.511402313760174e-05,
+      "loss": 2.9461,
+      "step": 11460
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5082180740798757e-05,
+      "loss": 2.8025,
+      "step": 11461
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5050371408684093e-05,
+      "loss": 2.699,
+      "step": 11462
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5018595143426683e-05,
+      "loss": 2.915,
+      "step": 11463
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.498685194719318e-05,
+      "loss": 3.1097,
+      "step": 11464
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4955141822148133e-05,
+      "loss": 2.9992,
+      "step": 11465
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4923464770453588e-05,
+      "loss": 2.9394,
+      "step": 11466
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4891820794269595e-05,
+      "loss": 2.841,
+      "step": 11467
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4860209895753818e-05,
+      "loss": 3.0186,
+      "step": 11468
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4828632077061643e-05,
+      "loss": 2.8228,
+      "step": 11469
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4797087340346294e-05,
+      "loss": 3.0871,
+      "step": 11470
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4765575687758549e-05,
+      "loss": 2.8771,
+      "step": 11471
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4734097121447354e-05,
+      "loss": 2.5843,
+      "step": 11472
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4702651643558774e-05,
+      "loss": 2.7704,
+      "step": 11473
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4671239256237202e-05,
+      "loss": 2.9431,
+      "step": 11474
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4639859961624368e-05,
+      "loss": 2.7719,
+      "step": 11475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4608513761860066e-05,
+      "loss": 2.9541,
+      "step": 11476
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.457720065908147e-05,
+      "loss": 2.8084,
+      "step": 11477
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4545920655423871e-05,
+      "loss": 2.9109,
+      "step": 11478
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4514673753020069e-05,
+      "loss": 2.7168,
+      "step": 11479
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4483459954000688e-05,
+      "loss": 2.9578,
+      "step": 11480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4452279260494083e-05,
+      "loss": 3.038,
+      "step": 11481
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4421131674626275e-05,
+      "loss": 2.8761,
+      "step": 11482
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4390017198521176e-05,
+      "loss": 2.8543,
+      "step": 11483
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4358935834300368e-05,
+      "loss": 3.1497,
+      "step": 11484
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4327887584083099e-05,
+      "loss": 3.039,
+      "step": 11485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.429687244998651e-05,
+      "loss": 2.7778,
+      "step": 11486
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.426589043412535e-05,
+      "loss": 2.8434,
+      "step": 11487
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4234941538612212e-05,
+      "loss": 2.8897,
+      "step": 11488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4204025765557294e-05,
+      "loss": 3.0403,
+      "step": 11489
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4173143117068743e-05,
+      "loss": 2.8566,
+      "step": 11490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.414229359525221e-05,
+      "loss": 2.7975,
+      "step": 11491
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4111477202211343e-05,
+      "loss": 3.0293,
+      "step": 11492
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.408069394004724e-05,
+      "loss": 2.8261,
+      "step": 11493
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4049943810858945e-05,
+      "loss": 2.9711,
+      "step": 11494
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.401922681674328e-05,
+      "loss": 2.8766,
+      "step": 11495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3988542959794625e-05,
+      "loss": 2.7964,
+      "step": 11496
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3957892242105253e-05,
+      "loss": 2.9601,
+      "step": 11497
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3927274665765044e-05,
+      "loss": 2.8566,
+      "step": 11498
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.389669023286172e-05,
+      "loss": 2.9103,
+      "step": 11499
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3866138945480778e-05,
+      "loss": 3.0071,
+      "step": 11500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3835620805705385e-05,
+      "loss": 2.8583,
+      "step": 11501
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3805135815616377e-05,
+      "loss": 2.9149,
+      "step": 11502
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3774683977292424e-05,
+      "loss": 2.8531,
+      "step": 11503
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3744265292809977e-05,
+      "loss": 2.8143,
+      "step": 11504
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3713879764243153e-05,
+      "loss": 2.8565,
+      "step": 11505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3683527393663798e-05,
+      "loss": 3.1022,
+      "step": 11506
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3653208183141586e-05,
+      "loss": 2.8335,
+      "step": 11507
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3622922134743753e-05,
+      "loss": 2.9683,
+      "step": 11508
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3592669250535483e-05,
+      "loss": 2.922,
+      "step": 11509
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3562449532579512e-05,
+      "loss": 2.9151,
+      "step": 11510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3532262982936638e-05,
+      "loss": 2.6508,
+      "step": 11511
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3502109603664825e-05,
+      "loss": 2.7635,
+      "step": 11512
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3471989396820428e-05,
+      "loss": 2.9839,
+      "step": 11513
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.344190236445697e-05,
+      "loss": 2.7161,
+      "step": 11514
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3411848508626145e-05,
+      "loss": 3.047,
+      "step": 11515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3381827831377202e-05,
+      "loss": 2.8607,
+      "step": 11516
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.335184033475706e-05,
+      "loss": 3.048,
+      "step": 11517
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.332188602081047e-05,
+      "loss": 2.8581,
+      "step": 11518
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3291964891579912e-05,
+      "loss": 2.7738,
+      "step": 11519
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3262076949105695e-05,
+      "loss": 2.6698,
+      "step": 11520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3232222195425526e-05,
+      "loss": 2.9451,
+      "step": 11521
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3202400632575273e-05,
+      "loss": 2.9885,
+      "step": 11522
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3172612262588312e-05,
+      "loss": 2.775,
+      "step": 11523
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3142857087495796e-05,
+      "loss": 2.8889,
+      "step": 11524
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3113135109326602e-05,
+      "loss": 3.0218,
+      "step": 11525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3083446330107384e-05,
+      "loss": 2.6397,
+      "step": 11526
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3053790751862471e-05,
+      "loss": 3.196,
+      "step": 11527
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.302416837661402e-05,
+      "loss": 2.8185,
+      "step": 11528
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2994579206381807e-05,
+      "loss": 3.0793,
+      "step": 11529
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2965023243183439e-05,
+      "loss": 2.805,
+      "step": 11530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2935500489034135e-05,
+      "loss": 2.8946,
+      "step": 11531
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2906010945947123e-05,
+      "loss": 2.9919,
+      "step": 11532
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2876554615932956e-05,
+      "loss": 2.8055,
+      "step": 11533
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.284713150100042e-05,
+      "loss": 2.9304,
+      "step": 11534
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2817741603155463e-05,
+      "loss": 2.9645,
+      "step": 11535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2788384924402374e-05,
+      "loss": 3.3367,
+      "step": 11536
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.275906146674255e-05,
+      "loss": 2.7187,
+      "step": 11537
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2729771232175724e-05,
+      "loss": 2.8986,
+      "step": 11538
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.270051422269891e-05,
+      "loss": 2.799,
+      "step": 11539
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2671290440307126e-05,
+      "loss": 3.0596,
+      "step": 11540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2642099886992996e-05,
+      "loss": 2.9622,
+      "step": 11541
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2612942564746931e-05,
+      "loss": 2.8177,
+      "step": 11542
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2583818475557008e-05,
+      "loss": 2.968,
+      "step": 11543
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2554727621409134e-05,
+      "loss": 2.9682,
+      "step": 11544
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2525670004286894e-05,
+      "loss": 2.8478,
+      "step": 11545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2496645626171588e-05,
+      "loss": 3.0376,
+      "step": 11546
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2467654489042302e-05,
+      "loss": 2.8303,
+      "step": 11547
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2438696594875842e-05,
+      "loss": 2.7501,
+      "step": 11548
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2409771945646686e-05,
+      "loss": 2.8292,
+      "step": 11549
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.23808805433272e-05,
+      "loss": 2.9732,
+      "step": 11550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2352022389887251e-05,
+      "loss": 2.8439,
+      "step": 11551
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2323197487294657e-05,
+      "loss": 2.878,
+      "step": 11552
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2294405837514844e-05,
+      "loss": 2.9749,
+      "step": 11553
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2265647442510908e-05,
+      "loss": 2.9711,
+      "step": 11554
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2236922304244002e-05,
+      "loss": 2.8068,
+      "step": 11555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2208230424672562e-05,
+      "loss": 2.9185,
+      "step": 11556
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2179571805753132e-05,
+      "loss": 2.7844,
+      "step": 11557
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2150946449439704e-05,
+      "loss": 3.1177,
+      "step": 11558
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2122354357684218e-05,
+      "loss": 2.8319,
+      "step": 11559
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2093795532436224e-05,
+      "loss": 3.1141,
+      "step": 11560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2065269975643111e-05,
+      "loss": 2.7987,
+      "step": 11561
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2036777689249823e-05,
+      "loss": 2.925,
+      "step": 11562
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2008318675199137e-05,
+      "loss": 2.9564,
+      "step": 11563
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1979892935431669e-05,
+      "loss": 3.1041,
+      "step": 11564
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.195150047188559e-05,
+      "loss": 3.0027,
+      "step": 11565
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1923141286496909e-05,
+      "loss": 2.713,
+      "step": 11566
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1894815381199298e-05,
+      "loss": 2.9171,
+      "step": 11567
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1866522757924215e-05,
+      "loss": 2.8541,
+      "step": 11568
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1838263418600781e-05,
+      "loss": 2.7961,
+      "step": 11569
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1810037365155902e-05,
+      "loss": 2.8169,
+      "step": 11570
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1781844599514313e-05,
+      "loss": 3.1147,
+      "step": 11571
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1753685123598146e-05,
+      "loss": 3.0087,
+      "step": 11572
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1725558939327751e-05,
+      "loss": 3.1026,
+      "step": 11573
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1697466048620763e-05,
+      "loss": 2.9615,
+      "step": 11574
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1669406453392817e-05,
+      "loss": 3.0805,
+      "step": 11575
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1641380155557101e-05,
+      "loss": 2.9144,
+      "step": 11576
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1613387157024757e-05,
+      "loss": 2.7039,
+      "step": 11577
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1585427459704367e-05,
+      "loss": 2.7685,
+      "step": 11578
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1557501065502518e-05,
+      "loss": 3.0176,
+      "step": 11579
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1529607976323353e-05,
+      "loss": 3.0773,
+      "step": 11580
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.150174819406874e-05,
+      "loss": 2.7955,
+      "step": 11581
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1473921720638436e-05,
+      "loss": 2.6753,
+      "step": 11582
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1446128557929757e-05,
+      "loss": 2.8416,
+      "step": 11583
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1418368707837856e-05,
+      "loss": 2.6961,
+      "step": 11584
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.139064217225555e-05,
+      "loss": 2.7114,
+      "step": 11585
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1362948953073382e-05,
+      "loss": 2.548,
+      "step": 11586
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1335289052179621e-05,
+      "loss": 2.9502,
+      "step": 11587
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1307662471460368e-05,
+      "loss": 2.8944,
+      "step": 11588
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1280069212799338e-05,
+      "loss": 2.7047,
+      "step": 11589
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1252509278078027e-05,
+      "loss": 2.7601,
+      "step": 11590
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1224982669175543e-05,
+      "loss": 3.0345,
+      "step": 11591
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1197489387968995e-05,
+      "loss": 2.9645,
+      "step": 11592
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1170029436332884e-05,
+      "loss": 2.6111,
+      "step": 11593
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1142602816139768e-05,
+      "loss": 2.8592,
+      "step": 11594
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.111520952925954e-05,
+      "loss": 2.7839,
+      "step": 11595
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.108784957756026e-05,
+      "loss": 2.9358,
+      "step": 11596
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1060522962907272e-05,
+      "loss": 2.7309,
+      "step": 11597
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1033229687164082e-05,
+      "loss": 2.8744,
+      "step": 11598
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1005969752191647e-05,
+      "loss": 2.7551,
+      "step": 11599
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0978743159848703e-05,
+      "loss": 2.8944,
+      "step": 11600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.095154991199171e-05,
+      "loss": 2.8566,
+      "step": 11601
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0924390010474905e-05,
+      "loss": 2.9942,
+      "step": 11602
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0897263457150197e-05,
+      "loss": 2.6913,
+      "step": 11603
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0870170253867273e-05,
+      "loss": 2.8288,
+      "step": 11604
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0843110402473433e-05,
+      "loss": 2.9981,
+      "step": 11605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0816083904813867e-05,
+      "loss": 3.0975,
+      "step": 11606
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0789090762731379e-05,
+      "loss": 2.9291,
+      "step": 11607
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0762130978066608e-05,
+      "loss": 2.9173,
+      "step": 11608
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.073520455265764e-05,
+      "loss": 3.1345,
+      "step": 11609
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0708311488340671e-05,
+      "loss": 2.7987,
+      "step": 11610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0681451786949348e-05,
+      "loss": 2.8473,
+      "step": 11611
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0654625450315203e-05,
+      "loss": 3.0919,
+      "step": 11612
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.062783248026733e-05,
+      "loss": 2.8175,
+      "step": 11613
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0601072878632656e-05,
+      "loss": 2.7237,
+      "step": 11614
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0574346647235889e-05,
+      "loss": 2.683,
+      "step": 11615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0547653787899347e-05,
+      "loss": 2.8154,
+      "step": 11616
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0520994302443077e-05,
+      "loss": 3.0298,
+      "step": 11617
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0494368192684955e-05,
+      "loss": 2.8284,
+      "step": 11618
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.046777546044042e-05,
+      "loss": 3.1387,
+      "step": 11619
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0441216107522855e-05,
+      "loss": 3.1055,
+      "step": 11620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0414690135743087e-05,
+      "loss": 3.1047,
+      "step": 11621
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.038819754690995e-05,
+      "loss": 2.8978,
+      "step": 11622
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.036173834282983e-05,
+      "loss": 2.6256,
+      "step": 11623
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0335312525306895e-05,
+      "loss": 2.7826,
+      "step": 11624
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0308920096142981e-05,
+      "loss": 2.7875,
+      "step": 11625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0282561057137707e-05,
+      "loss": 2.6939,
+      "step": 11626
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0256235410088355e-05,
+      "loss": 3.2268,
+      "step": 11627
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0229943156790045e-05,
+      "loss": 2.9242,
+      "step": 11628
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0203684299035509e-05,
+      "loss": 2.9335,
+      "step": 11629
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0177458838615205e-05,
+      "loss": 2.7859,
+      "step": 11630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0151266777317481e-05,
+      "loss": 3.0927,
+      "step": 11631
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0125108116928017e-05,
+      "loss": 2.7235,
+      "step": 11632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0098982859230776e-05,
+      "loss": 2.7761,
+      "step": 11633
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.007289100600689e-05,
+      "loss": 2.8049,
+      "step": 11634
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0046832559035546e-05,
+      "loss": 2.9161,
+      "step": 11635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0020807520093656e-05,
+      "loss": 2.9213,
+      "step": 11636
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.994815890955689e-06,
+      "loss": 2.6345,
+      "step": 11637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.968857673393895e-06,
+      "loss": 2.8905,
+      "step": 11638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.9429328691783e-06,
+      "loss": 2.7886,
+      "step": 11639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.917041480076605e-06,
+      "loss": 2.7365,
+      "step": 11640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.891183507854285e-06,
+      "loss": 2.7064,
+      "step": 11641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.865358954274429e-06,
+      "loss": 3.0247,
+      "step": 11642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.839567821097962e-06,
+      "loss": 2.71,
+      "step": 11643
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.813810110083476e-06,
+      "loss": 2.9168,
+      "step": 11644
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.788085822987236e-06,
+      "loss": 2.9948,
+      "step": 11645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.762394961563337e-06,
+      "loss": 2.9473,
+      "step": 11646
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.7367375275636e-06,
+      "loss": 2.762,
+      "step": 11647
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.711113522737347e-06,
+      "loss": 2.9669,
+      "step": 11648
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.68552294883196e-06,
+      "loss": 2.9357,
+      "step": 11649
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.659965807592152e-06,
+      "loss": 2.8207,
+      "step": 11650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.63444210076081e-06,
+      "loss": 2.6877,
+      "step": 11651
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.60895183007815e-06,
+      "loss": 2.9835,
+      "step": 11652
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.583494997282227e-06,
+      "loss": 2.9953,
+      "step": 11653
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.558071604108987e-06,
+      "loss": 2.9666,
+      "step": 11654
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.53268165229182e-06,
+      "loss": 3.001,
+      "step": 11655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.507325143562062e-06,
+      "loss": 3.0294,
+      "step": 11656
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.482002079648499e-06,
+      "loss": 2.782,
+      "step": 11657
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.456712462278084e-06,
+      "loss": 2.8156,
+      "step": 11658
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.431456293175044e-06,
+      "loss": 3.0392,
+      "step": 11659
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.406233574061506e-06,
+      "loss": 3.2153,
+      "step": 11660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.381044306657371e-06,
+      "loss": 2.7201,
+      "step": 11661
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.355888492680153e-06,
+      "loss": 3.0305,
+      "step": 11662
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.330766133845147e-06,
+      "loss": 3.0963,
+      "step": 11663
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.305677231865373e-06,
+      "loss": 2.7328,
+      "step": 11664
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.280621788451515e-06,
+      "loss": 3.0537,
+      "step": 11665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.255599805312044e-06,
+      "loss": 2.7785,
+      "step": 11666
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.230611284152978e-06,
+      "loss": 3.0407,
+      "step": 11667
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.20565622667846e-06,
+      "loss": 2.7697,
+      "step": 11668
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.180734634589794e-06,
+      "loss": 3.106,
+      "step": 11669
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.155846509586451e-06,
+      "loss": 2.7249,
+      "step": 11670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.130991853365411e-06,
+      "loss": 3.0619,
+      "step": 11671
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.106170667621483e-06,
+      "loss": 3.092,
+      "step": 11672
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.081382954046924e-06,
+      "loss": 2.8296,
+      "step": 11673
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.05662871433216e-06,
+      "loss": 2.8412,
+      "step": 11674
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.031907950165009e-06,
+      "loss": 3.0663,
+      "step": 11675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.007220663231008e-06,
+      "loss": 2.9691,
+      "step": 11676
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.982566855213591e-06,
+      "loss": 2.7126,
+      "step": 11677
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.957946527793692e-06,
+      "loss": 2.9398,
+      "step": 11678
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.93335968265019e-06,
+      "loss": 2.9862,
+      "step": 11679
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.908806321459517e-06,
+      "loss": 2.8188,
+      "step": 11680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.884286445895895e-06,
+      "loss": 3.112,
+      "step": 11681
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.859800057631206e-06,
+      "loss": 2.8036,
+      "step": 11682
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.835347158335117e-06,
+      "loss": 3.0337,
+      "step": 11683
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.810927749674902e-06,
+      "loss": 2.9634,
+      "step": 11684
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.786541833315732e-06,
+      "loss": 2.9371,
+      "step": 11685
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.762189410920328e-06,
+      "loss": 2.9563,
+      "step": 11686
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.737870484149202e-06,
+      "loss": 2.9868,
+      "step": 11687
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.713585054660577e-06,
+      "loss": 3.1477,
+      "step": 11688
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.689333124110354e-06,
+      "loss": 3.119,
+      "step": 11689
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.665114694152209e-06,
+      "loss": 2.8244,
+      "step": 11690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.640929766437544e-06,
+      "loss": 2.8309,
+      "step": 11691
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.616778342615317e-06,
+      "loss": 2.8197,
+      "step": 11692
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.592660424332433e-06,
+      "loss": 2.9361,
+      "step": 11693
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.568576013233298e-06,
+      "loss": 2.9294,
+      "step": 11694
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.544525110960267e-06,
+      "loss": 2.7235,
+      "step": 11695
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.520507719153192e-06,
+      "loss": 3.0492,
+      "step": 11696
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.496523839449767e-06,
+      "loss": 2.918,
+      "step": 11697
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.47257347348529e-06,
+      "loss": 2.7739,
+      "step": 11698
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.448656622892904e-06,
+      "loss": 2.7609,
+      "step": 11699
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.42477328930341e-06,
+      "loss": 2.9116,
+      "step": 11700
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.40092347434529e-06,
+      "loss": 2.9365,
+      "step": 11701
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.377107179644794e-06,
+      "loss": 3.0739,
+      "step": 11702
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.353324406825902e-06,
+      "loss": 2.8519,
+      "step": 11703
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.32957515751015e-06,
+      "loss": 2.8431,
+      "step": 11704
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.305859433317076e-06,
+      "loss": 2.8054,
+      "step": 11705
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.282177235863553e-06,
+      "loss": 2.8639,
+      "step": 11706
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.258528566764623e-06,
+      "loss": 2.9628,
+      "step": 11707
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.234913427632551e-06,
+      "loss": 2.605,
+      "step": 11708
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.211331820077828e-06,
+      "loss": 3.1242,
+      "step": 11709
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.18778374570811e-06,
+      "loss": 3.0105,
+      "step": 11710
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.164269206129282e-06,
+      "loss": 2.9481,
+      "step": 11711
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.140788202944616e-06,
+      "loss": 2.771,
+      "step": 11712
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.117340737755164e-06,
+      "loss": 2.8178,
+      "step": 11713
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.093926812159757e-06,
+      "loss": 3.0365,
+      "step": 11714
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.070546427754899e-06,
+      "loss": 2.8473,
+      "step": 11715
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.047199586134869e-06,
+      "loss": 2.8198,
+      "step": 11716
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.023886288891391e-06,
+      "loss": 3.0414,
+      "step": 11717
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.000606537614364e-06,
+      "loss": 2.9861,
+      "step": 11718
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.977360333891014e-06,
+      "loss": 3.0982,
+      "step": 11719
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.954147679306411e-06,
+      "loss": 2.9444,
+      "step": 11720
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.930968575443342e-06,
+      "loss": 2.9775,
+      "step": 11721
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.907823023882377e-06,
+      "loss": 3.1382,
+      "step": 11722
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.884711026201585e-06,
+      "loss": 3.0133,
+      "step": 11723
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.86163258397704e-06,
+      "loss": 2.8103,
+      "step": 11724
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.838587698782262e-06,
+      "loss": 2.8678,
+      "step": 11725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.815576372188604e-06,
+      "loss": 2.7269,
+      "step": 11726
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.792598605765089e-06,
+      "loss": 2.9139,
+      "step": 11727
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.76965440107863e-06,
+      "loss": 2.8834,
+      "step": 11728
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.746743759693531e-06,
+      "loss": 2.8279,
+      "step": 11729
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.723866683172155e-06,
+      "loss": 2.7738,
+      "step": 11730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.701023173074251e-06,
+      "loss": 3.0502,
+      "step": 11731
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.67821323095752e-06,
+      "loss": 3.0037,
+      "step": 11732
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.655436858377163e-06,
+      "loss": 3.0136,
+      "step": 11733
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.63269405688638e-06,
+      "loss": 2.8707,
+      "step": 11734
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.6099848280357655e-06,
+      "loss": 2.8875,
+      "step": 11735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.587309173373913e-06,
+      "loss": 2.7531,
+      "step": 11736
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.56466709444692e-06,
+      "loss": 2.7212,
+      "step": 11737
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.542058592798606e-06,
+      "loss": 2.9748,
+      "step": 11738
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.519483669970683e-06,
+      "loss": 2.8462,
+      "step": 11739
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.496942327502365e-06,
+      "loss": 2.8701,
+      "step": 11740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.474434566930644e-06,
+      "loss": 3.1437,
+      "step": 11741
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.451960389790291e-06,
+      "loss": 2.8597,
+      "step": 11742
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.429519797613693e-06,
+      "loss": 2.755,
+      "step": 11743
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.40711279193107e-06,
+      "loss": 2.8069,
+      "step": 11744
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.384739374270089e-06,
+      "loss": 2.6836,
+      "step": 11745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.362399546156584e-06,
+      "loss": 2.8012,
+      "step": 11746
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.3400933091135604e-06,
+      "loss": 3.0367,
+      "step": 11747
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.317820664662133e-06,
+      "loss": 3.1246,
+      "step": 11748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.2955816143209786e-06,
+      "loss": 3.1781,
+      "step": 11749
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.273376159606382e-06,
+      "loss": 3.1406,
+      "step": 11750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.251204302032633e-06,
+      "loss": 2.908,
+      "step": 11751
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.229066043111354e-06,
+      "loss": 2.8027,
+      "step": 11752
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.206961384352284e-06,
+      "loss": 3.0198,
+      "step": 11753
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.1848903272623835e-06,
+      "loss": 2.866,
+      "step": 11754
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.162852873346836e-06,
+      "loss": 2.9448,
+      "step": 11755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.140849024108165e-06,
+      "loss": 2.7616,
+      "step": 11756
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.118878781046834e-06,
+      "loss": 3.0234,
+      "step": 11757
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.0969421456607565e-06,
+      "loss": 2.7379,
+      "step": 11758
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.075039119445848e-06,
+      "loss": 2.9852,
+      "step": 11759
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.0531697038954675e-06,
+      "loss": 2.9574,
+      "step": 11760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.031333900500925e-06,
+      "loss": 2.9618,
+      "step": 11761
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.009531710751027e-06,
+      "loss": 2.872,
+      "step": 11762
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.987763136132475e-06,
+      "loss": 2.7745,
+      "step": 11763
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.966028178129468e-06,
+      "loss": 3.0855,
+      "step": 11764
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.944326838224102e-06,
+      "loss": 2.7899,
+      "step": 11765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.92265911789608e-06,
+      "loss": 2.8534,
+      "step": 11766
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.9010250186229415e-06,
+      "loss": 2.7384,
+      "step": 11767
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.879424541879676e-06,
+      "loss": 2.8032,
+      "step": 11768
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.8578576891392685e-06,
+      "loss": 2.8852,
+      "step": 11769
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.8363244618721566e-06,
+      "loss": 2.6869,
+      "step": 11770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.81482486154672e-06,
+      "loss": 2.8345,
+      "step": 11771
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.793358889628898e-06,
+      "loss": 2.4867,
+      "step": 11772
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.771926547582352e-06,
+      "loss": 3.0161,
+      "step": 11773
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.750527836868469e-06,
+      "loss": 3.1757,
+      "step": 11774
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.729162758946417e-06,
+      "loss": 2.6963,
+      "step": 11775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.707831315272861e-06,
+      "loss": 2.9764,
+      "step": 11776
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.686533507302417e-06,
+      "loss": 2.9869,
+      "step": 11777
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.665269336487312e-06,
+      "loss": 3.0645,
+      "step": 11778
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.6440388042773855e-06,
+      "loss": 2.8692,
+      "step": 11779
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.62284191212037e-06,
+      "loss": 3.0541,
+      "step": 11780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.601678661461497e-06,
+      "loss": 2.8895,
+      "step": 11781
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.580549053743834e-06,
+      "loss": 2.8948,
+      "step": 11782
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.559453090408174e-06,
+      "loss": 2.9494,
+      "step": 11783
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.538390772892977e-06,
+      "loss": 2.8374,
+      "step": 11784
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.517362102634316e-06,
+      "loss": 2.9974,
+      "step": 11785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.496367081066101e-06,
+      "loss": 2.8282,
+      "step": 11786
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.475405709619908e-06,
+      "loss": 2.9219,
+      "step": 11787
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.4544779897250935e-06,
+      "loss": 2.6508,
+      "step": 11788
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.433583922808461e-06,
+      "loss": 2.7878,
+      "step": 11789
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.412723510294871e-06,
+      "loss": 2.7842,
+      "step": 11790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.391896753606574e-06,
+      "loss": 2.708,
+      "step": 11791
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.371103654163768e-06,
+      "loss": 2.8794,
+      "step": 11792
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.350344213384207e-06,
+      "loss": 2.7544,
+      "step": 11793
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.32961843268337e-06,
+      "loss": 2.7956,
+      "step": 11794
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.308926313474572e-06,
+      "loss": 2.7074,
+      "step": 11795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.288267857168628e-06,
+      "loss": 2.6369,
+      "step": 11796
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.267643065174189e-06,
+      "loss": 2.9509,
+      "step": 11797
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.24705193889763e-06,
+      "loss": 2.9697,
+      "step": 11798
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.226494479742884e-06,
+      "loss": 2.9244,
+      "step": 11799
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.205970689111773e-06,
+      "loss": 2.7401,
+      "step": 11800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.185480568403678e-06,
+      "loss": 2.7843,
+      "step": 11801
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.165024119015761e-06,
+      "loss": 2.9027,
+      "step": 11802
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.144601342342848e-06,
+      "loss": 2.837,
+      "step": 11803
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.124212239777605e-06,
+      "loss": 2.7524,
+      "step": 11804
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.10385681271014e-06,
+      "loss": 2.9641,
+      "step": 11805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.083535062528511e-06,
+      "loss": 2.8855,
+      "step": 11806
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.063246990618277e-06,
+      "loss": 2.8921,
+      "step": 11807
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.042992598362884e-06,
+      "loss": 3.0079,
+      "step": 11808
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.022771887143397e-06,
+      "loss": 2.7407,
+      "step": 11809
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.0025848583385445e-06,
+      "loss": 2.9853,
+      "step": 11810
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.982431513324837e-06,
+      "loss": 3.0581,
+      "step": 11811
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.96231185347651e-06,
+      "loss": 2.97,
+      "step": 11812
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.942225880165297e-06,
+      "loss": 2.8843,
+      "step": 11813
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.9221735947609375e-06,
+      "loss": 2.8418,
+      "step": 11814
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.902154998630616e-06,
+      "loss": 3.0576,
+      "step": 11815
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.882170093139405e-06,
+      "loss": 2.8795,
+      "step": 11816
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8622188796499385e-06,
+      "loss": 3.0861,
+      "step": 11817
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.842301359522628e-06,
+      "loss": 2.7548,
+      "step": 11818
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.822417534115554e-06,
+      "loss": 2.8988,
+      "step": 11819
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.80256740478452e-06,
+      "loss": 3.024,
+      "step": 11820
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.782750972883111e-06,
+      "loss": 2.9127,
+      "step": 11821
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.762968239762467e-06,
+      "loss": 2.8075,
+      "step": 11822
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.743219206771455e-06,
+      "loss": 2.8417,
+      "step": 11823
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.723503875256775e-06,
+      "loss": 2.881,
+      "step": 11824
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.703822246562684e-06,
+      "loss": 2.9863,
+      "step": 11825
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.684174322031166e-06,
+      "loss": 2.8512,
+      "step": 11826
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.664560103002037e-06,
+      "loss": 2.8753,
+      "step": 11827
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6449795908126155e-06,
+      "loss": 2.9954,
+      "step": 11828
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.625432786798168e-06,
+      "loss": 2.7658,
+      "step": 11829
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6059196922912945e-06,
+      "loss": 2.8323,
+      "step": 11830
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.586440308622709e-06,
+      "loss": 3.0319,
+      "step": 11831
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.56699463712057e-06,
+      "loss": 2.8643,
+      "step": 11832
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.547582679110763e-06,
+      "loss": 3.0049,
+      "step": 11833
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.52820443591695e-06,
+      "loss": 3.0561,
+      "step": 11834
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.50885990886052e-06,
+      "loss": 2.8324,
+      "step": 11835
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.48954909926036e-06,
+      "loss": 2.7287,
+      "step": 11836
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.470272008433363e-06,
+      "loss": 2.8197,
+      "step": 11837
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.451028637693811e-06,
+      "loss": 3.0569,
+      "step": 11838
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.431818988353931e-06,
+      "loss": 3.18,
+      "step": 11839
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.412643061723566e-06,
+      "loss": 2.9356,
+      "step": 11840
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.393500859110168e-06,
+      "loss": 2.8853,
+      "step": 11841
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.374392381818971e-06,
+      "loss": 3.05,
+      "step": 11842
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.355317631153045e-06,
+      "loss": 2.923,
+      "step": 11843
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.336276608412905e-06,
+      "loss": 3.2077,
+      "step": 11844
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3172693148969e-06,
+      "loss": 2.7388,
+      "step": 11845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.29829575190105e-06,
+      "loss": 3.0627,
+      "step": 11846
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.279355920719153e-06,
+      "loss": 2.9802,
+      "step": 11847
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.260449822642621e-06,
+      "loss": 2.9168,
+      "step": 11848
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.241577458960534e-06,
+      "loss": 2.9303,
+      "step": 11849
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.222738830959806e-06,
+      "loss": 2.8544,
+      "step": 11850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.203933939924854e-06,
+      "loss": 2.9677,
+      "step": 11851
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.185162787138098e-06,
+      "loss": 2.8531,
+      "step": 11852
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.166425373879291e-06,
+      "loss": 3.0195,
+      "step": 11853
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1477217014261335e-06,
+      "loss": 2.7784,
+      "step": 11854
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.1290517710539384e-06,
+      "loss": 2.9262,
+      "step": 11855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.110415584035799e-06,
+      "loss": 2.8363,
+      "step": 11856
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0918131416424205e-06,
+      "loss": 3.1257,
+      "step": 11857
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.073244445142178e-06,
+      "loss": 2.8087,
+      "step": 11858
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.05470949580128e-06,
+      "loss": 2.9459,
+      "step": 11859
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0362082948834934e-06,
+      "loss": 3.1239,
+      "step": 11860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.017740843650365e-06,
+      "loss": 2.8003,
+      "step": 11861
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9993071433611095e-06,
+      "loss": 2.9114,
+      "step": 11862
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.980907195272611e-06,
+      "loss": 2.984,
+      "step": 11863
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.962541000639642e-06,
+      "loss": 2.9247,
+      "step": 11864
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.944208560714314e-06,
+      "loss": 2.65,
+      "step": 11865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.925909876746848e-06,
+      "loss": 2.6641,
+      "step": 11866
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.907644949984747e-06,
+      "loss": 2.868,
+      "step": 11867
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.889413781673679e-06,
+      "loss": 2.9427,
+      "step": 11868
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.871216373056486e-06,
+      "loss": 2.9389,
+      "step": 11869
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.853052725374174e-06,
+      "loss": 2.7041,
+      "step": 11870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.834922839865252e-06,
+      "loss": 2.8677,
+      "step": 11871
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.816826717765788e-06,
+      "loss": 2.996,
+      "step": 11872
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.798764360309793e-06,
+      "loss": 2.8399,
+      "step": 11873
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.780735768728895e-06,
+      "loss": 2.8807,
+      "step": 11874
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.762740944252275e-06,
+      "loss": 3.0574,
+      "step": 11875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.744779888107065e-06,
+      "loss": 2.9614,
+      "step": 11876
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7268526015178394e-06,
+      "loss": 3.0122,
+      "step": 11877
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.708959085707065e-06,
+      "loss": 3.018,
+      "step": 11878
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6910993418948225e-06,
+      "loss": 2.7043,
+      "step": 11879
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6732733712989136e-06,
+      "loss": 3.0351,
+      "step": 11880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.655481175134757e-06,
+      "loss": 2.6184,
+      "step": 11881
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.637722754615603e-06,
+      "loss": 3.0385,
+      "step": 11882
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.619998110952317e-06,
+      "loss": 2.9396,
+      "step": 11883
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.602307245353488e-06,
+      "loss": 2.9312,
+      "step": 11884
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.584650159025317e-06,
+      "loss": 2.7769,
+      "step": 11885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5670268531717854e-06,
+      "loss": 2.8132,
+      "step": 11886
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.54943732899471e-06,
+      "loss": 2.9487,
+      "step": 11887
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.531881587693243e-06,
+      "loss": 2.8353,
+      "step": 11888
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.514359630464593e-06,
+      "loss": 2.89,
+      "step": 11889
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.496871458503415e-06,
+      "loss": 2.9881,
+      "step": 11890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.479417073002201e-06,
+      "loss": 2.8974,
+      "step": 11891
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.461996475151165e-06,
+      "loss": 2.9126,
+      "step": 11892
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.444609666138022e-06,
+      "loss": 2.7173,
+      "step": 11893
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4272566471484365e-06,
+      "loss": 2.9326,
+      "step": 11894
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.409937419365517e-06,
+      "loss": 2.9457,
+      "step": 11895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.39265198397032e-06,
+      "loss": 2.964,
+      "step": 11896
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3754003421414005e-06,
+      "loss": 3.1325,
+      "step": 11897
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.358182495055097e-06,
+      "loss": 2.9008,
+      "step": 11898
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.340998443885469e-06,
+      "loss": 2.8569,
+      "step": 11899
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.323848189804136e-06,
+      "loss": 2.8628,
+      "step": 11900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.306731733980607e-06,
+      "loss": 2.8439,
+      "step": 11901
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.289649077581892e-06,
+      "loss": 2.8499,
+      "step": 11902
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.272600221772893e-06,
+      "loss": 3.1024,
+      "step": 11903
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.255585167716014e-06,
+      "loss": 2.8524,
+      "step": 11904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.238603916571549e-06,
+      "loss": 2.9748,
+      "step": 11905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.221656469497292e-06,
+      "loss": 2.9387,
+      "step": 11906
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.204742827648877e-06,
+      "loss": 2.9498,
+      "step": 11907
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.187862992179548e-06,
+      "loss": 2.8056,
+      "step": 11908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.171016964240271e-06,
+      "loss": 2.8742,
+      "step": 11909
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.154204744979795e-06,
+      "loss": 2.9272,
+      "step": 11910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.137426335544369e-06,
+      "loss": 2.8653,
+      "step": 11911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.12068173707808e-06,
+      "loss": 2.9877,
+      "step": 11912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.103970950722735e-06,
+      "loss": 3.0839,
+      "step": 11913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.087293977617757e-06,
+      "loss": 3.0989,
+      "step": 11914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.070650818900234e-06,
+      "loss": 2.839,
+      "step": 11915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.054041475704984e-06,
+      "loss": 2.9262,
+      "step": 11916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.037465949164654e-06,
+      "loss": 2.9542,
+      "step": 11917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.020924240409396e-06,
+      "loss": 2.7127,
+      "step": 11918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.004416350567086e-06,
+      "loss": 2.7281,
+      "step": 11919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9879422807633796e-06,
+      "loss": 2.9669,
+      "step": 11920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.971502032121543e-06,
+      "loss": 2.8679,
+      "step": 11921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.955095605762682e-06,
+      "loss": 2.9276,
+      "step": 11922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.938723002805289e-06,
+      "loss": 2.9337,
+      "step": 11923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.922384224365971e-06,
+      "loss": 3.0866,
+      "step": 11924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.906079271558671e-06,
+      "loss": 2.6878,
+      "step": 11925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.889808145495222e-06,
+      "loss": 2.831,
+      "step": 11926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8735708472850125e-06,
+      "loss": 2.8424,
+      "step": 11927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.857367378035326e-06,
+      "loss": 2.8377,
+      "step": 11928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8411977388508325e-06,
+      "loss": 3.1349,
+      "step": 11929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.825061930834261e-06,
+      "loss": 2.8734,
+      "step": 11930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.808959955085789e-06,
+      "loss": 3.1102,
+      "step": 11931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.792891812703314e-06,
+      "loss": 2.7411,
+      "step": 11932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7768575047824605e-06,
+      "loss": 2.7294,
+      "step": 11933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7608570324165757e-06,
+      "loss": 3.014,
+      "step": 11934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7448903966966764e-06,
+      "loss": 3.0344,
+      "step": 11935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7289575987114466e-06,
+      "loss": 2.8835,
+      "step": 11936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7130586395472956e-06,
+      "loss": 3.0029,
+      "step": 11937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6971935202883e-06,
+      "loss": 2.7621,
+      "step": 11938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6813622420162063e-06,
+      "loss": 2.9678,
+      "step": 11939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.665564805810595e-06,
+      "loss": 2.8851,
+      "step": 11940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.649801212748549e-06,
+      "loss": 2.7664,
+      "step": 11941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.634071463904931e-06,
+      "loss": 2.7741,
+      "step": 11942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.618375560352327e-06,
+      "loss": 3.0374,
+      "step": 11943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6027135031609924e-06,
+      "loss": 2.957,
+      "step": 11944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5870852933987397e-06,
+      "loss": 2.8567,
+      "step": 11945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.571490932131327e-06,
+      "loss": 3.1232,
+      "step": 11946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.555930420422071e-06,
+      "loss": 2.7607,
+      "step": 11947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5404037593319003e-06,
+      "loss": 2.9028,
+      "step": 11948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5249109499196352e-06,
+      "loss": 2.8979,
+      "step": 11949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5094519932415414e-06,
+      "loss": 2.9867,
+      "step": 11950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4940268903517756e-06,
+      "loss": 3.1045,
+      "step": 11951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4786356423021082e-06,
+      "loss": 2.7815,
+      "step": 11952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4632782501419767e-06,
+      "loss": 2.8886,
+      "step": 11953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.447954714918655e-06,
+      "loss": 2.5495,
+      "step": 11954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.432665037676863e-06,
+      "loss": 2.9611,
+      "step": 11955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.417409219459211e-06,
+      "loss": 2.8482,
+      "step": 11956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402187261305867e-06,
+      "loss": 2.7527,
+      "step": 11957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.386999164254889e-06,
+      "loss": 3.0697,
+      "step": 11958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.371844929341783e-06,
+      "loss": 2.9926,
+      "step": 11959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3567245575998882e-06,
+      "loss": 2.7761,
+      "step": 11960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3416380500602694e-06,
+      "loss": 3.011,
+      "step": 11961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.326585407751437e-06,
+      "loss": 2.726,
+      "step": 11962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.311566631700014e-06,
+      "loss": 2.9647,
+      "step": 11963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.296581722929903e-06,
+      "loss": 3.0125,
+      "step": 11964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.281630682462955e-06,
+      "loss": 3.0326,
+      "step": 11965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.26671351131852e-06,
+      "loss": 3.2058,
+      "step": 11966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.25183021051384e-06,
+      "loss": 2.8444,
+      "step": 11967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.23698078106377e-06,
+      "loss": 2.9074,
+      "step": 11968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.222165223980722e-06,
+      "loss": 3.1009,
+      "step": 11969
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2073835402749994e-06,
+      "loss": 2.8781,
+      "step": 11970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1926357309545184e-06,
+      "loss": 3.0011,
+      "step": 11971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1779217970248073e-06,
+      "loss": 2.8729,
+      "step": 11972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1632417394891755e-06,
+      "loss": 2.8895,
+      "step": 11973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1485955593486547e-06,
+      "loss": 2.9391,
+      "step": 11974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1339832576018357e-06,
+      "loss": 2.8738,
+      "step": 11975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.119404835245143e-06,
+      "loss": 2.9539,
+      "step": 11976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.10486029327256e-06,
+      "loss": 2.8604,
+      "step": 11977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0903496326758484e-06,
+      "loss": 2.991,
+      "step": 11978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.075872854444439e-06,
+      "loss": 2.6514,
+      "step": 11979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.061429959565487e-06,
+      "loss": 2.85,
+      "step": 11980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0470209490237045e-06,
+      "loss": 2.9709,
+      "step": 11981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.032645823801694e-06,
+      "loss": 3.0128,
+      "step": 11982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0183045848795608e-06,
+      "loss": 2.7805,
+      "step": 11983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.003997233235245e-06,
+      "loss": 2.6302,
+      "step": 11984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9897237698441883e-06,
+      "loss": 3.1445,
+      "step": 11985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9754841956797783e-06,
+      "loss": 3.0635,
+      "step": 11986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.961278511712906e-06,
+      "loss": 2.8238,
+      "step": 11987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9471067189122402e-06,
+      "loss": 3.0353,
+      "step": 11988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9329688182440085e-06,
+      "loss": 2.78,
+      "step": 11989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9188648106722724e-06,
+      "loss": 2.9692,
+      "step": 11990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9047946971587635e-06,
+      "loss": 2.761,
+      "step": 11991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8907584786627693e-06,
+      "loss": 2.7252,
+      "step": 11992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8767561561415247e-06,
+      "loss": 2.817,
+      "step": 11993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.862787730549654e-06,
+      "loss": 2.5554,
+      "step": 11994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.848853202839674e-06,
+      "loss": 2.8854,
+      "step": 11995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8349525739617133e-06,
+      "loss": 2.8749,
+      "step": 11996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8210858448636245e-06,
+      "loss": 2.9217,
+      "step": 11997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.807253016490874e-06,
+      "loss": 2.7057,
+      "step": 11998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.793454089786651e-06,
+      "loss": 2.9701,
+      "step": 11999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7796890656919817e-06,
+      "loss": 2.9817,
+      "step": 12000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7659579451453365e-06,
+      "loss": 2.9183,
+      "step": 12001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7522607290830224e-06,
+      "loss": 2.7968,
+      "step": 12002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.738597418439015e-06,
+      "loss": 2.7799,
+      "step": 12003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7249680141449574e-06,
+      "loss": 2.7119,
+      "step": 12004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.711372517130106e-06,
+      "loss": 2.9759,
+      "step": 12005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.697810928321609e-06,
+      "loss": 2.8622,
+      "step": 12006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.684283248644115e-06,
+      "loss": 2.9002,
+      "step": 12007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6707894790200526e-06,
+      "loss": 2.7625,
+      "step": 12008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.657329620369464e-06,
+      "loss": 2.8754,
+      "step": 12009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6439036736101707e-06,
+      "loss": 2.8524,
+      "step": 12010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6305116396576068e-06,
+      "loss": 2.8889,
+      "step": 12011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6171535194249308e-06,
+      "loss": 2.9476,
+      "step": 12012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.603829313823025e-06,
+      "loss": 3.0115,
+      "step": 12013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.59053902376033e-06,
+      "loss": 2.9577,
+      "step": 12014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5772826501431755e-06,
+      "loss": 2.9674,
+      "step": 12015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5640601938753396e-06,
+      "loss": 2.8448,
+      "step": 12016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.550871655858489e-06,
+      "loss": 2.9378,
+      "step": 12017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.537717036991849e-06,
+      "loss": 2.7117,
+      "step": 12018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5245963381724246e-06,
+      "loss": 2.8206,
+      "step": 12019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.511509560294889e-06,
+      "loss": 2.9056,
+      "step": 12020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.498456704251528e-06,
+      "loss": 2.8533,
+      "step": 12021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4854377709322974e-06,
+      "loss": 3.103,
+      "step": 12022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.472452761225097e-06,
+      "loss": 2.9193,
+      "step": 12023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4595016760151632e-06,
+      "loss": 2.9151,
+      "step": 12024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.446584516185679e-06,
+      "loss": 2.7353,
+      "step": 12025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.433701282617329e-06,
+      "loss": 3.0816,
+      "step": 12026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.420851976188687e-06,
+      "loss": 2.9286,
+      "step": 12027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4080365977757756e-06,
+      "loss": 3.0136,
+      "step": 12028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.395255148252451e-06,
+      "loss": 3.0813,
+      "step": 12029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3825076284902934e-06,
+      "loss": 2.9326,
+      "step": 12030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.369794039358497e-06,
+      "loss": 2.6549,
+      "step": 12031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3571143817239237e-06,
+      "loss": 2.9859,
+      "step": 12032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3444686564511043e-06,
+      "loss": 3.059,
+      "step": 12033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.331856864402404e-06,
+      "loss": 2.8358,
+      "step": 12034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.319279006437691e-06,
+      "loss": 2.9134,
+      "step": 12035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.306735083414668e-06,
+      "loss": 2.8104,
+      "step": 12036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2942250961885957e-06,
+      "loss": 2.9739,
+      "step": 12037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2817490456125133e-06,
+      "loss": 2.7256,
+      "step": 12038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2693069325371296e-06,
+      "loss": 2.8842,
+      "step": 12039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.256898757810766e-06,
+      "loss": 2.6893,
+      "step": 12040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2445245222795786e-06,
+      "loss": 3.1075,
+      "step": 12041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2321842267871707e-06,
+      "loss": 3.0042,
+      "step": 12042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.219877872175147e-06,
+      "loss": 3.1283,
+      "step": 12043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2076054592826132e-06,
+      "loss": 2.7143,
+      "step": 12044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1953669889462348e-06,
+      "loss": 2.8297,
+      "step": 12045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.183162462000676e-06,
+      "loss": 3.0955,
+      "step": 12046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1709918792779394e-06,
+      "loss": 2.8248,
+      "step": 12047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.158855241608082e-06,
+      "loss": 2.9226,
+      "step": 12048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.146752549818498e-06,
+      "loss": 2.9009,
+      "step": 12049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1346838047344718e-06,
+      "loss": 2.8379,
+      "step": 12050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1226490071790117e-06,
+      "loss": 2.7944,
+      "step": 12051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.110648157972628e-06,
+      "loss": 2.9431,
+      "step": 12052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.09868125793361e-06,
+      "loss": 2.9099,
+      "step": 12053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0867483078779724e-06,
+      "loss": 2.6995,
+      "step": 12054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0748493086193977e-06,
+      "loss": 3.1739,
+      "step": 12055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0629842609691807e-06,
+      "loss": 2.9169,
+      "step": 12056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.051153165736341e-06,
+      "loss": 2.9937,
+      "step": 12057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0393560237276785e-06,
+      "loss": 3.0915,
+      "step": 12058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0275928357474937e-06,
+      "loss": 3.0189,
+      "step": 12059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.015863602597978e-06,
+      "loss": 2.9078,
+      "step": 12060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0041683250788257e-06,
+      "loss": 2.9968,
+      "step": 12061
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9925070039875647e-06,
+      "loss": 2.6888,
+      "step": 12062
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9808796401192265e-06,
+      "loss": 2.8581,
+      "step": 12063
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9692862342667317e-06,
+      "loss": 2.7465,
+      "step": 12064
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9577267872205595e-06,
+      "loss": 3.0027,
+      "step": 12065
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.946201299768857e-06,
+      "loss": 2.9625,
+      "step": 12066
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.934709772697607e-06,
+      "loss": 3.0527,
+      "step": 12067
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9232522067902935e-06,
+      "loss": 2.9092,
+      "step": 12068
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.911828602828181e-06,
+      "loss": 2.7318,
+      "step": 12069
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9004389615902007e-06,
+      "loss": 3.0245,
+      "step": 12070
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8890832838530103e-06,
+      "loss": 2.9632,
+      "step": 12071
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8777615703908235e-06,
+      "loss": 2.8048,
+      "step": 12072
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8664738219756895e-06,
+      "loss": 3.0399,
+      "step": 12073
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.85522003937727e-06,
+      "loss": 2.9323,
+      "step": 12074
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8440002233628961e-06,
+      "loss": 2.8533,
+      "step": 12075
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8328143746976222e-06,
+      "loss": 2.7848,
+      "step": 12076
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8216624941441717e-06,
+      "loss": 2.8127,
+      "step": 12077
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8105445824629363e-06,
+      "loss": 2.9686,
+      "step": 12078
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7994606404120318e-06,
+      "loss": 2.5308,
+      "step": 12079
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7884106687471314e-06,
+      "loss": 2.6523,
+      "step": 12080
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.777394668221799e-06,
+      "loss": 2.9126,
+      "step": 12081
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.766412639587156e-06,
+      "loss": 2.8128,
+      "step": 12082
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7554645835919925e-06,
+      "loss": 2.8114,
+      "step": 12083
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7445505009828223e-06,
+      "loss": 2.8554,
+      "step": 12084
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7336703925038831e-06,
+      "loss": 3.0272,
+      "step": 12085
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7228242588969712e-06,
+      "loss": 2.724,
+      "step": 12086
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.712012100901661e-06,
+      "loss": 2.7476,
+      "step": 12087
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7012339192552517e-06,
+      "loss": 2.878,
+      "step": 12088
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6904897146926001e-06,
+      "loss": 2.965,
+      "step": 12089
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6797794879463425e-06,
+      "loss": 3.0203,
+      "step": 12090
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.669103239746783e-06,
+      "loss": 3.0887,
+      "step": 12091
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.65846097082184e-06,
+      "loss": 2.998,
+      "step": 12092
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6478526818972107e-06,
+      "loss": 2.531,
+      "step": 12093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.637278373696205e-06,
+      "loss": 2.8911,
+      "step": 12094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6267380469399128e-06,
+      "loss": 2.9756,
+      "step": 12095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6162317023469264e-06,
+      "loss": 2.8329,
+      "step": 12096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6057593406337279e-06,
+      "loss": 2.9765,
+      "step": 12097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5953209625143572e-06,
+      "loss": 3.1664,
+      "step": 12098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5849165687005784e-06,
+      "loss": 2.7207,
+      "step": 12099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574546159901824e-06,
+      "loss": 2.8814,
+      "step": 12100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5642097368251951e-06,
+      "loss": 3.0523,
+      "step": 12101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.553907300175461e-06,
+      "loss": 2.8808,
+      "step": 12102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5436388506551713e-06,
+      "loss": 3.0914,
+      "step": 12103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5334043889644877e-06,
+      "loss": 3.0013,
+      "step": 12104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5232039158012413e-06,
+      "loss": 2.8402,
+      "step": 12105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5130374318609307e-06,
+      "loss": 2.8572,
+      "step": 12106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5029049378368354e-06,
+      "loss": 2.7435,
+      "step": 12107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4928064344197912e-06,
+      "loss": 2.8332,
+      "step": 12108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4827419222983585e-06,
+      "loss": 2.6111,
+      "step": 12109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4727114021588773e-06,
+      "loss": 3.1133,
+      "step": 12110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4627148746853002e-06,
+      "loss": 2.8731,
+      "step": 12111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.452752340559138e-06,
+      "loss": 2.6989,
+      "step": 12112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4428238004597916e-06,
+      "loss": 2.7999,
+      "step": 12113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4329292550642193e-06,
+      "loss": 2.9656,
+      "step": 12114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4230687050471036e-06,
+      "loss": 3.0377,
+      "step": 12115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4132421510807402e-06,
+      "loss": 2.8906,
+      "step": 12116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4034495938352598e-06,
+      "loss": 2.7198,
+      "step": 12117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3936910339783504e-06,
+      "loss": 2.7276,
+      "step": 12118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.383966472175313e-06,
+      "loss": 2.8922,
+      "step": 12119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3742759090893953e-06,
+      "loss": 2.8516,
+      "step": 12120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.364619345381235e-06,
+      "loss": 2.7953,
+      "step": 12121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.354996781709361e-06,
+      "loss": 2.9689,
+      "step": 12122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.345408218729749e-06,
+      "loss": 2.9284,
+      "step": 12123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3358536570964308e-06,
+      "loss": 2.9414,
+      "step": 12124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3263330974606635e-06,
+      "loss": 2.7443,
+      "step": 12125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3168465404718167e-06,
+      "loss": 3.0858,
+      "step": 12126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.307393986776595e-06,
+      "loss": 3.0394,
+      "step": 12127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2979754370196495e-06,
+      "loss": 2.993,
+      "step": 12128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2885908918430779e-06,
+      "loss": 2.9384,
+      "step": 12129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2792403518868678e-06,
+      "loss": 2.8702,
+      "step": 12130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2699238177885652e-06,
+      "loss": 2.8101,
+      "step": 12131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2606412901833842e-06,
+      "loss": 2.7667,
+      "step": 12132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2513927697043182e-06,
+      "loss": 2.9735,
+      "step": 12133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2421782569820295e-06,
+      "loss": 2.9964,
+      "step": 12134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2329977526447378e-06,
+      "loss": 2.9427,
+      "step": 12135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2238512573184424e-06,
+      "loss": 3.1105,
+      "step": 12136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.214738771626811e-06,
+      "loss": 3.0388,
+      "step": 12137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.205660296191291e-06,
+      "loss": 2.8446,
+      "step": 12138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1966158316307207e-06,
+      "loss": 3.0906,
+      "step": 12139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.18760537856194e-06,
+      "loss": 2.9764,
+      "step": 12140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.178628937599291e-06,
+      "loss": 3.1013,
+      "step": 12141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1696865093548948e-06,
+      "loss": 2.6664,
+      "step": 12142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1607780944384305e-06,
+      "loss": 2.8281,
+      "step": 12143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1519036934573568e-06,
+      "loss": 2.8268,
+      "step": 12144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1430633070168006e-06,
+      "loss": 2.6764,
+      "step": 12145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1342569357195575e-06,
+      "loss": 2.8978,
+      "step": 12146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1254845801660363e-06,
+      "loss": 2.8013,
+      "step": 12147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1167462409544804e-06,
+      "loss": 3.0024,
+      "step": 12148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1080419186806357e-06,
+      "loss": 3.0488,
+      "step": 12149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0993716139381382e-06,
+      "loss": 2.9308,
+      "step": 12150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0907353273180155e-06,
+      "loss": 2.9514,
+      "step": 12151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.082133059409296e-06,
+      "loss": 3.0285,
+      "step": 12152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0735648107984552e-06,
+      "loss": 2.9146,
+      "step": 12153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.065030582069748e-06,
+      "loss": 2.5896,
+      "step": 12154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.056530373805098e-06,
+      "loss": 2.9678,
+      "step": 12155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0480641865840968e-06,
+      "loss": 3.1329,
+      "step": 12156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0396320209840048e-06,
+      "loss": 2.5513,
+      "step": 12157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.031233877579807e-06,
+      "loss": 3.0815,
+      "step": 12158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0228697569441003e-06,
+      "loss": 2.9592,
+      "step": 12159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0145396596472623e-06,
+      "loss": 3.0696,
+      "step": 12160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0062435862572273e-06,
+      "loss": 3.0207,
+      "step": 12161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.979815373397094e-07,
+      "loss": 3.0842,
+      "step": 12162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.897535134580359e-07,
+      "loss": 3.0745,
+      "step": 12163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.815595151733136e-07,
+      "loss": 2.989,
+      "step": 12164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.733995430442067e-07,
+      "loss": 2.9227,
+      "step": 12165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.65273597627103e-07,
+      "loss": 2.9397,
+      "step": 12166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.571816794761156e-07,
+      "loss": 2.9986,
+      "step": 12167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.491237891429694e-07,
+      "loss": 2.9853,
+      "step": 12168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.410999271770582e-07,
+      "loss": 2.8872,
+      "step": 12169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.33110094125611e-07,
+      "loss": 2.456,
+      "step": 12170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.251542905333588e-07,
+      "loss": 3.1279,
+      "step": 12171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.172325169427564e-07,
+      "loss": 2.9958,
+      "step": 12172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.093447738940385e-07,
+      "loss": 2.898,
+      "step": 12173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.014910619249417e-07,
+      "loss": 2.8575,
+      "step": 12174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.936713815710929e-07,
+      "loss": 2.8992,
+      "step": 12175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.858857333655656e-07,
+      "loss": 2.9659,
+      "step": 12176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.781341178393243e-07,
+      "loss": 2.7799,
+      "step": 12177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.704165355208349e-07,
+      "loss": 2.6868,
+      "step": 12178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.627329869363987e-07,
+      "loss": 2.7653,
+      "step": 12179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.5508347260993e-07,
+      "loss": 2.8029,
+      "step": 12180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.474679930629558e-07,
+      "loss": 2.9678,
+      "step": 12181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.398865488148388e-07,
+      "loss": 2.9791,
+      "step": 12182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.323391403824986e-07,
+      "loss": 2.706,
+      "step": 12183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.248257682805238e-07,
+      "loss": 3.0463,
+      "step": 12184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.173464330212266e-07,
+      "loss": 2.5717,
+      "step": 12185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.09901135114588e-07,
+      "loss": 2.8783,
+      "step": 12186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.024898750683684e-07,
+      "loss": 3.0653,
+      "step": 12187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.95112653387775e-07,
+      "loss": 2.876,
+      "step": 12188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.877694705759053e-07,
+      "loss": 2.5912,
+      "step": 12189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.804603271335253e-07,
+      "loss": 2.8255,
+      "step": 12190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.731852235589032e-07,
+      "loss": 2.6678,
+      "step": 12191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.65944160348142e-07,
+      "loss": 2.8695,
+      "step": 12192
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.587371379950137e-07,
+      "loss": 2.7773,
+      "step": 12193
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.515641569909027e-07,
+      "loss": 3.0989,
+      "step": 12194
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.444252178248623e-07,
+      "loss": 2.768,
+      "step": 12195
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.373203209837808e-07,
+      "loss": 3.032,
+      "step": 12196
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.302494669520487e-07,
+      "loss": 2.7141,
+      "step": 12197
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.232126562118357e-07,
+      "loss": 2.9098,
+      "step": 12198
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.162098892428692e-07,
+      "loss": 2.9354,
+      "step": 12199
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.092411665227116e-07,
+      "loss": 2.8708,
+      "step": 12200
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.023064885265384e-07,
+      "loss": 2.5857,
+      "step": 12201
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.954058557271936e-07,
+      "loss": 2.725,
+      "step": 12202
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.885392685951341e-07,
+      "loss": 2.9467,
+      "step": 12203
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.817067275987077e-07,
+      "loss": 2.8963,
+      "step": 12204
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.749082332036527e-07,
+      "loss": 3.0953,
+      "step": 12205
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.681437858735984e-07,
+      "loss": 2.8881,
+      "step": 12206
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.614133860698424e-07,
+      "loss": 2.9424,
+      "step": 12207
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.547170342511844e-07,
+      "loss": 3.0445,
+      "step": 12208
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.480547308743146e-07,
+      "loss": 2.8138,
+      "step": 12209
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.414264763934807e-07,
+      "loss": 2.997,
+      "step": 12210
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.348322712605992e-07,
+      "loss": 2.8045,
+      "step": 12211
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.282721159253658e-07,
+      "loss": 2.7124,
+      "step": 12212
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.217460108350892e-07,
+      "loss": 2.7592,
+      "step": 12213
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.152539564346915e-07,
+      "loss": 2.9756,
+      "step": 12214
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.087959531669296e-07,
+      "loss": 2.9673,
+      "step": 12215
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.023720014721179e-07,
+      "loss": 3.0176,
+      "step": 12216
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.959821017882394e-07,
+      "loss": 2.6984,
+      "step": 12217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.896262545510566e-07,
+      "loss": 2.942,
+      "step": 12218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.833044601938897e-07,
+      "loss": 2.7026,
+      "step": 12219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.770167191478382e-07,
+      "loss": 2.8473,
+      "step": 12220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.707630318416701e-07,
+      "loss": 2.7119,
+      "step": 12221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.645433987017113e-07,
+      "loss": 2.8794,
+      "step": 12222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.583578201521223e-07,
+      "loss": 3.0241,
+      "step": 12223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.522062966146768e-07,
+      "loss": 2.9213,
+      "step": 12224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.460888285087617e-07,
+      "loss": 3.0606,
+      "step": 12225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.400054162515988e-07,
+      "loss": 3.0534,
+      "step": 12226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.339560602579118e-07,
+      "loss": 2.9374,
+      "step": 12227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.279407609402042e-07,
+      "loss": 2.8075,
+      "step": 12228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.219595187086479e-07,
+      "loss": 2.9186,
+      "step": 12229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.160123339710831e-07,
+      "loss": 2.74,
+      "step": 12230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.100992071330191e-07,
+      "loss": 3.0252,
+      "step": 12231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.042201385976331e-07,
+      "loss": 2.9577,
+      "step": 12232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.983751287658822e-07,
+      "loss": 2.7672,
+      "step": 12233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9256417803617e-07,
+      "loss": 3.0758,
+      "step": 12234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.867872868048461e-07,
+      "loss": 2.8845,
+      "step": 12235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.810444554657067e-07,
+      "loss": 2.9073,
+      "step": 12236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7533568441043837e-07,
+      "loss": 2.7717,
+      "step": 12237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6966097402822983e-07,
+      "loss": 2.9346,
+      "step": 12238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6402032470604924e-07,
+      "loss": 2.8219,
+      "step": 12239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.5841373682853347e-07,
+      "loss": 3.0036,
+      "step": 12240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.528412107779323e-07,
+      "loss": 2.8957,
+      "step": 12241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.473027469342195e-07,
+      "loss": 2.7248,
+      "step": 12242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.417983456750374e-07,
+      "loss": 2.7938,
+      "step": 12243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.363280073757525e-07,
+      "loss": 2.813,
+      "step": 12244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.308917324092887e-07,
+      "loss": 2.8608,
+      "step": 12245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.254895211464049e-07,
+      "loss": 2.8369,
+      "step": 12246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2012137395541773e-07,
+      "loss": 2.7573,
+      "step": 12247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.147872912024231e-07,
+      "loss": 3.0588,
+      "step": 12248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0948727325101906e-07,
+      "loss": 2.4919,
+      "step": 12249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0422132046263883e-07,
+      "loss": 2.6249,
+      "step": 12250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.989894331963839e-07,
+      "loss": 2.8135,
+      "step": 12251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9379161180896904e-07,
+      "loss": 2.9519,
+      "step": 12252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.886278566548329e-07,
+      "loss": 2.995,
+      "step": 12253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8349816808602724e-07,
+      "loss": 2.6808,
+      "step": 12254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.784025464523833e-07,
+      "loss": 2.919,
+      "step": 12255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.733409921013453e-07,
+      "loss": 2.7951,
+      "step": 12256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.683135053779707e-07,
+      "loss": 2.6127,
+      "step": 12257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.633200866251518e-07,
+      "loss": 2.8132,
+      "step": 12258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.58360736183283e-07,
+      "loss": 2.6547,
+      "step": 12259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.534354543905938e-07,
+      "loss": 2.7588,
+      "step": 12260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4854424158292654e-07,
+      "loss": 3.0128,
+      "step": 12261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.436870980937368e-07,
+      "loss": 3.0389,
+      "step": 12262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3886402425425953e-07,
+      "loss": 2.8959,
+      "step": 12263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.340750203933429e-07,
+      "loss": 3.0752,
+      "step": 12264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2932008683755897e-07,
+      "loss": 2.6696,
+      "step": 12265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2459922391109285e-07,
+      "loss": 2.8025,
+      "step": 12266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1991243193579823e-07,
+      "loss": 3.1416,
+      "step": 12267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1525971123136376e-07,
+      "loss": 2.8079,
+      "step": 12268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1064106211498024e-07,
+      "loss": 2.9344,
+      "step": 12269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0605648490156237e-07,
+      "loss": 3.1269,
+      "step": 12270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0150597990374893e-07,
+      "loss": 2.8331,
+      "step": 12271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9698954743179184e-07,
+      "loss": 2.9271,
+      "step": 12272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.925071877936114e-07,
+      "loss": 2.9457,
+      "step": 12273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8805890129496306e-07,
+      "loss": 2.9896,
+      "step": 12274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8364468823904866e-07,
+      "loss": 3.002,
+      "step": 12275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7926454892690525e-07,
+      "loss": 2.9375,
+      "step": 12276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7491848365718277e-07,
+      "loss": 3.1074,
+      "step": 12277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7060649272625525e-07,
+      "loss": 2.788,
+      "step": 12278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.663285764280543e-07,
+      "loss": 2.9649,
+      "step": 12279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.620847350543465e-07,
+      "loss": 2.925,
+      "step": 12280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5787496889451146e-07,
+      "loss": 2.9486,
+      "step": 12281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.536992782355974e-07,
+      "loss": 3.0448,
+      "step": 12282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4955766336226536e-07,
+      "loss": 2.9239,
+      "step": 12283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4545012455695623e-07,
+      "loss": 2.8719,
+      "step": 12284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.413766620997793e-07,
+      "loss": 3.0818,
+      "step": 12285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.373372762684567e-07,
+      "loss": 2.8029,
+      "step": 12286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3333196733837937e-07,
+      "loss": 2.7613,
+      "step": 12287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2936073558271764e-07,
+      "loss": 2.705,
+      "step": 12288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2542358127225493e-07,
+      "loss": 2.9385,
+      "step": 12289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2152050467538764e-07,
+      "loss": 2.7506,
+      "step": 12290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1765150605829175e-07,
+      "loss": 2.9518,
+      "step": 12291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1381658568481178e-07,
+      "loss": 2.9373,
+      "step": 12292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.100157438163497e-07,
+      "loss": 2.8409,
+      "step": 12293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.062489807121426e-07,
+      "loss": 2.8553,
+      "step": 12294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0251629662904057e-07,
+      "loss": 2.6943,
+      "step": 12295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9881769182150678e-07,
+      "loss": 2.8701,
+      "step": 12296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9515316654178384e-07,
+      "loss": 2.8927,
+      "step": 12297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9152272103972747e-07,
+      "loss": 2.9114,
+      "step": 12298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8792635556280636e-07,
+      "loss": 2.776,
+      "step": 12299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.843640703563798e-07,
+      "loss": 2.7638,
+      "step": 12300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8083586566325362e-07,
+      "loss": 2.9844,
+      "step": 12301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7734174172401307e-07,
+      "loss": 2.9107,
+      "step": 12302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7388169877696758e-07,
+      "loss": 3.1043,
+      "step": 12303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.70455737057984e-07,
+      "loss": 2.8139,
+      "step": 12304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.670638568006533e-07,
+      "loss": 2.9539,
+      "step": 12305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6370605823629037e-07,
+      "loss": 3.0247,
+      "step": 12306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6038234159387876e-07,
+      "loss": 3.0052,
+      "step": 12307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5709270709995948e-07,
+      "loss": 2.8145,
+      "step": 12308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.538371549789086e-07,
+      "loss": 3.0655,
+      "step": 12309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.506156854527152e-07,
+      "loss": 2.7292,
+      "step": 12310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4742829874103692e-07,
+      "loss": 3.0279,
+      "step": 12311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.442749950611444e-07,
+      "loss": 2.8701,
+      "step": 12312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4115577462814333e-07,
+      "loss": 2.7154,
+      "step": 12313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3807063765464146e-07,
+      "loss": 2.8914,
+      "step": 12314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3501958435108153e-07,
+      "loss": 3.0301,
+      "step": 12315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3200261492540832e-07,
+      "loss": 2.7227,
+      "step": 12316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.290197295834017e-07,
+      "loss": 3.011,
+      "step": 12317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2607092852845448e-07,
+      "loss": 2.7321,
+      "step": 12318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2315621196157257e-07,
+      "loss": 2.7951,
+      "step": 12319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.202755800815969e-07,
+      "loss": 2.8761,
+      "step": 12320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1742903308487041e-07,
+      "loss": 2.9922,
+      "step": 12321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.146165711655156e-07,
+      "loss": 2.9248,
+      "step": 12322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.118381945153235e-07,
+      "loss": 3.089,
+      "step": 12323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0909390332369817e-07,
+      "loss": 2.7001,
+      "step": 12324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0638369777776768e-07,
+      "loss": 3.0324,
+      "step": 12325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0370757806232866e-07,
+      "loss": 2.7331,
+      "step": 12326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0106554435990178e-07,
+      "loss": 2.9128,
+      "step": 12327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.845759685062072e-08,
+      "loss": 2.7593,
+      "step": 12328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.58837357122877e-08,
+      "loss": 2.9314,
+      "step": 12329
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.334396112042898e-08,
+      "loss": 2.8256,
+      "step": 12330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.083827324818384e-08,
+      "loss": 2.994,
+      "step": 12331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.836667226641559e-08,
+      "loss": 2.9643,
+      "step": 12332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.592915834371162e-08,
+      "loss": 2.9931,
+      "step": 12333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.352573164616129e-08,
+      "loss": 3.1047,
+      "step": 12334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.115639233780003e-08,
+      "loss": 2.8997,
+      "step": 12335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.88211405800543e-08,
+      "loss": 2.9087,
+      "step": 12336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.651997653218556e-08,
+      "loss": 2.8379,
+      "step": 12337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.425290035117937e-08,
+      "loss": 3.0647,
+      "step": 12338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.201991219146775e-08,
+      "loss": 2.7103,
+      "step": 12339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.982101220548431e-08,
+      "loss": 2.9957,
+      "step": 12340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.765620054305366e-08,
+      "loss": 2.7932,
+      "step": 12341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.552547735177994e-08,
+      "loss": 2.8944,
+      "step": 12342
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.342884277699134e-08,
+      "loss": 2.9629,
+      "step": 12343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.13662969616291e-08,
+      "loss": 3.0657,
+      "step": 12344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.933784004635845e-08,
+      "loss": 2.7839,
+      "step": 12345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.734347216945768e-08,
+      "loss": 2.7979,
+      "step": 12346
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.5383193466984615e-08,
+      "loss": 2.7867,
+      "step": 12347
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.345700407249909e-08,
+      "loss": 3.0562,
+      "step": 12348
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.156490411739601e-08,
+      "loss": 2.7762,
+      "step": 12349
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.970689373068327e-08,
+      "loss": 2.9249,
+      "step": 12350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.788297303903733e-08,
+      "loss": 2.864,
+      "step": 12351
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.609314216685867e-08,
+      "loss": 3.0789,
+      "step": 12352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4337401236160813e-08,
+      "loss": 3.0119,
+      "step": 12353
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.2615750366681306e-08,
+      "loss": 2.836,
+      "step": 12354
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.092818967577072e-08,
+      "loss": 2.8303,
+      "step": 12355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.927471927855919e-08,
+      "loss": 2.7927,
+      "step": 12356
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.765533928773435e-08,
+      "loss": 2.6868,
+      "step": 12357
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.607004981376338e-08,
+      "loss": 2.7865,
+      "step": 12358
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.451885096472651e-08,
+      "loss": 2.7701,
+      "step": 12359
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.300174284637247e-08,
+      "loss": 2.9935,
+      "step": 12360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1518725562174054e-08,
+      "loss": 2.7806,
+      "step": 12361
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.006979921321706e-08,
+      "loss": 2.6287,
+      "step": 12362
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8654963898311348e-08,
+      "loss": 2.933,
+      "step": 12363
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.72742197139908e-08,
+      "loss": 2.7408,
+      "step": 12364
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.592756675429131e-08,
+      "loss": 2.8528,
+      "step": 12365
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4615005111139343e-08,
+      "loss": 2.9892,
+      "step": 12366
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3336534873963367e-08,
+      "loss": 2.6971,
+      "step": 12367
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2092156129971396e-08,
+      "loss": 2.7572,
+      "step": 12368
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0881868963984473e-08,
+      "loss": 2.9116,
+      "step": 12369
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9705673458603192e-08,
+      "loss": 3.0289,
+      "step": 12370
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.856356969393014e-08,
+      "loss": 2.9818,
+      "step": 12371
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7455557747847462e-08,
+      "loss": 2.8122,
+      "step": 12372
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.638163769601686e-08,
+      "loss": 2.9415,
+      "step": 12373
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.534180961154652e-08,
+      "loss": 2.9852,
+      "step": 12374
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4336073565379693e-08,
+      "loss": 2.7972,
+      "step": 12375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.336442962612816e-08,
+      "loss": 2.9383,
+      "step": 12376
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2426877859961216e-08,
+      "loss": 2.8103,
+      "step": 12377
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1523418330883218e-08,
+      "loss": 2.9792,
+      "step": 12378
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.065405110045603e-08,
+      "loss": 2.9805,
+      "step": 12379
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.818776228021076e-09,
+      "loss": 2.7114,
+      "step": 12380
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.017593770421773e-09,
+      "loss": 2.8677,
+      "step": 12381
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.250503782392115e-09,
+      "loss": 2.8568,
+      "step": 12382
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.51750631622361e-09,
+      "loss": 3.0028,
+      "step": 12383
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.8186014218207806e-09,
+      "loss": 2.9006,
+      "step": 12384
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.153789146923216e-09,
+      "loss": 2.6429,
+      "step": 12385
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.523069536772507e-09,
+      "loss": 2.6909,
+      "step": 12386
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.926442634500816e-09,
+      "loss": 2.9319,
+      "step": 12387
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.363908480686796e-09,
+      "loss": 2.7437,
+      "step": 12388
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8354671136886506e-09,
+      "loss": 2.8612,
+      "step": 12389
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3411185695886304e-09,
+      "loss": 2.8204,
+      "step": 12390
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8808628820820028e-09,
+      "loss": 2.9525,
+      "step": 12391
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4547000825325683e-09,
+      "loss": 2.6894,
+      "step": 12392
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0626302000281705e-09,
+      "loss": 3.0581,
+      "step": 12393
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7046532613251842e-09,
+      "loss": 2.864,
+      "step": 12394
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3807692907374937e-09,
+      "loss": 2.8374,
+      "step": 12395
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.090978310414048e-09,
+      "loss": 2.8573,
+      "step": 12396
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.352803401168174e-10,
+      "loss": 2.6636,
+      "step": 12397
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.13675397220792e-10,
+      "loss": 2.8159,
+      "step": 12398
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.261634969360273e-10,
+      "loss": 2.8024,
+      "step": 12399
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7274465197457687e-10,
+      "loss": 3.0661,
+      "step": 12400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5341887282804834e-10,
+      "loss": 2.6774,
+      "step": 12401
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.818616765658092e-11,
+      "loss": 2.4839,
+      "step": 12402
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.70465421778232e-11,
+      "loss": 2.9068,
+      "step": 12403
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 3.0942,
+      "step": 12404
+    },
+    {
+      "epoch": 1.0,
+      "step": 12404,
+      "total_flos": 5.258763499806392e+17,
+      "train_loss": 3.041371289221405,
+      "train_runtime": 11617.6217,
+      "train_samples_per_second": 51.248,
+      "train_steps_per_second": 1.068
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 12404,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 5.258763499806392e+17,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}