diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,36108 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.4383954154727796,
+  "global_step": 6000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9999312320916904e-05,
+      "loss": 1.4453,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999862464183381e-05,
+      "loss": 1.4104,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9997936962750715e-05,
+      "loss": 1.3941,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9997249283667624e-05,
+      "loss": 1.3744,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9996561604584526e-05,
+      "loss": 1.3616,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9995873925501435e-05,
+      "loss": 1.3432,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.999518624641834e-05,
+      "loss": 1.3249,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9994498567335246e-05,
+      "loss": 1.3046,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999381088825215e-05,
+      "loss": 1.2915,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9993123209169057e-05,
+      "loss": 1.2643,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999243553008596e-05,
+      "loss": 1.2743,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999174785100287e-05,
+      "loss": 1.2497,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999106017191977e-05,
+      "loss": 1.2248,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.999037249283668e-05,
+      "loss": 1.218,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.998968481375358e-05,
+      "loss": 1.1984,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.998899713467049e-05,
+      "loss": 1.1824,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.998830945558739e-05,
+      "loss": 1.1614,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99876217765043e-05,
+      "loss": 1.165,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.99869340974212e-05,
+      "loss": 1.1457,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.998624641833811e-05,
+      "loss": 1.1384,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9985558739255014e-05,
+      "loss": 1.1081,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.998487106017192e-05,
+      "loss": 1.0945,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9984183381088825e-05,
+      "loss": 1.0709,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9983495702005734e-05,
+      "loss": 1.051,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9982808022922636e-05,
+      "loss": 1.059,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9982120343839545e-05,
+      "loss": 1.02,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9981432664756447e-05,
+      "loss": 1.0222,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9980744985673355e-05,
+      "loss": 0.9961,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.998005730659026e-05,
+      "loss": 1.0428,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9979369627507166e-05,
+      "loss": 1.0169,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.997868194842407e-05,
+      "loss": 0.9827,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.997799426934098e-05,
+      "loss": 0.9293,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.997730659025788e-05,
+      "loss": 0.9287,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.997661891117479e-05,
+      "loss": 0.9355,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.997593123209169e-05,
+      "loss": 0.9111,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99752435530086e-05,
+      "loss": 0.9426,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.99745558739255e-05,
+      "loss": 0.9279,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.997386819484241e-05,
+      "loss": 0.8917,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.997318051575931e-05,
+      "loss": 0.8569,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.997249283667622e-05,
+      "loss": 0.8529,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9971805157593124e-05,
+      "loss": 0.8333,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.997111747851003e-05,
+      "loss": 0.8653,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9970429799426935e-05,
+      "loss": 0.8614,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9969742120343843e-05,
+      "loss": 0.8854,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9969054441260746e-05,
+      "loss": 0.7929,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9968366762177654e-05,
+      "loss": 0.8201,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9967679083094557e-05,
+      "loss": 0.8126,
+      "step": 47
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9966991404011465e-05,
+      "loss": 0.762,
+      "step": 48
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.996630372492837e-05,
+      "loss": 0.7702,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9965616045845276e-05,
+      "loss": 0.763,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.996492836676218e-05,
+      "loss": 0.7434,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.996424068767909e-05,
+      "loss": 0.7309,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.996355300859599e-05,
+      "loss": 0.7234,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99628653295129e-05,
+      "loss": 0.6848,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.99621776504298e-05,
+      "loss": 0.6685,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.996148997134671e-05,
+      "loss": 0.7326,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.996080229226361e-05,
+      "loss": 0.751,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.996011461318052e-05,
+      "loss": 0.6549,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.995942693409742e-05,
+      "loss": 0.6261,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.995873925501433e-05,
+      "loss": 0.6858,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.9958051575931234e-05,
+      "loss": 0.6387,
+      "step": 61
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995736389684814e-05,
+      "loss": 0.6242,
+      "step": 62
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9956676217765045e-05,
+      "loss": 0.5805,
+      "step": 63
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9955988538681953e-05,
+      "loss": 0.6216,
+      "step": 64
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9955300859598856e-05,
+      "loss": 0.5796,
+      "step": 65
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9954613180515764e-05,
+      "loss": 0.574,
+      "step": 66
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9953925501432667e-05,
+      "loss": 0.5548,
+      "step": 67
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9953237822349575e-05,
+      "loss": 0.582,
+      "step": 68
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995255014326648e-05,
+      "loss": 0.5712,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.9951862464183386e-05,
+      "loss": 0.5721,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.995117478510029e-05,
+      "loss": 0.5542,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99504871060172e-05,
+      "loss": 0.5815,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.99497994269341e-05,
+      "loss": 0.5403,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.994911174785101e-05,
+      "loss": 0.5289,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.994842406876791e-05,
+      "loss": 0.4968,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.994773638968482e-05,
+      "loss": 0.554,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.994704871060172e-05,
+      "loss": 0.5653,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.994636103151863e-05,
+      "loss": 0.548,
+      "step": 78
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.994567335243553e-05,
+      "loss": 0.4493,
+      "step": 79
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.994498567335244e-05,
+      "loss": 0.4536,
+      "step": 80
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9944297994269344e-05,
+      "loss": 0.5695,
+      "step": 81
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.994361031518625e-05,
+      "loss": 0.5091,
+      "step": 82
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9942922636103155e-05,
+      "loss": 0.5044,
+      "step": 83
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9942234957020063e-05,
+      "loss": 0.4425,
+      "step": 84
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9941547277936966e-05,
+      "loss": 0.4717,
+      "step": 85
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9940859598853874e-05,
+      "loss": 0.4729,
+      "step": 86
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9940171919770777e-05,
+      "loss": 0.4932,
+      "step": 87
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9939484240687685e-05,
+      "loss": 0.51,
+      "step": 88
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993879656160459e-05,
+      "loss": 0.4124,
+      "step": 89
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9938108882521496e-05,
+      "loss": 0.4801,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.99374212034384e-05,
+      "loss": 0.4645,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993673352435531e-05,
+      "loss": 0.3832,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993604584527221e-05,
+      "loss": 0.3685,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993535816618912e-05,
+      "loss": 0.389,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.993467048710602e-05,
+      "loss": 0.5481,
+      "step": 95
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.993398280802293e-05,
+      "loss": 0.5097,
+      "step": 96
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.993329512893983e-05,
+      "loss": 0.4567,
+      "step": 97
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.993260744985674e-05,
+      "loss": 0.4647,
+      "step": 98
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.993191977077364e-05,
+      "loss": 0.3918,
+      "step": 99
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.993123209169055e-05,
+      "loss": 0.4282,
+      "step": 100
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9930544412607454e-05,
+      "loss": 0.3838,
+      "step": 101
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.992985673352436e-05,
+      "loss": 0.3937,
+      "step": 102
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9929169054441265e-05,
+      "loss": 0.3886,
+      "step": 103
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9928481375358173e-05,
+      "loss": 0.4047,
+      "step": 104
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9927793696275076e-05,
+      "loss": 0.368,
+      "step": 105
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9927106017191984e-05,
+      "loss": 0.5107,
+      "step": 106
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9926418338108887e-05,
+      "loss": 0.3476,
+      "step": 107
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9925730659025795e-05,
+      "loss": 0.3618,
+      "step": 108
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.99250429799427e-05,
+      "loss": 0.3488,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.9924355300859606e-05,
+      "loss": 0.3496,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.992366762177651e-05,
+      "loss": 0.3626,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.992297994269342e-05,
+      "loss": 0.3933,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 5.992229226361032e-05,
+      "loss": 0.336,
+      "step": 113
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.992160458452723e-05,
+      "loss": 0.3209,
+      "step": 114
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.992091690544413e-05,
+      "loss": 0.2745,
+      "step": 115
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.992022922636104e-05,
+      "loss": 0.3366,
+      "step": 116
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.991954154727794e-05,
+      "loss": 0.3488,
+      "step": 117
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.991885386819485e-05,
+      "loss": 0.2988,
+      "step": 118
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.991816618911175e-05,
+      "loss": 0.2945,
+      "step": 119
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9917478510028655e-05,
+      "loss": 0.2951,
+      "step": 120
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9916790830945564e-05,
+      "loss": 0.3418,
+      "step": 121
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9916103151862466e-05,
+      "loss": 0.3414,
+      "step": 122
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9915415472779375e-05,
+      "loss": 0.4199,
+      "step": 123
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.991472779369628e-05,
+      "loss": 0.3871,
+      "step": 124
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9914040114613186e-05,
+      "loss": 0.3067,
+      "step": 125
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.991335243553009e-05,
+      "loss": 0.3412,
+      "step": 126
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.9912664756446997e-05,
+      "loss": 0.3385,
+      "step": 127
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99119770773639e-05,
+      "loss": 0.3048,
+      "step": 128
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99112893982808e-05,
+      "loss": 0.2971,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.99106017191977e-05,
+      "loss": 0.2899,
+      "step": 130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.990991404011461e-05,
+      "loss": 0.2689,
+      "step": 131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9909226361031514e-05,
+      "loss": 0.2765,
+      "step": 132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.990853868194842e-05,
+      "loss": 0.2742,
+      "step": 133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9907851002865325e-05,
+      "loss": 0.2415,
+      "step": 134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9907163323782234e-05,
+      "loss": 0.2356,
+      "step": 135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9906475644699136e-05,
+      "loss": 0.3307,
+      "step": 136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9905787965616045e-05,
+      "loss": 0.2273,
+      "step": 137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.990510028653295e-05,
+      "loss": 0.3424,
+      "step": 138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9904412607449856e-05,
+      "loss": 0.3039,
+      "step": 139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.990372492836676e-05,
+      "loss": 0.2834,
+      "step": 140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.990303724928367e-05,
+      "loss": 0.2431,
+      "step": 141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.990234957020057e-05,
+      "loss": 0.2577,
+      "step": 142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.990166189111748e-05,
+      "loss": 0.2399,
+      "step": 143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.990097421203438e-05,
+      "loss": 0.2923,
+      "step": 144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.990028653295129e-05,
+      "loss": 0.2527,
+      "step": 145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.989959885386819e-05,
+      "loss": 0.2253,
+      "step": 146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.98989111747851e-05,
+      "loss": 0.3546,
+      "step": 147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.9898223495702e-05,
+      "loss": 0.2669,
+      "step": 148
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.989753581661891e-05,
+      "loss": 0.2055,
+      "step": 149
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.989684813753581e-05,
+      "loss": 0.2274,
+      "step": 150
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.989616045845272e-05,
+      "loss": 0.1944,
+      "step": 151
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9895472779369624e-05,
+      "loss": 0.2582,
+      "step": 152
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.989478510028653e-05,
+      "loss": 0.2667,
+      "step": 153
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9894097421203435e-05,
+      "loss": 0.1573,
+      "step": 154
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9893409742120344e-05,
+      "loss": 0.1969,
+      "step": 155
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9892722063037246e-05,
+      "loss": 0.2237,
+      "step": 156
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9892034383954155e-05,
+      "loss": 0.3058,
+      "step": 157
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.989134670487106e-05,
+      "loss": 0.2552,
+      "step": 158
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.9890659025787966e-05,
+      "loss": 0.239,
+      "step": 159
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.988997134670487e-05,
+      "loss": 0.2113,
+      "step": 160
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.988928366762178e-05,
+      "loss": 0.3054,
+      "step": 161
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.988859598853868e-05,
+      "loss": 0.2344,
+      "step": 162
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.988790830945559e-05,
+      "loss": 0.2481,
+      "step": 163
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.988722063037249e-05,
+      "loss": 0.1425,
+      "step": 164
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.98865329512894e-05,
+      "loss": 0.2385,
+      "step": 165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.98858452722063e-05,
+      "loss": 0.1796,
+      "step": 166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.988515759312321e-05,
+      "loss": 0.1914,
+      "step": 167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.988446991404011e-05,
+      "loss": 0.2118,
+      "step": 168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.988378223495702e-05,
+      "loss": 0.2171,
+      "step": 169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.988309455587392e-05,
+      "loss": 0.1681,
+      "step": 170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.988240687679083e-05,
+      "loss": 0.1962,
+      "step": 171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9881719197707734e-05,
+      "loss": 0.1975,
+      "step": 172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.988103151862464e-05,
+      "loss": 0.1753,
+      "step": 173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9880343839541545e-05,
+      "loss": 0.2721,
+      "step": 174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9879656160458454e-05,
+      "loss": 0.1799,
+      "step": 175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9878968481375356e-05,
+      "loss": 0.182,
+      "step": 176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9878280802292265e-05,
+      "loss": 0.1501,
+      "step": 177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.987759312320917e-05,
+      "loss": 0.3676,
+      "step": 178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.9876905444126076e-05,
+      "loss": 0.2557,
+      "step": 179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.987621776504298e-05,
+      "loss": 0.2652,
+      "step": 180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.987553008595989e-05,
+      "loss": 0.2533,
+      "step": 181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.987484240687679e-05,
+      "loss": 0.2732,
+      "step": 182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.98741547277937e-05,
+      "loss": 0.1912,
+      "step": 183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.98734670487106e-05,
+      "loss": 0.2619,
+      "step": 184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.987277936962751e-05,
+      "loss": 0.1948,
+      "step": 185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.987209169054441e-05,
+      "loss": 0.1944,
+      "step": 186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.987140401146132e-05,
+      "loss": 0.2556,
+      "step": 187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.987071633237822e-05,
+      "loss": 0.2742,
+      "step": 188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.987002865329513e-05,
+      "loss": 0.1958,
+      "step": 189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.986934097421203e-05,
+      "loss": 0.219,
+      "step": 190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.986865329512894e-05,
+      "loss": 0.338,
+      "step": 191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9867965616045844e-05,
+      "loss": 0.1982,
+      "step": 192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.986727793696275e-05,
+      "loss": 0.1716,
+      "step": 193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9866590257879655e-05,
+      "loss": 0.2564,
+      "step": 194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9865902578796564e-05,
+      "loss": 0.1456,
+      "step": 195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9865214899713466e-05,
+      "loss": 0.2309,
+      "step": 196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9864527220630375e-05,
+      "loss": 0.3244,
+      "step": 197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.986383954154728e-05,
+      "loss": 0.1774,
+      "step": 198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.9863151862464186e-05,
+      "loss": 0.2406,
+      "step": 199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.986246418338109e-05,
+      "loss": 0.1904,
+      "step": 200
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9861776504298e-05,
+      "loss": 0.1784,
+      "step": 201
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.98610888252149e-05,
+      "loss": 0.177,
+      "step": 202
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.986040114613181e-05,
+      "loss": 0.174,
+      "step": 203
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985971346704871e-05,
+      "loss": 0.2394,
+      "step": 204
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985902578796562e-05,
+      "loss": 0.1645,
+      "step": 205
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985833810888252e-05,
+      "loss": 0.1625,
+      "step": 206
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985765042979943e-05,
+      "loss": 0.2518,
+      "step": 207
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985696275071633e-05,
+      "loss": 0.2166,
+      "step": 208
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985627507163324e-05,
+      "loss": 0.1953,
+      "step": 209
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985558739255014e-05,
+      "loss": 0.2996,
+      "step": 210
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985489971346705e-05,
+      "loss": 0.184,
+      "step": 211
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9854212034383954e-05,
+      "loss": 0.2016,
+      "step": 212
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985352435530086e-05,
+      "loss": 0.1686,
+      "step": 213
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9852836676217765e-05,
+      "loss": 0.1902,
+      "step": 214
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9852148997134674e-05,
+      "loss": 0.2011,
+      "step": 215
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9851461318051576e-05,
+      "loss": 0.2071,
+      "step": 216
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.9850773638968485e-05,
+      "loss": 0.2377,
+      "step": 217
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.985008595988539e-05,
+      "loss": 0.2537,
+      "step": 218
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9849398280802296e-05,
+      "loss": 0.1914,
+      "step": 219
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.98487106017192e-05,
+      "loss": 0.1959,
+      "step": 220
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984802292263611e-05,
+      "loss": 0.2338,
+      "step": 221
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984733524355301e-05,
+      "loss": 0.2696,
+      "step": 222
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984664756446992e-05,
+      "loss": 0.2298,
+      "step": 223
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984595988538682e-05,
+      "loss": 0.1687,
+      "step": 224
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984527220630373e-05,
+      "loss": 0.1857,
+      "step": 225
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984458452722063e-05,
+      "loss": 0.1692,
+      "step": 226
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984389684813754e-05,
+      "loss": 0.1894,
+      "step": 227
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984320916905444e-05,
+      "loss": 0.2299,
+      "step": 228
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984252148997135e-05,
+      "loss": 0.2177,
+      "step": 229
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984183381088825e-05,
+      "loss": 0.1783,
+      "step": 230
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.984114613180516e-05,
+      "loss": 0.2779,
+      "step": 231
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9840458452722064e-05,
+      "loss": 0.1645,
+      "step": 232
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.983977077363897e-05,
+      "loss": 0.1328,
+      "step": 233
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9839083094555875e-05,
+      "loss": 0.2176,
+      "step": 234
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.9838395415472784e-05,
+      "loss": 0.2471,
+      "step": 235
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9837707736389686e-05,
+      "loss": 0.1444,
+      "step": 236
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9837020057306595e-05,
+      "loss": 0.1931,
+      "step": 237
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98363323782235e-05,
+      "loss": 0.1686,
+      "step": 238
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.98356446991404e-05,
+      "loss": 0.1444,
+      "step": 239
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983495702005731e-05,
+      "loss": 0.2147,
+      "step": 240
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983426934097421e-05,
+      "loss": 0.2184,
+      "step": 241
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983358166189112e-05,
+      "loss": 0.1524,
+      "step": 242
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983289398280802e-05,
+      "loss": 0.2056,
+      "step": 243
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983220630372493e-05,
+      "loss": 0.2485,
+      "step": 244
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983151862464183e-05,
+      "loss": 0.1607,
+      "step": 245
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983083094555874e-05,
+      "loss": 0.1999,
+      "step": 246
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.983014326647564e-05,
+      "loss": 0.3184,
+      "step": 247
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982945558739255e-05,
+      "loss": 0.2132,
+      "step": 248
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9828767908309454e-05,
+      "loss": 0.2113,
+      "step": 249
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.982808022922636e-05,
+      "loss": 0.2697,
+      "step": 250
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9827392550143265e-05,
+      "loss": 0.1837,
+      "step": 251
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9826704871060174e-05,
+      "loss": 0.2367,
+      "step": 252
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.9826017191977076e-05,
+      "loss": 0.2209,
+      "step": 253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9825329512893985e-05,
+      "loss": 0.2175,
+      "step": 254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.982464183381089e-05,
+      "loss": 0.176,
+      "step": 255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9823954154727796e-05,
+      "loss": 0.3049,
+      "step": 256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.98232664756447e-05,
+      "loss": 0.1747,
+      "step": 257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.982257879656161e-05,
+      "loss": 0.205,
+      "step": 258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.982189111747851e-05,
+      "loss": 0.1843,
+      "step": 259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.982120343839542e-05,
+      "loss": 0.2379,
+      "step": 260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.982051575931232e-05,
+      "loss": 0.2222,
+      "step": 261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981982808022923e-05,
+      "loss": 0.1902,
+      "step": 262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981914040114613e-05,
+      "loss": 0.1963,
+      "step": 263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981845272206304e-05,
+      "loss": 0.2889,
+      "step": 264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981776504297994e-05,
+      "loss": 0.3008,
+      "step": 265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981707736389685e-05,
+      "loss": 0.1505,
+      "step": 266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981638968481375e-05,
+      "loss": 0.1801,
+      "step": 267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981570200573066e-05,
+      "loss": 0.1869,
+      "step": 268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.9815014326647564e-05,
+      "loss": 0.2073,
+      "step": 269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.981432664756447e-05,
+      "loss": 0.1716,
+      "step": 270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9813638968481375e-05,
+      "loss": 0.1229,
+      "step": 271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9812951289398284e-05,
+      "loss": 0.2138,
+      "step": 272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9812263610315186e-05,
+      "loss": 0.1739,
+      "step": 273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9811575931232095e-05,
+      "loss": 0.1784,
+      "step": 274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9810888252149e-05,
+      "loss": 0.1923,
+      "step": 275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.9810200573065906e-05,
+      "loss": 0.3102,
+      "step": 276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980951289398281e-05,
+      "loss": 0.1621,
+      "step": 277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980882521489972e-05,
+      "loss": 0.2948,
+      "step": 278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980813753581662e-05,
+      "loss": 0.213,
+      "step": 279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980744985673353e-05,
+      "loss": 0.1932,
+      "step": 280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980676217765043e-05,
+      "loss": 0.1362,
+      "step": 281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980607449856734e-05,
+      "loss": 0.3054,
+      "step": 282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980538681948424e-05,
+      "loss": 0.2724,
+      "step": 283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980469914040115e-05,
+      "loss": 0.1893,
+      "step": 284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980401146131805e-05,
+      "loss": 0.228,
+      "step": 285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980332378223496e-05,
+      "loss": 0.1738,
+      "step": 286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 5.980263610315186e-05,
+      "loss": 0.1171,
+      "step": 287
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.980194842406877e-05,
+      "loss": 0.1934,
+      "step": 288
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9801260744985674e-05,
+      "loss": 0.2257,
+      "step": 289
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.980057306590258e-05,
+      "loss": 0.1123,
+      "step": 290
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9799885386819485e-05,
+      "loss": 0.1878,
+      "step": 291
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9799197707736394e-05,
+      "loss": 0.2699,
+      "step": 292
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9798510028653296e-05,
+      "loss": 0.2399,
+      "step": 293
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9797822349570205e-05,
+      "loss": 0.1515,
+      "step": 294
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979713467048711e-05,
+      "loss": 0.12,
+      "step": 295
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.9796446991404016e-05,
+      "loss": 0.1506,
+      "step": 296
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979575931232092e-05,
+      "loss": 0.1228,
+      "step": 297
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979507163323783e-05,
+      "loss": 0.2085,
+      "step": 298
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979438395415473e-05,
+      "loss": 0.1893,
+      "step": 299
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979369627507164e-05,
+      "loss": 0.2368,
+      "step": 300
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979300859598854e-05,
+      "loss": 0.1665,
+      "step": 301
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979232091690545e-05,
+      "loss": 0.2553,
+      "step": 302
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979163323782235e-05,
+      "loss": 0.1714,
+      "step": 303
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979094555873926e-05,
+      "loss": 0.1541,
+      "step": 304
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 5.979025787965616e-05,
+      "loss": 0.1149,
+      "step": 305
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.978957020057307e-05,
+      "loss": 0.2305,
+      "step": 306
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.978888252148997e-05,
+      "loss": 0.1547,
+      "step": 307
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.978819484240688e-05,
+      "loss": 0.1621,
+      "step": 308
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9787507163323784e-05,
+      "loss": 0.2161,
+      "step": 309
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.978681948424069e-05,
+      "loss": 0.1832,
+      "step": 310
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9786131805157595e-05,
+      "loss": 0.2154,
+      "step": 311
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9785444126074504e-05,
+      "loss": 0.224,
+      "step": 312
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9784756446991406e-05,
+      "loss": 0.1805,
+      "step": 313
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9784068767908315e-05,
+      "loss": 0.2821,
+      "step": 314
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.978338108882522e-05,
+      "loss": 0.1361,
+      "step": 315
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.9782693409742126e-05,
+      "loss": 0.1352,
+      "step": 316
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.978200573065903e-05,
+      "loss": 0.2582,
+      "step": 317
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.978131805157594e-05,
+      "loss": 0.1733,
+      "step": 318
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.978063037249284e-05,
+      "loss": 0.2724,
+      "step": 319
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.977994269340975e-05,
+      "loss": 0.1075,
+      "step": 320
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.977925501432665e-05,
+      "loss": 0.1385,
+      "step": 321
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 5.977856733524356e-05,
+      "loss": 0.1924,
+      "step": 322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.977787965616046e-05,
+      "loss": 0.272,
+      "step": 323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.977719197707737e-05,
+      "loss": 0.1758,
+      "step": 324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.977650429799427e-05,
+      "loss": 0.1909,
+      "step": 325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.977581661891118e-05,
+      "loss": 0.1497,
+      "step": 326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.977512893982808e-05,
+      "loss": 0.1773,
+      "step": 327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.977444126074499e-05,
+      "loss": 0.1753,
+      "step": 328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9773753581661894e-05,
+      "loss": 0.1965,
+      "step": 329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.97730659025788e-05,
+      "loss": 0.1481,
+      "step": 330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9772378223495705e-05,
+      "loss": 0.1776,
+      "step": 331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9771690544412614e-05,
+      "loss": 0.1547,
+      "step": 332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9771002865329516e-05,
+      "loss": 0.1889,
+      "step": 333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9770315186246425e-05,
+      "loss": 0.19,
+      "step": 334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.976962750716333e-05,
+      "loss": 0.1498,
+      "step": 335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.9768939828080236e-05,
+      "loss": 0.1436,
+      "step": 336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.976825214899714e-05,
+      "loss": 0.2276,
+      "step": 337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.976756446991405e-05,
+      "loss": 0.249,
+      "step": 338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.976687679083095e-05,
+      "loss": 0.1699,
+      "step": 339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 5.976618911174786e-05,
+      "loss": 0.1637,
+      "step": 340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.976550143266476e-05,
+      "loss": 0.1806,
+      "step": 341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.976481375358167e-05,
+      "loss": 0.1991,
+      "step": 342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.976412607449857e-05,
+      "loss": 0.1599,
+      "step": 343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.976343839541548e-05,
+      "loss": 0.2169,
+      "step": 344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.976275071633238e-05,
+      "loss": 0.2298,
+      "step": 345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.976206303724929e-05,
+      "loss": 0.2505,
+      "step": 346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.976137535816619e-05,
+      "loss": 0.2006,
+      "step": 347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.97606876790831e-05,
+      "loss": 0.1542,
+      "step": 348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9760000000000004e-05,
+      "loss": 0.1692,
+      "step": 349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.975931232091691e-05,
+      "loss": 0.1433,
+      "step": 350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9758624641833815e-05,
+      "loss": 0.1753,
+      "step": 351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9757936962750724e-05,
+      "loss": 0.1313,
+      "step": 352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9757249283667626e-05,
+      "loss": 0.1198,
+      "step": 353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9756561604584535e-05,
+      "loss": 0.1474,
+      "step": 354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.975587392550144e-05,
+      "loss": 0.2345,
+      "step": 355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.9755186246418346e-05,
+      "loss": 0.1223,
+      "step": 356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 5.975449856733525e-05,
+      "loss": 0.2108,
+      "step": 357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.975381088825215e-05,
+      "loss": 0.2011,
+      "step": 358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.975312320916906e-05,
+      "loss": 0.0957,
+      "step": 359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.975243553008596e-05,
+      "loss": 0.1427,
+      "step": 360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.975174785100287e-05,
+      "loss": 0.1581,
+      "step": 361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.975106017191977e-05,
+      "loss": 0.1397,
+      "step": 362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.975037249283668e-05,
+      "loss": 0.171,
+      "step": 363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.974968481375358e-05,
+      "loss": 0.3295,
+      "step": 364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.974899713467049e-05,
+      "loss": 0.1431,
+      "step": 365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9748309455587394e-05,
+      "loss": 0.2408,
+      "step": 366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.97476217765043e-05,
+      "loss": 0.1175,
+      "step": 367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9746934097421205e-05,
+      "loss": 0.161,
+      "step": 368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9746246418338114e-05,
+      "loss": 0.1561,
+      "step": 369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9745558739255016e-05,
+      "loss": 0.1349,
+      "step": 370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9744871060171925e-05,
+      "loss": 0.166,
+      "step": 371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.974418338108883e-05,
+      "loss": 0.1455,
+      "step": 372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.9743495702005736e-05,
+      "loss": 0.1827,
+      "step": 373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.974280802292264e-05,
+      "loss": 0.1325,
+      "step": 374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 5.974212034383955e-05,
+      "loss": 0.1925,
+      "step": 375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.974143266475645e-05,
+      "loss": 0.1532,
+      "step": 376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.974074498567336e-05,
+      "loss": 0.1776,
+      "step": 377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.974005730659026e-05,
+      "loss": 0.1508,
+      "step": 378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.973936962750717e-05,
+      "loss": 0.209,
+      "step": 379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.973868194842407e-05,
+      "loss": 0.1788,
+      "step": 380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.973799426934098e-05,
+      "loss": 0.1827,
+      "step": 381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.973730659025788e-05,
+      "loss": 0.1318,
+      "step": 382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9736618911174784e-05,
+      "loss": 0.1422,
+      "step": 383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9735931232091686e-05,
+      "loss": 0.1512,
+      "step": 384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9735243553008595e-05,
+      "loss": 0.1357,
+      "step": 385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.97345558739255e-05,
+      "loss": 0.1137,
+      "step": 386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.9733868194842406e-05,
+      "loss": 0.227,
+      "step": 387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.973318051575931e-05,
+      "loss": 0.1464,
+      "step": 388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.973249283667622e-05,
+      "loss": 0.1474,
+      "step": 389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.973180515759312e-05,
+      "loss": 0.1024,
+      "step": 390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.973111747851003e-05,
+      "loss": 0.1596,
+      "step": 391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 5.973042979942693e-05,
+      "loss": 0.2358,
+      "step": 392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.972974212034384e-05,
+      "loss": 0.1715,
+      "step": 393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.972905444126074e-05,
+      "loss": 0.197,
+      "step": 394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.972836676217765e-05,
+      "loss": 0.1581,
+      "step": 395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.972767908309455e-05,
+      "loss": 0.135,
+      "step": 396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.972699140401146e-05,
+      "loss": 0.3343,
+      "step": 397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.972630372492836e-05,
+      "loss": 0.1639,
+      "step": 398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.972561604584527e-05,
+      "loss": 0.1605,
+      "step": 399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9724928366762174e-05,
+      "loss": 0.193,
+      "step": 400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.972424068767908e-05,
+      "loss": 0.1659,
+      "step": 401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9723553008595985e-05,
+      "loss": 0.1662,
+      "step": 402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9722865329512894e-05,
+      "loss": 0.1965,
+      "step": 403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9722177650429796e-05,
+      "loss": 0.1583,
+      "step": 404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9721489971346705e-05,
+      "loss": 0.1943,
+      "step": 405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.972080229226361e-05,
+      "loss": 0.2675,
+      "step": 406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.9720114613180516e-05,
+      "loss": 0.1724,
+      "step": 407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.971942693409742e-05,
+      "loss": 0.1836,
+      "step": 408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.971873925501433e-05,
+      "loss": 0.1852,
+      "step": 409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 5.971805157593123e-05,
+      "loss": 0.2409,
+      "step": 410
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.971736389684814e-05,
+      "loss": 0.1508,
+      "step": 411
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.971667621776504e-05,
+      "loss": 0.1536,
+      "step": 412
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.971598853868195e-05,
+      "loss": 0.1679,
+      "step": 413
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.971530085959885e-05,
+      "loss": 0.1484,
+      "step": 414
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.971461318051576e-05,
+      "loss": 0.1786,
+      "step": 415
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.971392550143266e-05,
+      "loss": 0.1528,
+      "step": 416
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.971323782234957e-05,
+      "loss": 0.1251,
+      "step": 417
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.971255014326647e-05,
+      "loss": 0.132,
+      "step": 418
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.971186246418338e-05,
+      "loss": 0.173,
+      "step": 419
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9711174785100284e-05,
+      "loss": 0.1692,
+      "step": 420
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.971048710601719e-05,
+      "loss": 0.1618,
+      "step": 421
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9709799426934095e-05,
+      "loss": 0.2132,
+      "step": 422
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9709111747851004e-05,
+      "loss": 0.1522,
+      "step": 423
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9708424068767906e-05,
+      "loss": 0.1638,
+      "step": 424
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9707736389684815e-05,
+      "loss": 0.222,
+      "step": 425
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.970704871060172e-05,
+      "loss": 0.21,
+      "step": 426
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 5.9706361031518626e-05,
+      "loss": 0.1399,
+      "step": 427
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.970567335243553e-05,
+      "loss": 0.1465,
+      "step": 428
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.970498567335244e-05,
+      "loss": 0.1746,
+      "step": 429
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.970429799426934e-05,
+      "loss": 0.2318,
+      "step": 430
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.970361031518625e-05,
+      "loss": 0.2051,
+      "step": 431
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.970292263610315e-05,
+      "loss": 0.195,
+      "step": 432
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.970223495702006e-05,
+      "loss": 0.152,
+      "step": 433
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.970154727793696e-05,
+      "loss": 0.1627,
+      "step": 434
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.970085959885387e-05,
+      "loss": 0.1335,
+      "step": 435
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.970017191977077e-05,
+      "loss": 0.2683,
+      "step": 436
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.969948424068768e-05,
+      "loss": 0.1443,
+      "step": 437
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.969879656160458e-05,
+      "loss": 0.1547,
+      "step": 438
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.969810888252149e-05,
+      "loss": 0.1581,
+      "step": 439
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9697421203438394e-05,
+      "loss": 0.1604,
+      "step": 440
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.96967335243553e-05,
+      "loss": 0.1912,
+      "step": 441
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9696045845272205e-05,
+      "loss": 0.0997,
+      "step": 442
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9695358166189114e-05,
+      "loss": 0.1898,
+      "step": 443
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 5.9694670487106016e-05,
+      "loss": 0.1503,
+      "step": 444
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9693982808022925e-05,
+      "loss": 0.1959,
+      "step": 445
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.969329512893983e-05,
+      "loss": 0.2347,
+      "step": 446
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9692607449856736e-05,
+      "loss": 0.1732,
+      "step": 447
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.969191977077364e-05,
+      "loss": 0.204,
+      "step": 448
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.969123209169055e-05,
+      "loss": 0.1749,
+      "step": 449
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.969054441260745e-05,
+      "loss": 0.1866,
+      "step": 450
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.968985673352436e-05,
+      "loss": 0.1335,
+      "step": 451
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.968916905444126e-05,
+      "loss": 0.1631,
+      "step": 452
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.968848137535817e-05,
+      "loss": 0.2241,
+      "step": 453
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.968779369627507e-05,
+      "loss": 0.1089,
+      "step": 454
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.968710601719198e-05,
+      "loss": 0.2322,
+      "step": 455
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.968641833810888e-05,
+      "loss": 0.1708,
+      "step": 456
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.968573065902579e-05,
+      "loss": 0.1437,
+      "step": 457
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.968504297994269e-05,
+      "loss": 0.1446,
+      "step": 458
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.96843553008596e-05,
+      "loss": 0.084,
+      "step": 459
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9683667621776504e-05,
+      "loss": 0.1557,
+      "step": 460
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.968297994269341e-05,
+      "loss": 0.1352,
+      "step": 461
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 5.9682292263610315e-05,
+      "loss": 0.1521,
+      "step": 462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9681604584527224e-05,
+      "loss": 0.1506,
+      "step": 463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9680916905444126e-05,
+      "loss": 0.1567,
+      "step": 464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9680229226361035e-05,
+      "loss": 0.0945,
+      "step": 465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.967954154727794e-05,
+      "loss": 0.1259,
+      "step": 466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9678853868194846e-05,
+      "loss": 0.1758,
+      "step": 467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.967816618911175e-05,
+      "loss": 0.0896,
+      "step": 468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.967747851002866e-05,
+      "loss": 0.2271,
+      "step": 469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.967679083094556e-05,
+      "loss": 0.0954,
+      "step": 470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.967610315186247e-05,
+      "loss": 0.1784,
+      "step": 471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.967541547277937e-05,
+      "loss": 0.1551,
+      "step": 472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.967472779369628e-05,
+      "loss": 0.1447,
+      "step": 473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.967404011461318e-05,
+      "loss": 0.2394,
+      "step": 474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.967335243553009e-05,
+      "loss": 0.2772,
+      "step": 475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.967266475644699e-05,
+      "loss": 0.2042,
+      "step": 476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9671977077363894e-05,
+      "loss": 0.1517,
+      "step": 477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.96712893982808e-05,
+      "loss": 0.1303,
+      "step": 478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 5.9670601719197705e-05,
+      "loss": 0.1949,
+      "step": 479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9669914040114614e-05,
+      "loss": 0.1489,
+      "step": 480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9669226361031516e-05,
+      "loss": 0.1362,
+      "step": 481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9668538681948425e-05,
+      "loss": 0.1949,
+      "step": 482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966785100286533e-05,
+      "loss": 0.141,
+      "step": 483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9667163323782236e-05,
+      "loss": 0.1284,
+      "step": 484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966647564469914e-05,
+      "loss": 0.1829,
+      "step": 485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966578796561605e-05,
+      "loss": 0.153,
+      "step": 486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966510028653295e-05,
+      "loss": 0.1276,
+      "step": 487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966441260744986e-05,
+      "loss": 0.2111,
+      "step": 488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966372492836676e-05,
+      "loss": 0.3088,
+      "step": 489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966303724928367e-05,
+      "loss": 0.1828,
+      "step": 490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966234957020057e-05,
+      "loss": 0.1579,
+      "step": 491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966166189111748e-05,
+      "loss": 0.1397,
+      "step": 492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966097421203438e-05,
+      "loss": 0.16,
+      "step": 493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.966028653295129e-05,
+      "loss": 0.1313,
+      "step": 494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.965959885386819e-05,
+      "loss": 0.1615,
+      "step": 495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.96589111747851e-05,
+      "loss": 0.2895,
+      "step": 496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 5.9658223495702004e-05,
+      "loss": 0.1667,
+      "step": 497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.965753581661891e-05,
+      "loss": 0.1691,
+      "step": 498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9656848137535815e-05,
+      "loss": 0.1256,
+      "step": 499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9656160458452724e-05,
+      "loss": 0.1784,
+      "step": 500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9655472779369626e-05,
+      "loss": 0.1355,
+      "step": 501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9654785100286535e-05,
+      "loss": 0.1781,
+      "step": 502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.965409742120344e-05,
+      "loss": 0.2338,
+      "step": 503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.9653409742120346e-05,
+      "loss": 0.1139,
+      "step": 504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.965272206303725e-05,
+      "loss": 0.1871,
+      "step": 505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.965203438395416e-05,
+      "loss": 0.1621,
+      "step": 506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.965134670487106e-05,
+      "loss": 0.1334,
+      "step": 507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.965065902578797e-05,
+      "loss": 0.1906,
+      "step": 508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.964997134670487e-05,
+      "loss": 0.1044,
+      "step": 509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.964928366762178e-05,
+      "loss": 0.1248,
+      "step": 510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.964859598853868e-05,
+      "loss": 0.2332,
+      "step": 511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.964790830945559e-05,
+      "loss": 0.1785,
+      "step": 512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.964722063037249e-05,
+      "loss": 0.1306,
+      "step": 513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 5.96465329512894e-05,
+      "loss": 0.1868,
+      "step": 514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.96458452722063e-05,
+      "loss": 0.145,
+      "step": 515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.964515759312321e-05,
+      "loss": 0.1745,
+      "step": 516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9644469914040114e-05,
+      "loss": 0.1365,
+      "step": 517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.964378223495702e-05,
+      "loss": 0.1399,
+      "step": 518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9643094555873925e-05,
+      "loss": 0.1204,
+      "step": 519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9642406876790834e-05,
+      "loss": 0.1551,
+      "step": 520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9641719197707736e-05,
+      "loss": 0.116,
+      "step": 521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9641031518624645e-05,
+      "loss": 0.3553,
+      "step": 522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.964034383954155e-05,
+      "loss": 0.1592,
+      "step": 523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.9639656160458456e-05,
+      "loss": 0.1256,
+      "step": 524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.963896848137536e-05,
+      "loss": 0.1187,
+      "step": 525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.963828080229227e-05,
+      "loss": 0.1975,
+      "step": 526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.963759312320917e-05,
+      "loss": 0.2619,
+      "step": 527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.963690544412608e-05,
+      "loss": 0.1045,
+      "step": 528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.963621776504298e-05,
+      "loss": 0.144,
+      "step": 529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.963553008595989e-05,
+      "loss": 0.1828,
+      "step": 530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.963484240687679e-05,
+      "loss": 0.1285,
+      "step": 531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 5.96341547277937e-05,
+      "loss": 0.1418,
+      "step": 532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.96334670487106e-05,
+      "loss": 0.1526,
+      "step": 533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.963277936962751e-05,
+      "loss": 0.126,
+      "step": 534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.963209169054441e-05,
+      "loss": 0.1754,
+      "step": 535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.963140401146132e-05,
+      "loss": 0.1441,
+      "step": 536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9630716332378224e-05,
+      "loss": 0.1584,
+      "step": 537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.963002865329513e-05,
+      "loss": 0.2284,
+      "step": 538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9629340974212035e-05,
+      "loss": 0.1103,
+      "step": 539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9628653295128944e-05,
+      "loss": 0.1997,
+      "step": 540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9627965616045846e-05,
+      "loss": 0.2078,
+      "step": 541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9627277936962755e-05,
+      "loss": 0.1651,
+      "step": 542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.962659025787966e-05,
+      "loss": 0.1498,
+      "step": 543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.9625902578796566e-05,
+      "loss": 0.114,
+      "step": 544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.962521489971347e-05,
+      "loss": 0.1343,
+      "step": 545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.962452722063038e-05,
+      "loss": 0.1681,
+      "step": 546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.962383954154728e-05,
+      "loss": 0.2442,
+      "step": 547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.962315186246419e-05,
+      "loss": 0.1697,
+      "step": 548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 5.962246418338109e-05,
+      "loss": 0.126,
+      "step": 549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9621776504298e-05,
+      "loss": 0.1193,
+      "step": 550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.96210888252149e-05,
+      "loss": 0.2145,
+      "step": 551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.962040114613181e-05,
+      "loss": 0.1572,
+      "step": 552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.961971346704871e-05,
+      "loss": 0.0909,
+      "step": 553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.961902578796562e-05,
+      "loss": 0.2248,
+      "step": 554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.961833810888252e-05,
+      "loss": 0.1328,
+      "step": 555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.961765042979943e-05,
+      "loss": 0.1382,
+      "step": 556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9616962750716334e-05,
+      "loss": 0.1795,
+      "step": 557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.961627507163324e-05,
+      "loss": 0.1228,
+      "step": 558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9615587392550145e-05,
+      "loss": 0.1341,
+      "step": 559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9614899713467054e-05,
+      "loss": 0.1079,
+      "step": 560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9614212034383956e-05,
+      "loss": 0.1296,
+      "step": 561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9613524355300865e-05,
+      "loss": 0.1149,
+      "step": 562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.961283667621777e-05,
+      "loss": 0.0885,
+      "step": 563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.9612148997134676e-05,
+      "loss": 0.12,
+      "step": 564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.961146131805158e-05,
+      "loss": 0.0928,
+      "step": 565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.961077363896849e-05,
+      "loss": 0.1105,
+      "step": 566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 5.961008595988539e-05,
+      "loss": 0.2385,
+      "step": 567
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.96093982808023e-05,
+      "loss": 0.0923,
+      "step": 568
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.96087106017192e-05,
+      "loss": 0.0988,
+      "step": 569
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.960802292263611e-05,
+      "loss": 0.1513,
+      "step": 570
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.960733524355301e-05,
+      "loss": 0.1414,
+      "step": 571
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.960664756446992e-05,
+      "loss": 0.262,
+      "step": 572
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.960595988538682e-05,
+      "loss": 0.137,
+      "step": 573
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.960527220630373e-05,
+      "loss": 0.1255,
+      "step": 574
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.960458452722063e-05,
+      "loss": 0.0927,
+      "step": 575
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.960389684813754e-05,
+      "loss": 0.1155,
+      "step": 576
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9603209169054444e-05,
+      "loss": 0.1182,
+      "step": 577
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.960252148997135e-05,
+      "loss": 0.1472,
+      "step": 578
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9601833810888255e-05,
+      "loss": 0.2142,
+      "step": 579
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9601146131805164e-05,
+      "loss": 0.1532,
+      "step": 580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9600458452722066e-05,
+      "loss": 0.1244,
+      "step": 581
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9599770773638975e-05,
+      "loss": 0.1511,
+      "step": 582
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.959908309455588e-05,
+      "loss": 0.1971,
+      "step": 583
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 5.9598395415472786e-05,
+      "loss": 0.1044,
+      "step": 584
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.959770773638969e-05,
+      "loss": 0.1789,
+      "step": 585
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.95970200573066e-05,
+      "loss": 0.1267,
+      "step": 586
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.95963323782235e-05,
+      "loss": 0.1251,
+      "step": 587
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.959564469914041e-05,
+      "loss": 0.1628,
+      "step": 588
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.959495702005731e-05,
+      "loss": 0.1857,
+      "step": 589
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.959426934097422e-05,
+      "loss": 0.1495,
+      "step": 590
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.959358166189112e-05,
+      "loss": 0.1452,
+      "step": 591
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.959289398280803e-05,
+      "loss": 0.0975,
+      "step": 592
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.959220630372493e-05,
+      "loss": 0.1487,
+      "step": 593
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.959151862464184e-05,
+      "loss": 0.1791,
+      "step": 594
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.959083094555874e-05,
+      "loss": 0.1531,
+      "step": 595
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9590143266475645e-05,
+      "loss": 0.1405,
+      "step": 596
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9589455587392554e-05,
+      "loss": 0.1328,
+      "step": 597
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9588767908309456e-05,
+      "loss": 0.1525,
+      "step": 598
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9588080229226365e-05,
+      "loss": 0.1616,
+      "step": 599
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.958739255014327e-05,
+      "loss": 0.1647,
+      "step": 600
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.9586704871060176e-05,
+      "loss": 0.1919,
+      "step": 601
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 5.958601719197708e-05,
+      "loss": 0.1267,
+      "step": 602
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.958532951289399e-05,
+      "loss": 0.1782,
+      "step": 603
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.958464183381089e-05,
+      "loss": 0.1401,
+      "step": 604
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.95839541547278e-05,
+      "loss": 0.1154,
+      "step": 605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.95832664756447e-05,
+      "loss": 0.1837,
+      "step": 606
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.958257879656161e-05,
+      "loss": 0.1788,
+      "step": 607
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.958189111747851e-05,
+      "loss": 0.1801,
+      "step": 608
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.958120343839542e-05,
+      "loss": 0.1256,
+      "step": 609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.958051575931232e-05,
+      "loss": 0.1244,
+      "step": 610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.957982808022923e-05,
+      "loss": 0.1661,
+      "step": 611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.957914040114613e-05,
+      "loss": 0.1342,
+      "step": 612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.957845272206304e-05,
+      "loss": 0.1385,
+      "step": 613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.9577765042979944e-05,
+      "loss": 0.14,
+      "step": 614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.957707736389685e-05,
+      "loss": 0.148,
+      "step": 615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.9576389684813755e-05,
+      "loss": 0.2178,
+      "step": 616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.9575702005730664e-05,
+      "loss": 0.168,
+      "step": 617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.9575014326647566e-05,
+      "loss": 0.0941,
+      "step": 618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 5.9574326647564475e-05,
+      "loss": 0.1644,
+      "step": 619
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.957363896848138e-05,
+      "loss": 0.0632,
+      "step": 620
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.9572951289398286e-05,
+      "loss": 0.1138,
+      "step": 621
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.957226361031519e-05,
+      "loss": 0.1045,
+      "step": 622
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.95715759312321e-05,
+      "loss": 0.1027,
+      "step": 623
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.9570888252149e-05,
+      "loss": 0.2079,
+      "step": 624
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.957020057306591e-05,
+      "loss": 0.104,
+      "step": 625
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.956951289398281e-05,
+      "loss": 0.1254,
+      "step": 626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.956882521489972e-05,
+      "loss": 0.1373,
+      "step": 627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.956813753581662e-05,
+      "loss": 0.1334,
+      "step": 628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.956744985673353e-05,
+      "loss": 0.1247,
+      "step": 629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.956676217765043e-05,
+      "loss": 0.145,
+      "step": 630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.956607449856734e-05,
+      "loss": 0.1296,
+      "step": 631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.956538681948424e-05,
+      "loss": 0.1464,
+      "step": 632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.956469914040115e-05,
+      "loss": 0.163,
+      "step": 633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.9564011461318054e-05,
+      "loss": 0.1338,
+      "step": 634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.956332378223496e-05,
+      "loss": 0.1553,
+      "step": 635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 5.9562636103151865e-05,
+      "loss": 0.2076,
+      "step": 636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.956194842406877e-05,
+      "loss": 0.1345,
+      "step": 637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.956126074498567e-05,
+      "loss": 0.1918,
+      "step": 638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.956057306590258e-05,
+      "loss": 0.3314,
+      "step": 639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.955988538681948e-05,
+      "loss": 0.1369,
+      "step": 640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.955919770773639e-05,
+      "loss": 0.1366,
+      "step": 641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.955851002865329e-05,
+      "loss": 0.1611,
+      "step": 642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.95578223495702e-05,
+      "loss": 0.1794,
+      "step": 643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.95571346704871e-05,
+      "loss": 0.1565,
+      "step": 644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.955644699140401e-05,
+      "loss": 0.1092,
+      "step": 645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.955575931232091e-05,
+      "loss": 0.2177,
+      "step": 646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.955507163323782e-05,
+      "loss": 0.1229,
+      "step": 647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.9554383954154724e-05,
+      "loss": 0.1634,
+      "step": 648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.955369627507163e-05,
+      "loss": 0.1535,
+      "step": 649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.9553008595988535e-05,
+      "loss": 0.1121,
+      "step": 650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.9552320916905444e-05,
+      "loss": 0.1457,
+      "step": 651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.9551633237822346e-05,
+      "loss": 0.1222,
+      "step": 652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.9550945558739255e-05,
+      "loss": 0.093,
+      "step": 653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 5.955025787965616e-05,
+      "loss": 0.1303,
+      "step": 654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.9549570200573066e-05,
+      "loss": 0.234,
+      "step": 655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.954888252148997e-05,
+      "loss": 0.1844,
+      "step": 656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.954819484240688e-05,
+      "loss": 0.1841,
+      "step": 657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.954750716332378e-05,
+      "loss": 0.1337,
+      "step": 658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.954681948424069e-05,
+      "loss": 0.1516,
+      "step": 659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.954613180515759e-05,
+      "loss": 0.1199,
+      "step": 660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.95454441260745e-05,
+      "loss": 0.165,
+      "step": 661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.95447564469914e-05,
+      "loss": 0.1689,
+      "step": 662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.954406876790831e-05,
+      "loss": 0.188,
+      "step": 663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.954338108882521e-05,
+      "loss": 0.1338,
+      "step": 664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.954269340974212e-05,
+      "loss": 0.1795,
+      "step": 665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.954200573065902e-05,
+      "loss": 0.1061,
+      "step": 666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.954131805157593e-05,
+      "loss": 0.1759,
+      "step": 667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.9540630372492834e-05,
+      "loss": 0.1478,
+      "step": 668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.953994269340974e-05,
+      "loss": 0.1337,
+      "step": 669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.9539255014326645e-05,
+      "loss": 0.1261,
+      "step": 670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 5.9538567335243554e-05,
+      "loss": 0.114,
+      "step": 671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.9537879656160456e-05,
+      "loss": 0.1972,
+      "step": 672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.9537191977077365e-05,
+      "loss": 0.1756,
+      "step": 673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.953650429799427e-05,
+      "loss": 0.1425,
+      "step": 674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.9535816618911176e-05,
+      "loss": 0.1027,
+      "step": 675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.953512893982808e-05,
+      "loss": 0.1608,
+      "step": 676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.953444126074499e-05,
+      "loss": 0.1344,
+      "step": 677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.953375358166189e-05,
+      "loss": 0.114,
+      "step": 678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.95330659025788e-05,
+      "loss": 0.1356,
+      "step": 679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.95323782234957e-05,
+      "loss": 0.1944,
+      "step": 680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.953169054441261e-05,
+      "loss": 0.121,
+      "step": 681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.953100286532951e-05,
+      "loss": 0.2994,
+      "step": 682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.953031518624642e-05,
+      "loss": 0.1491,
+      "step": 683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.952962750716332e-05,
+      "loss": 0.165,
+      "step": 684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.952893982808023e-05,
+      "loss": 0.1249,
+      "step": 685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.952825214899713e-05,
+      "loss": 0.102,
+      "step": 686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.952756446991404e-05,
+      "loss": 0.2067,
+      "step": 687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.9526876790830944e-05,
+      "loss": 0.1722,
+      "step": 688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 5.952618911174785e-05,
+      "loss": 0.2066,
+      "step": 689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9525501432664755e-05,
+      "loss": 0.1305,
+      "step": 690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9524813753581664e-05,
+      "loss": 0.1237,
+      "step": 691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9524126074498566e-05,
+      "loss": 0.1407,
+      "step": 692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9523438395415475e-05,
+      "loss": 0.12,
+      "step": 693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.952275071633238e-05,
+      "loss": 0.1719,
+      "step": 694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.9522063037249286e-05,
+      "loss": 0.0771,
+      "step": 695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.952137535816619e-05,
+      "loss": 0.0854,
+      "step": 696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.95206876790831e-05,
+      "loss": 0.1404,
+      "step": 697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.952e-05,
+      "loss": 0.11,
+      "step": 698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.951931232091691e-05,
+      "loss": 0.1977,
+      "step": 699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.951862464183381e-05,
+      "loss": 0.1206,
+      "step": 700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.951793696275072e-05,
+      "loss": 0.1362,
+      "step": 701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.951724928366762e-05,
+      "loss": 0.13,
+      "step": 702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.951656160458453e-05,
+      "loss": 0.1386,
+      "step": 703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.951587392550143e-05,
+      "loss": 0.1154,
+      "step": 704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.951518624641834e-05,
+      "loss": 0.1819,
+      "step": 705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 5.951449856733524e-05,
+      "loss": 0.0933,
+      "step": 706
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.951381088825215e-05,
+      "loss": 0.182,
+      "step": 707
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9513123209169054e-05,
+      "loss": 0.105,
+      "step": 708
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.951243553008596e-05,
+      "loss": 0.1205,
+      "step": 709
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9511747851002865e-05,
+      "loss": 0.1397,
+      "step": 710
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9511060171919774e-05,
+      "loss": 0.1234,
+      "step": 711
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9510372492836676e-05,
+      "loss": 0.1178,
+      "step": 712
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9509684813753585e-05,
+      "loss": 0.1189,
+      "step": 713
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.950899713467049e-05,
+      "loss": 0.1221,
+      "step": 714
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.950830945558739e-05,
+      "loss": 0.114,
+      "step": 715
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.95076217765043e-05,
+      "loss": 0.0896,
+      "step": 716
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.95069340974212e-05,
+      "loss": 0.1394,
+      "step": 717
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.950624641833811e-05,
+      "loss": 0.102,
+      "step": 718
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.950555873925501e-05,
+      "loss": 0.1006,
+      "step": 719
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.950487106017192e-05,
+      "loss": 0.098,
+      "step": 720
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.950418338108882e-05,
+      "loss": 0.1338,
+      "step": 721
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.950349570200573e-05,
+      "loss": 0.1109,
+      "step": 722
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.9502808022922633e-05,
+      "loss": 0.1388,
+      "step": 723
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 5.950212034383954e-05,
+      "loss": 0.1275,
+      "step": 724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.9501432664756444e-05,
+      "loss": 0.1082,
+      "step": 725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.950074498567335e-05,
+      "loss": 0.1555,
+      "step": 726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.9500057306590255e-05,
+      "loss": 0.1711,
+      "step": 727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.9499369627507164e-05,
+      "loss": 0.0959,
+      "step": 728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.9498681948424066e-05,
+      "loss": 0.2358,
+      "step": 729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.9497994269340975e-05,
+      "loss": 0.128,
+      "step": 730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.949730659025788e-05,
+      "loss": 0.1318,
+      "step": 731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.9496618911174786e-05,
+      "loss": 0.1928,
+      "step": 732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.949593123209169e-05,
+      "loss": 0.117,
+      "step": 733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.94952435530086e-05,
+      "loss": 0.1606,
+      "step": 734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.94945558739255e-05,
+      "loss": 0.1235,
+      "step": 735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.949386819484241e-05,
+      "loss": 0.1016,
+      "step": 736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.949318051575931e-05,
+      "loss": 0.2032,
+      "step": 737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.949249283667622e-05,
+      "loss": 0.1756,
+      "step": 738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.949180515759312e-05,
+      "loss": 0.1141,
+      "step": 739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.949111747851003e-05,
+      "loss": 0.1301,
+      "step": 740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 5.949042979942693e-05,
+      "loss": 0.1059,
+      "step": 741
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.948974212034384e-05,
+      "loss": 0.1668,
+      "step": 742
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9489054441260743e-05,
+      "loss": 0.1361,
+      "step": 743
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.948836676217765e-05,
+      "loss": 0.1275,
+      "step": 744
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9487679083094554e-05,
+      "loss": 0.17,
+      "step": 745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.948699140401146e-05,
+      "loss": 0.1066,
+      "step": 746
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9486303724928365e-05,
+      "loss": 0.1204,
+      "step": 747
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9485616045845274e-05,
+      "loss": 0.106,
+      "step": 748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9484928366762176e-05,
+      "loss": 0.1579,
+      "step": 749
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9484240687679085e-05,
+      "loss": 0.1288,
+      "step": 750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.948355300859599e-05,
+      "loss": 0.1335,
+      "step": 751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.9482865329512896e-05,
+      "loss": 0.1494,
+      "step": 752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.94821776504298e-05,
+      "loss": 0.1054,
+      "step": 753
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.948148997134671e-05,
+      "loss": 0.209,
+      "step": 754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.948080229226361e-05,
+      "loss": 0.1547,
+      "step": 755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.948011461318052e-05,
+      "loss": 0.138,
+      "step": 756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.947942693409742e-05,
+      "loss": 0.2068,
+      "step": 757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.947873925501433e-05,
+      "loss": 0.1512,
+      "step": 758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 5.947805157593123e-05,
+      "loss": 0.1098,
+      "step": 759
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.947736389684814e-05,
+      "loss": 0.0815,
+      "step": 760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.947667621776504e-05,
+      "loss": 0.155,
+      "step": 761
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.947598853868195e-05,
+      "loss": 0.1173,
+      "step": 762
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9475300859598853e-05,
+      "loss": 0.2412,
+      "step": 763
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.947461318051576e-05,
+      "loss": 0.1232,
+      "step": 764
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9473925501432664e-05,
+      "loss": 0.116,
+      "step": 765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.947323782234957e-05,
+      "loss": 0.116,
+      "step": 766
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9472550143266475e-05,
+      "loss": 0.1199,
+      "step": 767
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9471862464183384e-05,
+      "loss": 0.1952,
+      "step": 768
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9471174785100286e-05,
+      "loss": 0.1306,
+      "step": 769
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9470487106017195e-05,
+      "loss": 0.105,
+      "step": 770
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.94697994269341e-05,
+      "loss": 0.1604,
+      "step": 771
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9469111747851006e-05,
+      "loss": 0.0783,
+      "step": 772
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.946842406876791e-05,
+      "loss": 0.1417,
+      "step": 773
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.946773638968482e-05,
+      "loss": 0.1231,
+      "step": 774
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.946704871060172e-05,
+      "loss": 0.1585,
+      "step": 775
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.946636103151863e-05,
+      "loss": 0.0774,
+      "step": 776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.946567335243553e-05,
+      "loss": 0.2028,
+      "step": 777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.946498567335244e-05,
+      "loss": 0.1401,
+      "step": 778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.946429799426934e-05,
+      "loss": 0.1345,
+      "step": 779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.946361031518625e-05,
+      "loss": 0.0996,
+      "step": 780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.946292263610315e-05,
+      "loss": 0.1216,
+      "step": 781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.946223495702006e-05,
+      "loss": 0.1876,
+      "step": 782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.946154727793696e-05,
+      "loss": 0.1257,
+      "step": 783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.946085959885387e-05,
+      "loss": 0.1645,
+      "step": 784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9460171919770774e-05,
+      "loss": 0.121,
+      "step": 785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.945948424068768e-05,
+      "loss": 0.1127,
+      "step": 786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9458796561604585e-05,
+      "loss": 0.1196,
+      "step": 787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9458108882521494e-05,
+      "loss": 0.163,
+      "step": 788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9457421203438396e-05,
+      "loss": 0.0863,
+      "step": 789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9456733524355305e-05,
+      "loss": 0.1319,
+      "step": 790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.945604584527221e-05,
+      "loss": 0.0913,
+      "step": 791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9455358166189116e-05,
+      "loss": 0.1004,
+      "step": 792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.945467048710602e-05,
+      "loss": 0.1075,
+      "step": 793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945398280802293e-05,
+      "loss": 0.14,
+      "step": 794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945329512893983e-05,
+      "loss": 0.0902,
+      "step": 795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945260744985674e-05,
+      "loss": 0.1291,
+      "step": 796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945191977077364e-05,
+      "loss": 0.0879,
+      "step": 797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945123209169055e-05,
+      "loss": 0.1411,
+      "step": 798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945054441260745e-05,
+      "loss": 0.1882,
+      "step": 799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944985673352436e-05,
+      "loss": 0.167,
+      "step": 800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944916905444126e-05,
+      "loss": 0.13,
+      "step": 801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944848137535817e-05,
+      "loss": 0.1493,
+      "step": 802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944779369627507e-05,
+      "loss": 0.1972,
+      "step": 803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944710601719198e-05,
+      "loss": 0.1903,
+      "step": 804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9446418338108884e-05,
+      "loss": 0.1085,
+      "step": 805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944573065902579e-05,
+      "loss": 0.1267,
+      "step": 806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9445042979942695e-05,
+      "loss": 0.126,
+      "step": 807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9444355300859604e-05,
+      "loss": 0.1236,
+      "step": 808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9443667621776506e-05,
+      "loss": 0.1114,
+      "step": 809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9442979942693415e-05,
+      "loss": 0.1774,
+      "step": 810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944229226361032e-05,
+      "loss": 0.1543,
+      "step": 811
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9441604584527226e-05,
+      "loss": 0.1243,
+      "step": 812
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.944091690544413e-05,
+      "loss": 0.1069,
+      "step": 813
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.944022922636104e-05,
+      "loss": 0.1199,
+      "step": 814
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943954154727794e-05,
+      "loss": 0.1009,
+      "step": 815
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943885386819485e-05,
+      "loss": 0.1172,
+      "step": 816
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943816618911175e-05,
+      "loss": 0.1705,
+      "step": 817
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943747851002866e-05,
+      "loss": 0.0924,
+      "step": 818
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943679083094556e-05,
+      "loss": 0.119,
+      "step": 819
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943610315186247e-05,
+      "loss": 0.1745,
+      "step": 820
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943541547277937e-05,
+      "loss": 0.0743,
+      "step": 821
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943472779369628e-05,
+      "loss": 0.121,
+      "step": 822
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943404011461318e-05,
+      "loss": 0.1519,
+      "step": 823
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.943335243553009e-05,
+      "loss": 0.1214,
+      "step": 824
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9432664756446994e-05,
+      "loss": 0.1356,
+      "step": 825
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.94319770773639e-05,
+      "loss": 0.0836,
+      "step": 826
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9431289398280805e-05,
+      "loss": 0.1488,
+      "step": 827
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.9430601719197714e-05,
+      "loss": 0.0884,
+      "step": 828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9429914040114616e-05,
+      "loss": 0.1531,
+      "step": 829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9429226361031525e-05,
+      "loss": 0.1455,
+      "step": 830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.942853868194843e-05,
+      "loss": 0.1064,
+      "step": 831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9427851002865336e-05,
+      "loss": 0.1709,
+      "step": 832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.942716332378224e-05,
+      "loss": 0.1323,
+      "step": 833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.942647564469914e-05,
+      "loss": 0.1104,
+      "step": 834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.942578796561605e-05,
+      "loss": 0.1705,
+      "step": 835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.942510028653295e-05,
+      "loss": 0.2293,
+      "step": 836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.942441260744986e-05,
+      "loss": 0.1557,
+      "step": 837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.942372492836676e-05,
+      "loss": 0.0975,
+      "step": 838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.942303724928367e-05,
+      "loss": 0.1444,
+      "step": 839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9422349570200573e-05,
+      "loss": 0.1191,
+      "step": 840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.942166189111748e-05,
+      "loss": 0.2165,
+      "step": 841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9420974212034384e-05,
+      "loss": 0.1872,
+      "step": 842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.942028653295129e-05,
+      "loss": 0.1242,
+      "step": 843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9419598853868195e-05,
+      "loss": 0.1508,
+      "step": 844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9418911174785104e-05,
+      "loss": 0.1081,
+      "step": 845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.9418223495702006e-05,
+      "loss": 0.1122,
+      "step": 846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.9417535816618915e-05,
+      "loss": 0.1283,
+      "step": 847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.941684813753582e-05,
+      "loss": 0.1187,
+      "step": 848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.9416160458452726e-05,
+      "loss": 0.1069,
+      "step": 849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.941547277936963e-05,
+      "loss": 0.121,
+      "step": 850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.941478510028654e-05,
+      "loss": 0.1883,
+      "step": 851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.941409742120344e-05,
+      "loss": 0.1149,
+      "step": 852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.941340974212035e-05,
+      "loss": 0.1136,
+      "step": 853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.941272206303725e-05,
+      "loss": 0.1519,
+      "step": 854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.941203438395416e-05,
+      "loss": 0.0987,
+      "step": 855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.941134670487106e-05,
+      "loss": 0.1123,
+      "step": 856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.941065902578797e-05,
+      "loss": 0.1603,
+      "step": 857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.940997134670487e-05,
+      "loss": 0.1334,
+      "step": 858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.940928366762178e-05,
+      "loss": 0.1052,
+      "step": 859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.9408595988538683e-05,
+      "loss": 0.1674,
+      "step": 860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.940790830945559e-05,
+      "loss": 0.1139,
+      "step": 861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.9407220630372494e-05,
+      "loss": 0.25,
+      "step": 862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.94065329512894e-05,
+      "loss": 0.1223,
+      "step": 863
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9405845272206305e-05,
+      "loss": 0.1166,
+      "step": 864
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9405157593123214e-05,
+      "loss": 0.096,
+      "step": 865
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9404469914040116e-05,
+      "loss": 0.0819,
+      "step": 866
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9403782234957025e-05,
+      "loss": 0.1171,
+      "step": 867
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.940309455587393e-05,
+      "loss": 0.11,
+      "step": 868
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9402406876790836e-05,
+      "loss": 0.0727,
+      "step": 869
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.940171919770774e-05,
+      "loss": 0.0946,
+      "step": 870
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.940103151862465e-05,
+      "loss": 0.1406,
+      "step": 871
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.940034383954155e-05,
+      "loss": 0.0886,
+      "step": 872
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.939965616045846e-05,
+      "loss": 0.0789,
+      "step": 873
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.939896848137536e-05,
+      "loss": 0.1049,
+      "step": 874
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.939828080229227e-05,
+      "loss": 0.1646,
+      "step": 875
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.939759312320917e-05,
+      "loss": 0.225,
+      "step": 876
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.939690544412608e-05,
+      "loss": 0.1032,
+      "step": 877
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.939621776504298e-05,
+      "loss": 0.2014,
+      "step": 878
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.939553008595989e-05,
+      "loss": 0.1095,
+      "step": 879
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.9394842406876793e-05,
+      "loss": 0.1654,
+      "step": 880
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.93941547277937e-05,
+      "loss": 0.1058,
+      "step": 881
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9393467048710604e-05,
+      "loss": 0.1554,
+      "step": 882
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.939277936962751e-05,
+      "loss": 0.1025,
+      "step": 883
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9392091690544415e-05,
+      "loss": 0.1425,
+      "step": 884
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9391404011461324e-05,
+      "loss": 0.1194,
+      "step": 885
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9390716332378226e-05,
+      "loss": 0.1296,
+      "step": 886
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9390028653295135e-05,
+      "loss": 0.1452,
+      "step": 887
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.938934097421204e-05,
+      "loss": 0.1289,
+      "step": 888
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9388653295128946e-05,
+      "loss": 0.183,
+      "step": 889
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.938796561604585e-05,
+      "loss": 0.1097,
+      "step": 890
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.938727793696275e-05,
+      "loss": 0.0936,
+      "step": 891
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.938659025787965e-05,
+      "loss": 0.1253,
+      "step": 892
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.938590257879656e-05,
+      "loss": 0.1148,
+      "step": 893
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9385214899713464e-05,
+      "loss": 0.1607,
+      "step": 894
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.938452722063037e-05,
+      "loss": 0.2705,
+      "step": 895
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9383839541547275e-05,
+      "loss": 0.143,
+      "step": 896
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9383151862464184e-05,
+      "loss": 0.1592,
+      "step": 897
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.9382464183381086e-05,
+      "loss": 0.104,
+      "step": 898
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.9381776504297995e-05,
+      "loss": 0.1557,
+      "step": 899
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.93810888252149e-05,
+      "loss": 0.1264,
+      "step": 900
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.9380401146131806e-05,
+      "loss": 0.1235,
+      "step": 901
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937971346704871e-05,
+      "loss": 0.1144,
+      "step": 902
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.9379025787965617e-05,
+      "loss": 0.204,
+      "step": 903
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937833810888252e-05,
+      "loss": 0.1191,
+      "step": 904
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937765042979943e-05,
+      "loss": 0.1211,
+      "step": 905
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937696275071633e-05,
+      "loss": 0.1213,
+      "step": 906
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937627507163324e-05,
+      "loss": 0.1152,
+      "step": 907
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937558739255014e-05,
+      "loss": 0.1143,
+      "step": 908
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937489971346705e-05,
+      "loss": 0.1191,
+      "step": 909
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937421203438395e-05,
+      "loss": 0.1514,
+      "step": 910
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937352435530086e-05,
+      "loss": 0.1058,
+      "step": 911
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937283667621776e-05,
+      "loss": 0.1392,
+      "step": 912
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937214899713467e-05,
+      "loss": 0.0895,
+      "step": 913
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.9371461318051574e-05,
+      "loss": 0.1443,
+      "step": 914
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.937077363896848e-05,
+      "loss": 0.106,
+      "step": 915
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.9370085959885385e-05,
+      "loss": 0.0911,
+      "step": 916
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9369398280802294e-05,
+      "loss": 0.1222,
+      "step": 917
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9368710601719196e-05,
+      "loss": 0.1722,
+      "step": 918
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9368022922636105e-05,
+      "loss": 0.216,
+      "step": 919
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.936733524355301e-05,
+      "loss": 0.0909,
+      "step": 920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9366647564469916e-05,
+      "loss": 0.0688,
+      "step": 921
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.936595988538682e-05,
+      "loss": 0.0994,
+      "step": 922
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9365272206303727e-05,
+      "loss": 0.1198,
+      "step": 923
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.936458452722063e-05,
+      "loss": 0.2022,
+      "step": 924
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.936389684813754e-05,
+      "loss": 0.1492,
+      "step": 925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.936320916905444e-05,
+      "loss": 0.1061,
+      "step": 926
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.936252148997135e-05,
+      "loss": 0.1799,
+      "step": 927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.936183381088825e-05,
+      "loss": 0.0925,
+      "step": 928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.936114613180516e-05,
+      "loss": 0.1473,
+      "step": 929
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.936045845272206e-05,
+      "loss": 0.1133,
+      "step": 930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.935977077363897e-05,
+      "loss": 0.0927,
+      "step": 931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.935908309455587e-05,
+      "loss": 0.1451,
+      "step": 932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.935839541547278e-05,
+      "loss": 0.1695,
+      "step": 933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9357707736389684e-05,
+      "loss": 0.1023,
+      "step": 934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.935702005730659e-05,
+      "loss": 0.1411,
+      "step": 935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9356332378223495e-05,
+      "loss": 0.1132,
+      "step": 936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9355644699140403e-05,
+      "loss": 0.1016,
+      "step": 937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9354957020057306e-05,
+      "loss": 0.1073,
+      "step": 938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9354269340974214e-05,
+      "loss": 0.1458,
+      "step": 939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9353581661891117e-05,
+      "loss": 0.1755,
+      "step": 940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9352893982808025e-05,
+      "loss": 0.0717,
+      "step": 941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.935220630372493e-05,
+      "loss": 0.1327,
+      "step": 942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9351518624641836e-05,
+      "loss": 0.1338,
+      "step": 943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.935083094555874e-05,
+      "loss": 0.1382,
+      "step": 944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.935014326647565e-05,
+      "loss": 0.0859,
+      "step": 945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.934945558739255e-05,
+      "loss": 0.1491,
+      "step": 946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.934876790830946e-05,
+      "loss": 0.1364,
+      "step": 947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.934808022922636e-05,
+      "loss": 0.1135,
+      "step": 948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.934739255014327e-05,
+      "loss": 0.1102,
+      "step": 949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.934670487106017e-05,
+      "loss": 0.0917,
+      "step": 950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9346017191977074e-05,
+      "loss": 0.1483,
+      "step": 951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.934532951289398e-05,
+      "loss": 0.1121,
+      "step": 952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.9344641833810885e-05,
+      "loss": 0.116,
+      "step": 953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.9343954154727794e-05,
+      "loss": 0.1756,
+      "step": 954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.9343266475644696e-05,
+      "loss": 0.1303,
+      "step": 955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.9342578796561605e-05,
+      "loss": 0.1471,
+      "step": 956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.934189111747851e-05,
+      "loss": 0.165,
+      "step": 957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.9341203438395416e-05,
+      "loss": 0.1279,
+      "step": 958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.934051575931232e-05,
+      "loss": 0.1084,
+      "step": 959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.9339828080229227e-05,
+      "loss": 0.115,
+      "step": 960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.933914040114613e-05,
+      "loss": 0.1436,
+      "step": 961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.933845272206304e-05,
+      "loss": 0.1516,
+      "step": 962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.933776504297994e-05,
+      "loss": 0.1282,
+      "step": 963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.933707736389685e-05,
+      "loss": 0.1458,
+      "step": 964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.933638968481375e-05,
+      "loss": 0.1402,
+      "step": 965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.933570200573066e-05,
+      "loss": 0.1417,
+      "step": 966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.933501432664756e-05,
+      "loss": 0.1125,
+      "step": 967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 5.933432664756447e-05,
+      "loss": 0.2129,
+      "step": 968
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.933363896848137e-05,
+      "loss": 0.0835,
+      "step": 969
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.933295128939828e-05,
+      "loss": 0.1902,
+      "step": 970
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.9332263610315184e-05,
+      "loss": 0.1253,
+      "step": 971
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.933157593123209e-05,
+      "loss": 0.1286,
+      "step": 972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.9330888252148995e-05,
+      "loss": 0.1398,
+      "step": 973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.9330200573065904e-05,
+      "loss": 0.1027,
+      "step": 974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.9329512893982806e-05,
+      "loss": 0.0946,
+      "step": 975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.9328825214899715e-05,
+      "loss": 0.1024,
+      "step": 976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.932813753581662e-05,
+      "loss": 0.1762,
+      "step": 977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.9327449856733526e-05,
+      "loss": 0.1645,
+      "step": 978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.932676217765043e-05,
+      "loss": 0.1142,
+      "step": 979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.9326074498567337e-05,
+      "loss": 0.1654,
+      "step": 980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.932538681948424e-05,
+      "loss": 0.1386,
+      "step": 981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.932469914040115e-05,
+      "loss": 0.1857,
+      "step": 982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.932401146131805e-05,
+      "loss": 0.0985,
+      "step": 983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.932332378223496e-05,
+      "loss": 0.1557,
+      "step": 984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 5.932263610315186e-05,
+      "loss": 0.1306,
+      "step": 985
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.932194842406877e-05,
+      "loss": 0.1095,
+      "step": 986
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.932126074498567e-05,
+      "loss": 0.0639,
+      "step": 987
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.932057306590258e-05,
+      "loss": 0.1103,
+      "step": 988
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.931988538681948e-05,
+      "loss": 0.0936,
+      "step": 989
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.931919770773639e-05,
+      "loss": 0.0974,
+      "step": 990
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9318510028653294e-05,
+      "loss": 0.1196,
+      "step": 991
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.93178223495702e-05,
+      "loss": 0.1163,
+      "step": 992
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9317134670487105e-05,
+      "loss": 0.1406,
+      "step": 993
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9316446991404014e-05,
+      "loss": 0.1869,
+      "step": 994
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9315759312320916e-05,
+      "loss": 0.1086,
+      "step": 995
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9315071633237825e-05,
+      "loss": 0.0771,
+      "step": 996
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.931438395415473e-05,
+      "loss": 0.1014,
+      "step": 997
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9313696275071636e-05,
+      "loss": 0.1254,
+      "step": 998
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.931300859598854e-05,
+      "loss": 0.1525,
+      "step": 999
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.9312320916905447e-05,
+      "loss": 0.0712,
+      "step": 1000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.931163323782235e-05,
+      "loss": 0.1259,
+      "step": 1001
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.931094555873926e-05,
+      "loss": 0.0771,
+      "step": 1002
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 5.931025787965616e-05,
+      "loss": 0.1191,
+      "step": 1003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.930957020057307e-05,
+      "loss": 0.1204,
+      "step": 1004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.930888252148997e-05,
+      "loss": 0.129,
+      "step": 1005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.930819484240688e-05,
+      "loss": 0.1139,
+      "step": 1006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.930750716332378e-05,
+      "loss": 0.1306,
+      "step": 1007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.930681948424069e-05,
+      "loss": 0.1557,
+      "step": 1008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.930613180515759e-05,
+      "loss": 0.1303,
+      "step": 1009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.93054441260745e-05,
+      "loss": 0.0918,
+      "step": 1010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9304756446991404e-05,
+      "loss": 0.1379,
+      "step": 1011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.930406876790831e-05,
+      "loss": 0.0958,
+      "step": 1012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9303381088825215e-05,
+      "loss": 0.0915,
+      "step": 1013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9302693409742124e-05,
+      "loss": 0.0684,
+      "step": 1014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9302005730659026e-05,
+      "loss": 0.0839,
+      "step": 1015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9301318051575935e-05,
+      "loss": 0.1053,
+      "step": 1016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.930063037249284e-05,
+      "loss": 0.0623,
+      "step": 1017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9299942693409746e-05,
+      "loss": 0.1323,
+      "step": 1018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.929925501432665e-05,
+      "loss": 0.1575,
+      "step": 1019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 5.9298567335243557e-05,
+      "loss": 0.0945,
+      "step": 1020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.929787965616046e-05,
+      "loss": 0.087,
+      "step": 1021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.929719197707737e-05,
+      "loss": 0.18,
+      "step": 1022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.929650429799427e-05,
+      "loss": 0.1105,
+      "step": 1023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.929581661891118e-05,
+      "loss": 0.1165,
+      "step": 1024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.929512893982808e-05,
+      "loss": 0.1707,
+      "step": 1025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.929444126074499e-05,
+      "loss": 0.1138,
+      "step": 1026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.929375358166189e-05,
+      "loss": 0.1372,
+      "step": 1027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.92930659025788e-05,
+      "loss": 0.1519,
+      "step": 1028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.92923782234957e-05,
+      "loss": 0.1028,
+      "step": 1029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.929169054441261e-05,
+      "loss": 0.2215,
+      "step": 1030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.9291002865329514e-05,
+      "loss": 0.1688,
+      "step": 1031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.929031518624642e-05,
+      "loss": 0.1429,
+      "step": 1032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.9289627507163325e-05,
+      "loss": 0.1264,
+      "step": 1033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.9288939828080234e-05,
+      "loss": 0.1187,
+      "step": 1034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.9288252148997136e-05,
+      "loss": 0.1523,
+      "step": 1035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.9287564469914045e-05,
+      "loss": 0.126,
+      "step": 1036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.928687679083095e-05,
+      "loss": 0.0904,
+      "step": 1037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 5.9286189111747856e-05,
+      "loss": 0.1211,
+      "step": 1038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.928550143266476e-05,
+      "loss": 0.0934,
+      "step": 1039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9284813753581667e-05,
+      "loss": 0.063,
+      "step": 1040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.928412607449857e-05,
+      "loss": 0.1271,
+      "step": 1041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.928343839541548e-05,
+      "loss": 0.0923,
+      "step": 1042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.928275071633238e-05,
+      "loss": 0.1076,
+      "step": 1043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.928206303724929e-05,
+      "loss": 0.1193,
+      "step": 1044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.928137535816619e-05,
+      "loss": 0.0885,
+      "step": 1045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.92806876790831e-05,
+      "loss": 0.0585,
+      "step": 1046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.928e-05,
+      "loss": 0.1171,
+      "step": 1047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.927931232091691e-05,
+      "loss": 0.0909,
+      "step": 1048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.927862464183381e-05,
+      "loss": 0.1161,
+      "step": 1049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.927793696275072e-05,
+      "loss": 0.1787,
+      "step": 1050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9277249283667624e-05,
+      "loss": 0.104,
+      "step": 1051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.927656160458453e-05,
+      "loss": 0.099,
+      "step": 1052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9275873925501435e-05,
+      "loss": 0.1278,
+      "step": 1053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9275186246418344e-05,
+      "loss": 0.1124,
+      "step": 1054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 5.9274498567335246e-05,
+      "loss": 0.0922,
+      "step": 1055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9273810888252155e-05,
+      "loss": 0.1559,
+      "step": 1056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.927312320916906e-05,
+      "loss": 0.1478,
+      "step": 1057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9272435530085966e-05,
+      "loss": 0.1098,
+      "step": 1058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.927174785100287e-05,
+      "loss": 0.1156,
+      "step": 1059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9271060171919777e-05,
+      "loss": 0.1145,
+      "step": 1060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.927037249283668e-05,
+      "loss": 0.1057,
+      "step": 1061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.926968481375359e-05,
+      "loss": 0.1429,
+      "step": 1062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.926899713467049e-05,
+      "loss": 0.1141,
+      "step": 1063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.92683094555874e-05,
+      "loss": 0.1688,
+      "step": 1064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.92676217765043e-05,
+      "loss": 0.1431,
+      "step": 1065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.926693409742121e-05,
+      "loss": 0.0973,
+      "step": 1066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.926624641833811e-05,
+      "loss": 0.128,
+      "step": 1067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.926555873925502e-05,
+      "loss": 0.1326,
+      "step": 1068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.926487106017192e-05,
+      "loss": 0.0826,
+      "step": 1069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9264183381088825e-05,
+      "loss": 0.0952,
+      "step": 1070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9263495702005734e-05,
+      "loss": 0.1157,
+      "step": 1071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9262808022922636e-05,
+      "loss": 0.0848,
+      "step": 1072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 5.9262120343839545e-05,
+      "loss": 0.1099,
+      "step": 1073
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.926143266475645e-05,
+      "loss": 0.1139,
+      "step": 1074
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9260744985673356e-05,
+      "loss": 0.0695,
+      "step": 1075
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.926005730659026e-05,
+      "loss": 0.1214,
+      "step": 1076
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.925936962750717e-05,
+      "loss": 0.0874,
+      "step": 1077
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.925868194842407e-05,
+      "loss": 0.0911,
+      "step": 1078
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.925799426934098e-05,
+      "loss": 0.0884,
+      "step": 1079
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.925730659025788e-05,
+      "loss": 0.1487,
+      "step": 1080
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.925661891117479e-05,
+      "loss": 0.0975,
+      "step": 1081
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.925593123209169e-05,
+      "loss": 0.0959,
+      "step": 1082
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.92552435530086e-05,
+      "loss": 0.1147,
+      "step": 1083
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.92545558739255e-05,
+      "loss": 0.2442,
+      "step": 1084
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.925386819484241e-05,
+      "loss": 0.0963,
+      "step": 1085
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.925318051575931e-05,
+      "loss": 0.2553,
+      "step": 1086
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.925249283667622e-05,
+      "loss": 0.0906,
+      "step": 1087
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9251805157593124e-05,
+      "loss": 0.0975,
+      "step": 1088
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.925111747851003e-05,
+      "loss": 0.1883,
+      "step": 1089
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 5.9250429799426935e-05,
+      "loss": 0.0959,
+      "step": 1090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9249742120343844e-05,
+      "loss": 0.1142,
+      "step": 1091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9249054441260746e-05,
+      "loss": 0.1185,
+      "step": 1092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9248366762177655e-05,
+      "loss": 0.1592,
+      "step": 1093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.924767908309456e-05,
+      "loss": 0.0998,
+      "step": 1094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9246991404011466e-05,
+      "loss": 0.1473,
+      "step": 1095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.924630372492837e-05,
+      "loss": 0.1306,
+      "step": 1096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9245616045845277e-05,
+      "loss": 0.0873,
+      "step": 1097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.924492836676218e-05,
+      "loss": 0.1051,
+      "step": 1098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.924424068767909e-05,
+      "loss": 0.1034,
+      "step": 1099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.924355300859599e-05,
+      "loss": 0.1,
+      "step": 1100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.92428653295129e-05,
+      "loss": 0.1892,
+      "step": 1101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.92421776504298e-05,
+      "loss": 0.1243,
+      "step": 1102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.924148997134671e-05,
+      "loss": 0.1679,
+      "step": 1103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.924080229226361e-05,
+      "loss": 0.1366,
+      "step": 1104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.924011461318052e-05,
+      "loss": 0.1141,
+      "step": 1105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.923942693409742e-05,
+      "loss": 0.092,
+      "step": 1106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.923873925501433e-05,
+      "loss": 0.0837,
+      "step": 1107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 5.9238051575931234e-05,
+      "loss": 0.1091,
+      "step": 1108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923736389684814e-05,
+      "loss": 0.0975,
+      "step": 1109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9236676217765045e-05,
+      "loss": 0.1186,
+      "step": 1110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9235988538681954e-05,
+      "loss": 0.1114,
+      "step": 1111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9235300859598856e-05,
+      "loss": 0.0959,
+      "step": 1112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9234613180515765e-05,
+      "loss": 0.0828,
+      "step": 1113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923392550143267e-05,
+      "loss": 0.1634,
+      "step": 1114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9233237822349576e-05,
+      "loss": 0.0928,
+      "step": 1115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923255014326648e-05,
+      "loss": 0.1153,
+      "step": 1116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9231862464183387e-05,
+      "loss": 0.0586,
+      "step": 1117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923117478510029e-05,
+      "loss": 0.1116,
+      "step": 1118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.92304871060172e-05,
+      "loss": 0.0732,
+      "step": 1119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.92297994269341e-05,
+      "loss": 0.1051,
+      "step": 1120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922911174785101e-05,
+      "loss": 0.1011,
+      "step": 1121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922842406876791e-05,
+      "loss": 0.1516,
+      "step": 1122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922773638968482e-05,
+      "loss": 0.1123,
+      "step": 1123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922704871060172e-05,
+      "loss": 0.1044,
+      "step": 1124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922636103151863e-05,
+      "loss": 0.1043,
+      "step": 1125
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.922567335243553e-05,
+      "loss": 0.0852,
+      "step": 1126
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.922498567335244e-05,
+      "loss": 0.0708,
+      "step": 1127
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9224297994269344e-05,
+      "loss": 0.1084,
+      "step": 1128
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.922361031518625e-05,
+      "loss": 0.0716,
+      "step": 1129
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9222922636103155e-05,
+      "loss": 0.181,
+      "step": 1130
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9222234957020064e-05,
+      "loss": 0.0673,
+      "step": 1131
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9221547277936966e-05,
+      "loss": 0.1645,
+      "step": 1132
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9220859598853875e-05,
+      "loss": 0.1176,
+      "step": 1133
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.922017191977078e-05,
+      "loss": 0.158,
+      "step": 1134
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9219484240687686e-05,
+      "loss": 0.1321,
+      "step": 1135
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.921879656160459e-05,
+      "loss": 0.1094,
+      "step": 1136
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9218108882521497e-05,
+      "loss": 0.0864,
+      "step": 1137
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.92174212034384e-05,
+      "loss": 0.0956,
+      "step": 1138
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.921673352435531e-05,
+      "loss": 0.0877,
+      "step": 1139
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.921604584527221e-05,
+      "loss": 0.0935,
+      "step": 1140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.921535816618912e-05,
+      "loss": 0.128,
+      "step": 1141
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.921467048710602e-05,
+      "loss": 0.0928,
+      "step": 1142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.921398280802293e-05,
+      "loss": 0.0877,
+      "step": 1143
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.921329512893983e-05,
+      "loss": 0.0878,
+      "step": 1144
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9212607449856734e-05,
+      "loss": 0.1692,
+      "step": 1145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9211919770773636e-05,
+      "loss": 0.1802,
+      "step": 1146
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9211232091690545e-05,
+      "loss": 0.241,
+      "step": 1147
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.921054441260745e-05,
+      "loss": 0.143,
+      "step": 1148
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.9209856733524356e-05,
+      "loss": 0.1253,
+      "step": 1149
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.920916905444126e-05,
+      "loss": 0.0982,
+      "step": 1150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.920848137535817e-05,
+      "loss": 0.0774,
+      "step": 1151
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.920779369627507e-05,
+      "loss": 0.1442,
+      "step": 1152
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.920710601719198e-05,
+      "loss": 0.1013,
+      "step": 1153
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.920641833810888e-05,
+      "loss": 0.1787,
+      "step": 1154
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.920573065902579e-05,
+      "loss": 0.0731,
+      "step": 1155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.920504297994269e-05,
+      "loss": 0.0852,
+      "step": 1156
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.92043553008596e-05,
+      "loss": 0.1454,
+      "step": 1157
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.92036676217765e-05,
+      "loss": 0.1529,
+      "step": 1158
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.920297994269341e-05,
+      "loss": 0.0948,
+      "step": 1159
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.920229226361031e-05,
+      "loss": 0.0769,
+      "step": 1160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.920160458452722e-05,
+      "loss": 0.0933,
+      "step": 1161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9200916905444124e-05,
+      "loss": 0.1037,
+      "step": 1162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.920022922636103e-05,
+      "loss": 0.112,
+      "step": 1163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9199541547277935e-05,
+      "loss": 0.2109,
+      "step": 1164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9198853868194844e-05,
+      "loss": 0.1071,
+      "step": 1165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9198166189111746e-05,
+      "loss": 0.1659,
+      "step": 1166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9197478510028655e-05,
+      "loss": 0.1328,
+      "step": 1167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.919679083094556e-05,
+      "loss": 0.1224,
+      "step": 1168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.9196103151862466e-05,
+      "loss": 0.1402,
+      "step": 1169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.919541547277937e-05,
+      "loss": 0.1206,
+      "step": 1170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.919472779369628e-05,
+      "loss": 0.071,
+      "step": 1171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.919404011461318e-05,
+      "loss": 0.1399,
+      "step": 1172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.919335243553009e-05,
+      "loss": 0.1454,
+      "step": 1173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.919266475644699e-05,
+      "loss": 0.0985,
+      "step": 1174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.91919770773639e-05,
+      "loss": 0.0677,
+      "step": 1175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.91912893982808e-05,
+      "loss": 0.0847,
+      "step": 1176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.919060171919771e-05,
+      "loss": 0.0902,
+      "step": 1177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.918991404011461e-05,
+      "loss": 0.066,
+      "step": 1178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.918922636103152e-05,
+      "loss": 0.1247,
+      "step": 1179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.918853868194842e-05,
+      "loss": 0.0895,
+      "step": 1180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.918785100286533e-05,
+      "loss": 0.0778,
+      "step": 1181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.9187163323782234e-05,
+      "loss": 0.1682,
+      "step": 1182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.918647564469914e-05,
+      "loss": 0.0998,
+      "step": 1183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.9185787965616045e-05,
+      "loss": 0.1334,
+      "step": 1184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.9185100286532954e-05,
+      "loss": 0.0752,
+      "step": 1185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.9184412607449856e-05,
+      "loss": 0.1268,
+      "step": 1186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.9183724928366765e-05,
+      "loss": 0.1359,
+      "step": 1187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.918303724928367e-05,
+      "loss": 0.1526,
+      "step": 1188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.918234957020057e-05,
+      "loss": 0.0896,
+      "step": 1189
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.918166189111748e-05,
+      "loss": 0.1356,
+      "step": 1190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.918097421203438e-05,
+      "loss": 0.171,
+      "step": 1191
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.918028653295129e-05,
+      "loss": 0.0988,
+      "step": 1192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.917959885386819e-05,
+      "loss": 0.0864,
+      "step": 1193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.91789111747851e-05,
+      "loss": 0.1129,
+      "step": 1194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.9178223495702e-05,
+      "loss": 0.1017,
+      "step": 1195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.917753581661891e-05,
+      "loss": 0.0669,
+      "step": 1196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.917684813753581e-05,
+      "loss": 0.0912,
+      "step": 1197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.917616045845272e-05,
+      "loss": 0.0662,
+      "step": 1198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9175472779369624e-05,
+      "loss": 0.1101,
+      "step": 1199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.917478510028653e-05,
+      "loss": 0.1111,
+      "step": 1200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9174097421203435e-05,
+      "loss": 0.0752,
+      "step": 1201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9173409742120344e-05,
+      "loss": 0.1144,
+      "step": 1202
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9172722063037246e-05,
+      "loss": 0.0886,
+      "step": 1203
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9172034383954155e-05,
+      "loss": 0.1092,
+      "step": 1204
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.917134670487106e-05,
+      "loss": 0.1543,
+      "step": 1205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.9170659025787966e-05,
+      "loss": 0.173,
+      "step": 1206
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.916997134670487e-05,
+      "loss": 0.0713,
+      "step": 1207
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.916928366762178e-05,
+      "loss": 0.2334,
+      "step": 1208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.916859598853868e-05,
+      "loss": 0.1271,
+      "step": 1209
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.916790830945559e-05,
+      "loss": 0.1146,
+      "step": 1210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.916722063037249e-05,
+      "loss": 0.142,
+      "step": 1211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 5.91665329512894e-05,
+      "loss": 0.1644,
+      "step": 1212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.91658452722063e-05,
+      "loss": 0.1372,
+      "step": 1213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.916515759312321e-05,
+      "loss": 0.1641,
+      "step": 1214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.916446991404011e-05,
+      "loss": 0.078,
+      "step": 1215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.916378223495702e-05,
+      "loss": 0.07,
+      "step": 1216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.916309455587392e-05,
+      "loss": 0.0858,
+      "step": 1217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.916240687679083e-05,
+      "loss": 0.1314,
+      "step": 1218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9161719197707734e-05,
+      "loss": 0.0937,
+      "step": 1219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.916103151862464e-05,
+      "loss": 0.1231,
+      "step": 1220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9160343839541545e-05,
+      "loss": 0.1669,
+      "step": 1221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9159656160458454e-05,
+      "loss": 0.1122,
+      "step": 1222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9158968481375356e-05,
+      "loss": 0.0971,
+      "step": 1223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9158280802292265e-05,
+      "loss": 0.1198,
+      "step": 1224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.915759312320917e-05,
+      "loss": 0.1112,
+      "step": 1225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.9156905444126076e-05,
+      "loss": 0.1049,
+      "step": 1226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.915621776504298e-05,
+      "loss": 0.0849,
+      "step": 1227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.915553008595989e-05,
+      "loss": 0.1028,
+      "step": 1228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.915484240687679e-05,
+      "loss": 0.1395,
+      "step": 1229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 5.91541547277937e-05,
+      "loss": 0.1688,
+      "step": 1230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.91534670487106e-05,
+      "loss": 0.1149,
+      "step": 1231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.915277936962751e-05,
+      "loss": 0.1048,
+      "step": 1232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.915209169054441e-05,
+      "loss": 0.1244,
+      "step": 1233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.915140401146132e-05,
+      "loss": 0.1232,
+      "step": 1234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.915071633237822e-05,
+      "loss": 0.1237,
+      "step": 1235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.915002865329513e-05,
+      "loss": 0.1271,
+      "step": 1236
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.914934097421203e-05,
+      "loss": 0.0822,
+      "step": 1237
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.914865329512894e-05,
+      "loss": 0.0822,
+      "step": 1238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9147965616045844e-05,
+      "loss": 0.1513,
+      "step": 1239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.914727793696275e-05,
+      "loss": 0.1139,
+      "step": 1240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9146590257879655e-05,
+      "loss": 0.1354,
+      "step": 1241
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9145902578796564e-05,
+      "loss": 0.1283,
+      "step": 1242
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9145214899713466e-05,
+      "loss": 0.078,
+      "step": 1243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9144527220630375e-05,
+      "loss": 0.1103,
+      "step": 1244
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.914383954154728e-05,
+      "loss": 0.1063,
+      "step": 1245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.9143151862464186e-05,
+      "loss": 0.1389,
+      "step": 1246
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 5.914246418338109e-05,
+      "loss": 0.1516,
+      "step": 1247
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9141776504298e-05,
+      "loss": 0.0835,
+      "step": 1248
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.91410888252149e-05,
+      "loss": 0.1551,
+      "step": 1249
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.914040114613181e-05,
+      "loss": 0.1061,
+      "step": 1250
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913971346704871e-05,
+      "loss": 0.1218,
+      "step": 1251
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913902578796562e-05,
+      "loss": 0.0991,
+      "step": 1252
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913833810888252e-05,
+      "loss": 0.06,
+      "step": 1253
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913765042979943e-05,
+      "loss": 0.1334,
+      "step": 1254
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913696275071633e-05,
+      "loss": 0.1496,
+      "step": 1255
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913627507163324e-05,
+      "loss": 0.1073,
+      "step": 1256
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913558739255014e-05,
+      "loss": 0.0912,
+      "step": 1257
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913489971346705e-05,
+      "loss": 0.1352,
+      "step": 1258
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9134212034383954e-05,
+      "loss": 0.0776,
+      "step": 1259
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913352435530086e-05,
+      "loss": 0.0871,
+      "step": 1260
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9132836676217765e-05,
+      "loss": 0.1599,
+      "step": 1261
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9132148997134674e-05,
+      "loss": 0.1357,
+      "step": 1262
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9131461318051576e-05,
+      "loss": 0.0875,
+      "step": 1263
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.9130773638968485e-05,
+      "loss": 0.1818,
+      "step": 1264
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 5.913008595988539e-05,
+      "loss": 0.1162,
+      "step": 1265
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.9129398280802296e-05,
+      "loss": 0.0656,
+      "step": 1266
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.91287106017192e-05,
+      "loss": 0.1008,
+      "step": 1267
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912802292263611e-05,
+      "loss": 0.0723,
+      "step": 1268
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912733524355301e-05,
+      "loss": 0.0881,
+      "step": 1269
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912664756446992e-05,
+      "loss": 0.1045,
+      "step": 1270
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912595988538682e-05,
+      "loss": 0.1201,
+      "step": 1271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912527220630373e-05,
+      "loss": 0.1483,
+      "step": 1272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912458452722063e-05,
+      "loss": 0.1054,
+      "step": 1273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912389684813754e-05,
+      "loss": 0.0954,
+      "step": 1274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912320916905444e-05,
+      "loss": 0.0898,
+      "step": 1275
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912252148997135e-05,
+      "loss": 0.1074,
+      "step": 1276
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912183381088825e-05,
+      "loss": 0.1006,
+      "step": 1277
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.912114613180516e-05,
+      "loss": 0.1103,
+      "step": 1278
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.9120458452722064e-05,
+      "loss": 0.0779,
+      "step": 1279
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.911977077363897e-05,
+      "loss": 0.1073,
+      "step": 1280
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.9119083094555875e-05,
+      "loss": 0.1115,
+      "step": 1281
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 5.9118395415472784e-05,
+      "loss": 0.0912,
+      "step": 1282
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.9117707736389686e-05,
+      "loss": 0.0666,
+      "step": 1283
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.9117020057306595e-05,
+      "loss": 0.0471,
+      "step": 1284
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.91163323782235e-05,
+      "loss": 0.1246,
+      "step": 1285
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.9115644699140406e-05,
+      "loss": 0.0799,
+      "step": 1286
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.911495702005731e-05,
+      "loss": 0.1028,
+      "step": 1287
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.911426934097422e-05,
+      "loss": 0.1567,
+      "step": 1288
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.911358166189112e-05,
+      "loss": 0.0956,
+      "step": 1289
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.911289398280803e-05,
+      "loss": 0.0892,
+      "step": 1290
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.911220630372493e-05,
+      "loss": 0.0707,
+      "step": 1291
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.911151862464184e-05,
+      "loss": 0.0913,
+      "step": 1292
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.911083094555874e-05,
+      "loss": 0.0994,
+      "step": 1293
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.911014326647565e-05,
+      "loss": 0.1064,
+      "step": 1294
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.910945558739255e-05,
+      "loss": 0.0828,
+      "step": 1295
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.910876790830946e-05,
+      "loss": 0.1402,
+      "step": 1296
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.910808022922636e-05,
+      "loss": 0.1317,
+      "step": 1297
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.910739255014327e-05,
+      "loss": 0.0931,
+      "step": 1298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.9106704871060174e-05,
+      "loss": 0.1209,
+      "step": 1299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 5.910601719197708e-05,
+      "loss": 0.1241,
+      "step": 1300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9105329512893985e-05,
+      "loss": 0.1619,
+      "step": 1301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9104641833810894e-05,
+      "loss": 0.0847,
+      "step": 1302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9103954154727796e-05,
+      "loss": 0.11,
+      "step": 1303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9103266475644705e-05,
+      "loss": 0.0842,
+      "step": 1304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.910257879656161e-05,
+      "loss": 0.0741,
+      "step": 1305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9101891117478516e-05,
+      "loss": 0.1122,
+      "step": 1306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.910120343839542e-05,
+      "loss": 0.0903,
+      "step": 1307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.910051575931232e-05,
+      "loss": 0.0918,
+      "step": 1308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.909982808022923e-05,
+      "loss": 0.097,
+      "step": 1309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.909914040114613e-05,
+      "loss": 0.1052,
+      "step": 1310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.909845272206304e-05,
+      "loss": 0.0652,
+      "step": 1311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.909776504297994e-05,
+      "loss": 0.089,
+      "step": 1312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.909707736389685e-05,
+      "loss": 0.0769,
+      "step": 1313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.909638968481375e-05,
+      "loss": 0.0629,
+      "step": 1314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.909570200573066e-05,
+      "loss": 0.0858,
+      "step": 1315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.9095014326647564e-05,
+      "loss": 0.0926,
+      "step": 1316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 5.909432664756447e-05,
+      "loss": 0.0975,
+      "step": 1317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9093638968481375e-05,
+      "loss": 0.1327,
+      "step": 1318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9092951289398284e-05,
+      "loss": 0.0811,
+      "step": 1319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9092263610315186e-05,
+      "loss": 0.1097,
+      "step": 1320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9091575931232095e-05,
+      "loss": 0.0805,
+      "step": 1321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9090888252149e-05,
+      "loss": 0.0818,
+      "step": 1322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.9090200573065906e-05,
+      "loss": 0.0893,
+      "step": 1323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908951289398281e-05,
+      "loss": 0.1735,
+      "step": 1324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908882521489972e-05,
+      "loss": 0.0901,
+      "step": 1325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908813753581662e-05,
+      "loss": 0.0813,
+      "step": 1326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908744985673353e-05,
+      "loss": 0.0653,
+      "step": 1327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908676217765043e-05,
+      "loss": 0.124,
+      "step": 1328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908607449856734e-05,
+      "loss": 0.1407,
+      "step": 1329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908538681948424e-05,
+      "loss": 0.0942,
+      "step": 1330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908469914040115e-05,
+      "loss": 0.1135,
+      "step": 1331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908401146131805e-05,
+      "loss": 0.1024,
+      "step": 1332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908332378223496e-05,
+      "loss": 0.1222,
+      "step": 1333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 5.908263610315186e-05,
+      "loss": 0.0672,
+      "step": 1334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.908194842406877e-05,
+      "loss": 0.113,
+      "step": 1335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.9081260744985674e-05,
+      "loss": 0.0955,
+      "step": 1336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.908057306590258e-05,
+      "loss": 0.0771,
+      "step": 1337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.9079885386819485e-05,
+      "loss": 0.1565,
+      "step": 1338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.9079197707736394e-05,
+      "loss": 0.1024,
+      "step": 1339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.9078510028653296e-05,
+      "loss": 0.1005,
+      "step": 1340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.9077822349570205e-05,
+      "loss": 0.0825,
+      "step": 1341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.907713467048711e-05,
+      "loss": 0.0906,
+      "step": 1342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.9076446991404016e-05,
+      "loss": 0.0801,
+      "step": 1343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.907575931232092e-05,
+      "loss": 0.1013,
+      "step": 1344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.907507163323783e-05,
+      "loss": 0.0758,
+      "step": 1345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.907438395415473e-05,
+      "loss": 0.1275,
+      "step": 1346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.907369627507164e-05,
+      "loss": 0.1186,
+      "step": 1347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.907300859598854e-05,
+      "loss": 0.1513,
+      "step": 1348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.907232091690545e-05,
+      "loss": 0.1363,
+      "step": 1349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.907163323782235e-05,
+      "loss": 0.0917,
+      "step": 1350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.907094555873926e-05,
+      "loss": 0.1207,
+      "step": 1351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.907025787965616e-05,
+      "loss": 0.1171,
+      "step": 1352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.906957020057307e-05,
+      "loss": 0.074,
+      "step": 1353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.906888252148997e-05,
+      "loss": 0.1112,
+      "step": 1354
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.906819484240688e-05,
+      "loss": 0.179,
+      "step": 1355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9067507163323784e-05,
+      "loss": 0.0877,
+      "step": 1356
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.906681948424069e-05,
+      "loss": 0.1219,
+      "step": 1357
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9066131805157595e-05,
+      "loss": 0.1248,
+      "step": 1358
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9065444126074504e-05,
+      "loss": 0.0769,
+      "step": 1359
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9064756446991406e-05,
+      "loss": 0.097,
+      "step": 1360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9064068767908315e-05,
+      "loss": 0.103,
+      "step": 1361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.906338108882522e-05,
+      "loss": 0.0857,
+      "step": 1362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9062693409742126e-05,
+      "loss": 0.1097,
+      "step": 1363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.906200573065903e-05,
+      "loss": 0.2848,
+      "step": 1364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.906131805157594e-05,
+      "loss": 0.1135,
+      "step": 1365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.906063037249284e-05,
+      "loss": 0.0968,
+      "step": 1366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.905994269340975e-05,
+      "loss": 0.073,
+      "step": 1367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.905925501432665e-05,
+      "loss": 0.1062,
+      "step": 1368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.905856733524356e-05,
+      "loss": 0.1429,
+      "step": 1369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.905787965616046e-05,
+      "loss": 0.0857,
+      "step": 1370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.905719197707737e-05,
+      "loss": 0.1003,
+      "step": 1371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.905650429799427e-05,
+      "loss": 0.0717,
+      "step": 1372
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.905581661891118e-05,
+      "loss": 0.0867,
+      "step": 1373
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.905512893982808e-05,
+      "loss": 0.0825,
+      "step": 1374
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.905444126074499e-05,
+      "loss": 0.0888,
+      "step": 1375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.9053753581661894e-05,
+      "loss": 0.1158,
+      "step": 1376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.90530659025788e-05,
+      "loss": 0.1146,
+      "step": 1377
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.9052378223495705e-05,
+      "loss": 0.1502,
+      "step": 1378
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.9051690544412614e-05,
+      "loss": 0.0649,
+      "step": 1379
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.9051002865329516e-05,
+      "loss": 0.1224,
+      "step": 1380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.9050315186246425e-05,
+      "loss": 0.1282,
+      "step": 1381
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.904962750716333e-05,
+      "loss": 0.1092,
+      "step": 1382
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.9048939828080236e-05,
+      "loss": 0.0651,
+      "step": 1383
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.904825214899714e-05,
+      "loss": 0.1351,
+      "step": 1384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.904756446991405e-05,
+      "loss": 0.1426,
+      "step": 1385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.904687679083095e-05,
+      "loss": 0.1014,
+      "step": 1386
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.904618911174786e-05,
+      "loss": 0.0935,
+      "step": 1387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.904550143266476e-05,
+      "loss": 0.16,
+      "step": 1388
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.904481375358167e-05,
+      "loss": 0.0889,
+      "step": 1389
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.904412607449857e-05,
+      "loss": 0.0995,
+      "step": 1390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.904343839541548e-05,
+      "loss": 0.0737,
+      "step": 1391
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.904275071633238e-05,
+      "loss": 0.0927,
+      "step": 1392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.904206303724929e-05,
+      "loss": 0.1209,
+      "step": 1393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.904137535816619e-05,
+      "loss": 0.0895,
+      "step": 1394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.90406876790831e-05,
+      "loss": 0.0586,
+      "step": 1395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9040000000000004e-05,
+      "loss": 0.084,
+      "step": 1396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.903931232091691e-05,
+      "loss": 0.1129,
+      "step": 1397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.9038624641833815e-05,
+      "loss": 0.1097,
+      "step": 1398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.903793696275072e-05,
+      "loss": 0.0778,
+      "step": 1399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.903724928366762e-05,
+      "loss": 0.1328,
+      "step": 1400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.903656160458453e-05,
+      "loss": 0.1227,
+      "step": 1401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.903587392550143e-05,
+      "loss": 0.0849,
+      "step": 1402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.903518624641834e-05,
+      "loss": 0.1005,
+      "step": 1403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.903449856733524e-05,
+      "loss": 0.1114,
+      "step": 1404
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.903381088825215e-05,
+      "loss": 0.1115,
+      "step": 1405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.903312320916905e-05,
+      "loss": 0.1122,
+      "step": 1406
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.903243553008596e-05,
+      "loss": 0.1126,
+      "step": 1407
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.903174785100286e-05,
+      "loss": 0.091,
+      "step": 1408
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.903106017191977e-05,
+      "loss": 0.0758,
+      "step": 1409
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.9030372492836674e-05,
+      "loss": 0.1744,
+      "step": 1410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.902968481375358e-05,
+      "loss": 0.1152,
+      "step": 1411
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.9028997134670485e-05,
+      "loss": 0.0706,
+      "step": 1412
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.9028309455587394e-05,
+      "loss": 0.1402,
+      "step": 1413
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.9027621776504296e-05,
+      "loss": 0.0674,
+      "step": 1414
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.9026934097421205e-05,
+      "loss": 0.1201,
+      "step": 1415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.902624641833811e-05,
+      "loss": 0.1125,
+      "step": 1416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.9025558739255016e-05,
+      "loss": 0.1276,
+      "step": 1417
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.902487106017192e-05,
+      "loss": 0.1117,
+      "step": 1418
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.902418338108883e-05,
+      "loss": 0.1283,
+      "step": 1419
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.902349570200573e-05,
+      "loss": 0.0897,
+      "step": 1420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.902280802292264e-05,
+      "loss": 0.1038,
+      "step": 1421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.902212034383954e-05,
+      "loss": 0.1148,
+      "step": 1422
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.902143266475645e-05,
+      "loss": 0.1165,
+      "step": 1423
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.902074498567335e-05,
+      "loss": 0.1009,
+      "step": 1424
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.902005730659026e-05,
+      "loss": 0.1153,
+      "step": 1425
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.901936962750716e-05,
+      "loss": 0.1109,
+      "step": 1426
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.9018681948424064e-05,
+      "loss": 0.0601,
+      "step": 1427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.901799426934097e-05,
+      "loss": 0.0913,
+      "step": 1428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.9017306590257875e-05,
+      "loss": 0.0827,
+      "step": 1429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.9016618911174784e-05,
+      "loss": 0.082,
+      "step": 1430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.9015931232091686e-05,
+      "loss": 0.0662,
+      "step": 1431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.9015243553008595e-05,
+      "loss": 0.0554,
+      "step": 1432
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.90145558739255e-05,
+      "loss": 0.0757,
+      "step": 1433
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.9013868194842406e-05,
+      "loss": 0.1913,
+      "step": 1434
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.901318051575931e-05,
+      "loss": 0.0739,
+      "step": 1435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.901249283667622e-05,
+      "loss": 0.1001,
+      "step": 1436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.901180515759312e-05,
+      "loss": 0.0836,
+      "step": 1437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.901111747851003e-05,
+      "loss": 0.1009,
+      "step": 1438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.901042979942693e-05,
+      "loss": 0.0832,
+      "step": 1439
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.900974212034384e-05,
+      "loss": 0.1067,
+      "step": 1440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.900905444126074e-05,
+      "loss": 0.1191,
+      "step": 1441
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.900836676217765e-05,
+      "loss": 0.0684,
+      "step": 1442
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.900767908309455e-05,
+      "loss": 0.1084,
+      "step": 1443
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.900699140401146e-05,
+      "loss": 0.111,
+      "step": 1444
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.900630372492836e-05,
+      "loss": 0.0622,
+      "step": 1445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.900561604584527e-05,
+      "loss": 0.105,
+      "step": 1446
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.9004928366762174e-05,
+      "loss": 0.1529,
+      "step": 1447
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.900424068767908e-05,
+      "loss": 0.0973,
+      "step": 1448
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.9003553008595985e-05,
+      "loss": 0.108,
+      "step": 1449
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.9002865329512894e-05,
+      "loss": 0.0668,
+      "step": 1450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.9002177650429796e-05,
+      "loss": 0.0779,
+      "step": 1451
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.9001489971346705e-05,
+      "loss": 0.0994,
+      "step": 1452
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.900080229226361e-05,
+      "loss": 0.0739,
+      "step": 1453
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.9000114613180516e-05,
+      "loss": 0.0985,
+      "step": 1454
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.899942693409742e-05,
+      "loss": 0.0841,
+      "step": 1455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.899873925501433e-05,
+      "loss": 0.133,
+      "step": 1456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.899805157593123e-05,
+      "loss": 0.0796,
+      "step": 1457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.899736389684814e-05,
+      "loss": 0.0808,
+      "step": 1458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.899667621776504e-05,
+      "loss": 0.0504,
+      "step": 1459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.899598853868195e-05,
+      "loss": 0.0862,
+      "step": 1460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.899530085959885e-05,
+      "loss": 0.1477,
+      "step": 1461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.899461318051576e-05,
+      "loss": 0.104,
+      "step": 1462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.899392550143266e-05,
+      "loss": 0.1233,
+      "step": 1463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.899323782234957e-05,
+      "loss": 0.1062,
+      "step": 1464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.899255014326647e-05,
+      "loss": 0.094,
+      "step": 1465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.899186246418338e-05,
+      "loss": 0.0999,
+      "step": 1466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.8991174785100284e-05,
+      "loss": 0.1117,
+      "step": 1467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.899048710601719e-05,
+      "loss": 0.1001,
+      "step": 1468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.8989799426934095e-05,
+      "loss": 0.095,
+      "step": 1469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.8989111747851004e-05,
+      "loss": 0.0854,
+      "step": 1470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.8988424068767906e-05,
+      "loss": 0.0573,
+      "step": 1471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.8987736389684815e-05,
+      "loss": 0.1127,
+      "step": 1472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.898704871060172e-05,
+      "loss": 0.0991,
+      "step": 1473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.8986361031518626e-05,
+      "loss": 0.0796,
+      "step": 1474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.898567335243553e-05,
+      "loss": 0.106,
+      "step": 1475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.898498567335244e-05,
+      "loss": 0.08,
+      "step": 1476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.898429799426934e-05,
+      "loss": 0.086,
+      "step": 1477
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.898361031518625e-05,
+      "loss": 0.109,
+      "step": 1478
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.898292263610315e-05,
+      "loss": 0.118,
+      "step": 1479
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.898223495702006e-05,
+      "loss": 0.1484,
+      "step": 1480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.898154727793696e-05,
+      "loss": 0.125,
+      "step": 1481
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.898085959885387e-05,
+      "loss": 0.1856,
+      "step": 1482
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.898017191977077e-05,
+      "loss": 0.0868,
+      "step": 1483
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.897948424068768e-05,
+      "loss": 0.1252,
+      "step": 1484
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.897879656160458e-05,
+      "loss": 0.083,
+      "step": 1485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.897810888252149e-05,
+      "loss": 0.0853,
+      "step": 1486
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8977421203438394e-05,
+      "loss": 0.0557,
+      "step": 1487
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.89767335243553e-05,
+      "loss": 0.0891,
+      "step": 1488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8976045845272205e-05,
+      "loss": 0.0935,
+      "step": 1489
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8975358166189114e-05,
+      "loss": 0.1467,
+      "step": 1490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8974670487106016e-05,
+      "loss": 0.0881,
+      "step": 1491
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.8973982808022925e-05,
+      "loss": 0.1265,
+      "step": 1492
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.897329512893983e-05,
+      "loss": 0.0638,
+      "step": 1493
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.8972607449856736e-05,
+      "loss": 0.0549,
+      "step": 1494
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.897191977077364e-05,
+      "loss": 0.0681,
+      "step": 1495
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.897123209169055e-05,
+      "loss": 0.1308,
+      "step": 1496
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.897054441260745e-05,
+      "loss": 0.0826,
+      "step": 1497
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.896985673352436e-05,
+      "loss": 0.1083,
+      "step": 1498
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.896916905444126e-05,
+      "loss": 0.0937,
+      "step": 1499
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.896848137535817e-05,
+      "loss": 0.0979,
+      "step": 1500
+    },
+    {
+      "epoch": 0.86,
+      "eval_loss": 0.07367682456970215,
+      "eval_mean_accuracy": 0.5818018321040256,
+      "eval_mean_iou": 0.5386925715894331,
+      "eval_overall_accuracy": 0.9762552197774251,
+      "eval_per_category_accuracy": [
+        0.9965245679092865,
+        0.4762083763185019,
+        0.4004900198360951,
+        0.45398436435221884
+      ],
+      "eval_per_category_iou": [
+        0.9775795191794591,
+        0.40313300827802795,
+        0.3741727199354318,
+        0.39988503896481326
+      ],
+      "eval_runtime": 410.083,
+      "eval_samples_per_second": 8.779,
+      "eval_steps_per_second": 1.097,
+      "step": 1500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.896779369627507e-05,
+      "loss": 0.0886,
+      "step": 1501
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.896710601719198e-05,
+      "loss": 0.1137,
+      "step": 1502
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.896641833810888e-05,
+      "loss": 0.0952,
+      "step": 1503
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.896573065902579e-05,
+      "loss": 0.058,
+      "step": 1504
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.896504297994269e-05,
+      "loss": 0.0711,
+      "step": 1505
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.89643553008596e-05,
+      "loss": 0.1084,
+      "step": 1506
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.8963667621776504e-05,
+      "loss": 0.0538,
+      "step": 1507
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.896297994269341e-05,
+      "loss": 0.13,
+      "step": 1508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.8962292263610315e-05,
+      "loss": 0.0978,
+      "step": 1509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.8961604584527224e-05,
+      "loss": 0.0515,
+      "step": 1510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.8960916905444126e-05,
+      "loss": 0.1103,
+      "step": 1511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.8960229226361035e-05,
+      "loss": 0.1517,
+      "step": 1512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.895954154727794e-05,
+      "loss": 0.0795,
+      "step": 1513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.8958853868194846e-05,
+      "loss": 0.0886,
+      "step": 1514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.895816618911175e-05,
+      "loss": 0.1134,
+      "step": 1515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.895747851002866e-05,
+      "loss": 0.07,
+      "step": 1516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.895679083094556e-05,
+      "loss": 0.1834,
+      "step": 1517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.895610315186247e-05,
+      "loss": 0.0889,
+      "step": 1518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.895541547277937e-05,
+      "loss": 0.0794,
+      "step": 1519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.895472779369628e-05,
+      "loss": 0.1179,
+      "step": 1520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.895404011461318e-05,
+      "loss": 0.2487,
+      "step": 1521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.895335243553009e-05,
+      "loss": 0.0568,
+      "step": 1522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.895266475644699e-05,
+      "loss": 0.0838,
+      "step": 1523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.89519770773639e-05,
+      "loss": 0.2585,
+      "step": 1524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.89512893982808e-05,
+      "loss": 0.1141,
+      "step": 1525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 5.895060171919771e-05,
+      "loss": 0.097,
+      "step": 1526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8949914040114614e-05,
+      "loss": 0.074,
+      "step": 1527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894922636103152e-05,
+      "loss": 0.0764,
+      "step": 1528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8948538681948425e-05,
+      "loss": 0.0789,
+      "step": 1529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8947851002865334e-05,
+      "loss": 0.0846,
+      "step": 1530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8947163323782236e-05,
+      "loss": 0.0903,
+      "step": 1531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8946475644699145e-05,
+      "loss": 0.1022,
+      "step": 1532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894578796561605e-05,
+      "loss": 0.1704,
+      "step": 1533
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.8945100286532956e-05,
+      "loss": 0.0939,
+      "step": 1534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894441260744986e-05,
+      "loss": 0.1035,
+      "step": 1535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894372492836677e-05,
+      "loss": 0.0878,
+      "step": 1536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894303724928367e-05,
+      "loss": 0.095,
+      "step": 1537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894234957020058e-05,
+      "loss": 0.0969,
+      "step": 1538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894166189111748e-05,
+      "loss": 0.1846,
+      "step": 1539
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894097421203439e-05,
+      "loss": 0.0787,
+      "step": 1540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.894028653295129e-05,
+      "loss": 0.1069,
+      "step": 1541
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.89395988538682e-05,
+      "loss": 0.1085,
+      "step": 1542
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.89389111747851e-05,
+      "loss": 0.0962,
+      "step": 1543
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 5.893822349570201e-05,
+      "loss": 0.0648,
+      "step": 1544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.893753581661891e-05,
+      "loss": 0.0833,
+      "step": 1545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8936848137535815e-05,
+      "loss": 0.0892,
+      "step": 1546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8936160458452724e-05,
+      "loss": 0.0603,
+      "step": 1547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8935472779369626e-05,
+      "loss": 0.1466,
+      "step": 1548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8934785100286535e-05,
+      "loss": 0.1213,
+      "step": 1549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.893409742120344e-05,
+      "loss": 0.1097,
+      "step": 1550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8933409742120346e-05,
+      "loss": 0.0849,
+      "step": 1551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.893272206303725e-05,
+      "loss": 0.1205,
+      "step": 1552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.893203438395416e-05,
+      "loss": 0.0656,
+      "step": 1553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.893134670487106e-05,
+      "loss": 0.093,
+      "step": 1554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.893065902578797e-05,
+      "loss": 0.1113,
+      "step": 1555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.892997134670487e-05,
+      "loss": 0.0968,
+      "step": 1556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.892928366762178e-05,
+      "loss": 0.0819,
+      "step": 1557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.892859598853868e-05,
+      "loss": 0.0782,
+      "step": 1558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.892790830945559e-05,
+      "loss": 0.0937,
+      "step": 1559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.892722063037249e-05,
+      "loss": 0.0892,
+      "step": 1560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.89265329512894e-05,
+      "loss": 0.1399,
+      "step": 1561
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.89258452722063e-05,
+      "loss": 0.1096,
+      "step": 1562
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.892515759312321e-05,
+      "loss": 0.0984,
+      "step": 1563
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8924469914040114e-05,
+      "loss": 0.1491,
+      "step": 1564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.892378223495702e-05,
+      "loss": 0.1292,
+      "step": 1565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8923094555873925e-05,
+      "loss": 0.1002,
+      "step": 1566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8922406876790834e-05,
+      "loss": 0.0723,
+      "step": 1567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8921719197707736e-05,
+      "loss": 0.0893,
+      "step": 1568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8921031518624645e-05,
+      "loss": 0.1037,
+      "step": 1569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.892034383954155e-05,
+      "loss": 0.1135,
+      "step": 1570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.8919656160458456e-05,
+      "loss": 0.0982,
+      "step": 1571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.891896848137536e-05,
+      "loss": 0.0999,
+      "step": 1572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.891828080229227e-05,
+      "loss": 0.0468,
+      "step": 1573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.891759312320917e-05,
+      "loss": 0.097,
+      "step": 1574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.891690544412608e-05,
+      "loss": 0.1245,
+      "step": 1575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.891621776504298e-05,
+      "loss": 0.1263,
+      "step": 1576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.891553008595989e-05,
+      "loss": 0.0885,
+      "step": 1577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.891484240687679e-05,
+      "loss": 0.096,
+      "step": 1578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.89141547277937e-05,
+      "loss": 0.1302,
+      "step": 1579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.89134670487106e-05,
+      "loss": 0.1771,
+      "step": 1580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.891277936962751e-05,
+      "loss": 0.1194,
+      "step": 1581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.891209169054441e-05,
+      "loss": 0.0773,
+      "step": 1582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.891140401146132e-05,
+      "loss": 0.0768,
+      "step": 1583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.8910716332378224e-05,
+      "loss": 0.0936,
+      "step": 1584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.891002865329513e-05,
+      "loss": 0.109,
+      "step": 1585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.8909340974212035e-05,
+      "loss": 0.0708,
+      "step": 1586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.8908653295128944e-05,
+      "loss": 0.1087,
+      "step": 1587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.8907965616045846e-05,
+      "loss": 0.1125,
+      "step": 1588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.8907277936962755e-05,
+      "loss": 0.0925,
+      "step": 1589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.890659025787966e-05,
+      "loss": 0.072,
+      "step": 1590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.8905902578796566e-05,
+      "loss": 0.1425,
+      "step": 1591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.890521489971347e-05,
+      "loss": 0.0528,
+      "step": 1592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.890452722063038e-05,
+      "loss": 0.1647,
+      "step": 1593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.890383954154728e-05,
+      "loss": 0.0987,
+      "step": 1594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.890315186246419e-05,
+      "loss": 0.0866,
+      "step": 1595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 5.890246418338109e-05,
+      "loss": 0.1287,
+      "step": 1596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.8901776504298e-05,
+      "loss": 0.0679,
+      "step": 1597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.89010888252149e-05,
+      "loss": 0.0865,
+      "step": 1598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.890040114613181e-05,
+      "loss": 0.1093,
+      "step": 1599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.889971346704871e-05,
+      "loss": 0.0581,
+      "step": 1600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.889902578796562e-05,
+      "loss": 0.0604,
+      "step": 1601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.889833810888252e-05,
+      "loss": 0.0855,
+      "step": 1602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.889765042979943e-05,
+      "loss": 0.072,
+      "step": 1603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.8896962750716334e-05,
+      "loss": 0.0869,
+      "step": 1604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.889627507163324e-05,
+      "loss": 0.075,
+      "step": 1605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.8895587392550145e-05,
+      "loss": 0.0935,
+      "step": 1606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.8894899713467054e-05,
+      "loss": 0.1234,
+      "step": 1607
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.8894212034383956e-05,
+      "loss": 0.1154,
+      "step": 1608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.8893524355300865e-05,
+      "loss": 0.0582,
+      "step": 1609
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.889283667621777e-05,
+      "loss": 0.0827,
+      "step": 1610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.8892148997134676e-05,
+      "loss": 0.106,
+      "step": 1611
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.889146131805158e-05,
+      "loss": 0.0743,
+      "step": 1612
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.889077363896849e-05,
+      "loss": 0.0724,
+      "step": 1613
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 5.889008595988539e-05,
+      "loss": 0.0746,
+      "step": 1614
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.88893982808023e-05,
+      "loss": 0.0907,
+      "step": 1615
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.88887106017192e-05,
+      "loss": 0.1436,
+      "step": 1616
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.888802292263611e-05,
+      "loss": 0.0931,
+      "step": 1617
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.888733524355301e-05,
+      "loss": 0.0694,
+      "step": 1618
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.888664756446992e-05,
+      "loss": 0.0546,
+      "step": 1619
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.888595988538682e-05,
+      "loss": 0.1025,
+      "step": 1620
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.888527220630373e-05,
+      "loss": 0.0747,
+      "step": 1621
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.888458452722063e-05,
+      "loss": 0.0656,
+      "step": 1622
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.888389684813754e-05,
+      "loss": 0.0805,
+      "step": 1623
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8883209169054444e-05,
+      "loss": 0.108,
+      "step": 1624
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.888252148997135e-05,
+      "loss": 0.1102,
+      "step": 1625
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8881833810888255e-05,
+      "loss": 0.1141,
+      "step": 1626
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8881146131805164e-05,
+      "loss": 0.0814,
+      "step": 1627
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8880458452722066e-05,
+      "loss": 0.0611,
+      "step": 1628
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8879770773638975e-05,
+      "loss": 0.0676,
+      "step": 1629
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.887908309455588e-05,
+      "loss": 0.0845,
+      "step": 1630
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.8878395415472786e-05,
+      "loss": 0.0564,
+      "step": 1631
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.887770773638969e-05,
+      "loss": 0.0945,
+      "step": 1632
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.88770200573066e-05,
+      "loss": 0.0908,
+      "step": 1633
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.88763323782235e-05,
+      "loss": 0.069,
+      "step": 1634
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.887564469914041e-05,
+      "loss": 0.1221,
+      "step": 1635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.887495702005731e-05,
+      "loss": 0.1074,
+      "step": 1636
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.887426934097422e-05,
+      "loss": 0.0802,
+      "step": 1637
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.887358166189112e-05,
+      "loss": 0.0845,
+      "step": 1638
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.887289398280803e-05,
+      "loss": 0.0819,
+      "step": 1639
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.887220630372493e-05,
+      "loss": 0.0451,
+      "step": 1640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.887151862464184e-05,
+      "loss": 0.0956,
+      "step": 1641
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.887083094555874e-05,
+      "loss": 0.1024,
+      "step": 1642
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.887014326647565e-05,
+      "loss": 0.0879,
+      "step": 1643
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.8869455587392554e-05,
+      "loss": 0.0805,
+      "step": 1644
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.886876790830946e-05,
+      "loss": 0.1098,
+      "step": 1645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.8868080229226365e-05,
+      "loss": 0.0863,
+      "step": 1646
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.8867392550143274e-05,
+      "loss": 0.0667,
+      "step": 1647
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.8866704871060176e-05,
+      "loss": 0.0578,
+      "step": 1648
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.8866017191977085e-05,
+      "loss": 0.0726,
+      "step": 1649
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.886532951289399e-05,
+      "loss": 0.072,
+      "step": 1650
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8864641833810896e-05,
+      "loss": 0.1083,
+      "step": 1651
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.88639541547278e-05,
+      "loss": 0.1957,
+      "step": 1652
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.88632664756447e-05,
+      "loss": 0.0946,
+      "step": 1653
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.88625787965616e-05,
+      "loss": 0.1983,
+      "step": 1654
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.886189111747851e-05,
+      "loss": 0.1104,
+      "step": 1655
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8861203438395413e-05,
+      "loss": 0.1196,
+      "step": 1656
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.886051575931232e-05,
+      "loss": 0.0866,
+      "step": 1657
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8859828080229224e-05,
+      "loss": 0.0843,
+      "step": 1658
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.885914040114613e-05,
+      "loss": 0.0651,
+      "step": 1659
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8858452722063035e-05,
+      "loss": 0.1403,
+      "step": 1660
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8857765042979944e-05,
+      "loss": 0.0983,
+      "step": 1661
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8857077363896846e-05,
+      "loss": 0.0905,
+      "step": 1662
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8856389684813755e-05,
+      "loss": 0.1525,
+      "step": 1663
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.885570200573066e-05,
+      "loss": 0.0829,
+      "step": 1664
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.885501432664756e-05,
+      "loss": 0.1372,
+      "step": 1665
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.885432664756447e-05,
+      "loss": 0.1074,
+      "step": 1666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.885363896848137e-05,
+      "loss": 0.1007,
+      "step": 1667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.885295128939828e-05,
+      "loss": 0.073,
+      "step": 1668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.885226361031518e-05,
+      "loss": 0.0908,
+      "step": 1669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.885157593123209e-05,
+      "loss": 0.0696,
+      "step": 1670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.885088825214899e-05,
+      "loss": 0.236,
+      "step": 1671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.88502005730659e-05,
+      "loss": 0.061,
+      "step": 1672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8849512893982804e-05,
+      "loss": 0.0959,
+      "step": 1673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.884882521489971e-05,
+      "loss": 0.0758,
+      "step": 1674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8848137535816615e-05,
+      "loss": 0.1883,
+      "step": 1675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8847449856733523e-05,
+      "loss": 0.0913,
+      "step": 1676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8846762177650426e-05,
+      "loss": 0.1014,
+      "step": 1677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8846074498567334e-05,
+      "loss": 0.1111,
+      "step": 1678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8845386819484237e-05,
+      "loss": 0.0826,
+      "step": 1679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8844699140401145e-05,
+      "loss": 0.0949,
+      "step": 1680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.884401146131805e-05,
+      "loss": 0.0909,
+      "step": 1681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.8843323782234956e-05,
+      "loss": 0.0832,
+      "step": 1682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.884263610315186e-05,
+      "loss": 0.0727,
+      "step": 1683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.884194842406877e-05,
+      "loss": 0.0706,
+      "step": 1684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.884126074498567e-05,
+      "loss": 0.076,
+      "step": 1685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.884057306590258e-05,
+      "loss": 0.0743,
+      "step": 1686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.883988538681948e-05,
+      "loss": 0.1161,
+      "step": 1687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.883919770773639e-05,
+      "loss": 0.0746,
+      "step": 1688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.883851002865329e-05,
+      "loss": 0.0634,
+      "step": 1689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.88378223495702e-05,
+      "loss": 0.0789,
+      "step": 1690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.88371346704871e-05,
+      "loss": 0.1005,
+      "step": 1691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.883644699140401e-05,
+      "loss": 0.1015,
+      "step": 1692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8835759312320913e-05,
+      "loss": 0.0516,
+      "step": 1693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.883507163323782e-05,
+      "loss": 0.0847,
+      "step": 1694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8834383954154724e-05,
+      "loss": 0.0656,
+      "step": 1695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.883369627507163e-05,
+      "loss": 0.0647,
+      "step": 1696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8833008595988535e-05,
+      "loss": 0.0567,
+      "step": 1697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8832320916905444e-05,
+      "loss": 0.0685,
+      "step": 1698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8831633237822346e-05,
+      "loss": 0.0986,
+      "step": 1699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8830945558739255e-05,
+      "loss": 0.0881,
+      "step": 1700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.883025787965616e-05,
+      "loss": 0.1251,
+      "step": 1701
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.8829570200573066e-05,
+      "loss": 0.0891,
+      "step": 1702
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.882888252148997e-05,
+      "loss": 0.053,
+      "step": 1703
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.882819484240688e-05,
+      "loss": 0.1177,
+      "step": 1704
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.882750716332378e-05,
+      "loss": 0.0966,
+      "step": 1705
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.882681948424069e-05,
+      "loss": 0.0417,
+      "step": 1706
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.882613180515759e-05,
+      "loss": 0.0795,
+      "step": 1707
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.88254441260745e-05,
+      "loss": 0.0827,
+      "step": 1708
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.88247564469914e-05,
+      "loss": 0.0818,
+      "step": 1709
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.882406876790831e-05,
+      "loss": 0.0797,
+      "step": 1710
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.882338108882521e-05,
+      "loss": 0.0685,
+      "step": 1711
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.882269340974212e-05,
+      "loss": 0.0717,
+      "step": 1712
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.8822005730659023e-05,
+      "loss": 0.0751,
+      "step": 1713
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.882131805157593e-05,
+      "loss": 0.0759,
+      "step": 1714
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.8820630372492834e-05,
+      "loss": 0.1091,
+      "step": 1715
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.881994269340974e-05,
+      "loss": 0.1052,
+      "step": 1716
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.8819255014326645e-05,
+      "loss": 0.0802,
+      "step": 1717
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.8818567335243554e-05,
+      "loss": 0.0744,
+      "step": 1718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8817879656160456e-05,
+      "loss": 0.0608,
+      "step": 1719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8817191977077365e-05,
+      "loss": 0.0907,
+      "step": 1720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.881650429799427e-05,
+      "loss": 0.0534,
+      "step": 1721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8815816618911176e-05,
+      "loss": 0.1141,
+      "step": 1722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.881512893982808e-05,
+      "loss": 0.0892,
+      "step": 1723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.881444126074499e-05,
+      "loss": 0.1177,
+      "step": 1724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.881375358166189e-05,
+      "loss": 0.0948,
+      "step": 1725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.88130659025788e-05,
+      "loss": 0.0484,
+      "step": 1726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.88123782234957e-05,
+      "loss": 0.1085,
+      "step": 1727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.881169054441261e-05,
+      "loss": 0.0698,
+      "step": 1728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.881100286532951e-05,
+      "loss": 0.0877,
+      "step": 1729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.881031518624642e-05,
+      "loss": 0.074,
+      "step": 1730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.880962750716332e-05,
+      "loss": 0.1931,
+      "step": 1731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.880893982808023e-05,
+      "loss": 0.1419,
+      "step": 1732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8808252148997133e-05,
+      "loss": 0.0562,
+      "step": 1733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.880756446991404e-05,
+      "loss": 0.0727,
+      "step": 1734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8806876790830944e-05,
+      "loss": 0.1103,
+      "step": 1735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.880618911174785e-05,
+      "loss": 0.1075,
+      "step": 1736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8805501432664755e-05,
+      "loss": 0.1447,
+      "step": 1737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8804813753581664e-05,
+      "loss": 0.1219,
+      "step": 1738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8804126074498566e-05,
+      "loss": 0.0693,
+      "step": 1739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8803438395415475e-05,
+      "loss": 0.1056,
+      "step": 1740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.880275071633238e-05,
+      "loss": 0.1068,
+      "step": 1741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8802063037249286e-05,
+      "loss": 0.0712,
+      "step": 1742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.880137535816619e-05,
+      "loss": 0.0906,
+      "step": 1743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.88006876790831e-05,
+      "loss": 0.0744,
+      "step": 1744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.88e-05,
+      "loss": 0.0517,
+      "step": 1745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.879931232091691e-05,
+      "loss": 0.0942,
+      "step": 1746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.879862464183381e-05,
+      "loss": 0.0747,
+      "step": 1747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.879793696275072e-05,
+      "loss": 0.0702,
+      "step": 1748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.879724928366762e-05,
+      "loss": 0.0584,
+      "step": 1749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.879656160458453e-05,
+      "loss": 0.0733,
+      "step": 1750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.879587392550143e-05,
+      "loss": 0.0672,
+      "step": 1751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.879518624641834e-05,
+      "loss": 0.0977,
+      "step": 1752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8794498567335243e-05,
+      "loss": 0.0735,
+      "step": 1753
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.879381088825215e-05,
+      "loss": 0.1201,
+      "step": 1754
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8793123209169054e-05,
+      "loss": 0.0905,
+      "step": 1755
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.879243553008596e-05,
+      "loss": 0.0696,
+      "step": 1756
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8791747851002865e-05,
+      "loss": 0.0958,
+      "step": 1757
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8791060171919774e-05,
+      "loss": 0.074,
+      "step": 1758
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8790372492836676e-05,
+      "loss": 0.1063,
+      "step": 1759
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8789684813753585e-05,
+      "loss": 0.1125,
+      "step": 1760
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.878899713467049e-05,
+      "loss": 0.0722,
+      "step": 1761
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.8788309455587396e-05,
+      "loss": 0.0751,
+      "step": 1762
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.87876217765043e-05,
+      "loss": 0.0644,
+      "step": 1763
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.878693409742121e-05,
+      "loss": 0.054,
+      "step": 1764
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.878624641833811e-05,
+      "loss": 0.0606,
+      "step": 1765
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.878555873925502e-05,
+      "loss": 0.0943,
+      "step": 1766
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.878487106017192e-05,
+      "loss": 0.106,
+      "step": 1767
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.878418338108883e-05,
+      "loss": 0.1,
+      "step": 1768
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.878349570200573e-05,
+      "loss": 0.0892,
+      "step": 1769
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.878280802292264e-05,
+      "loss": 0.0547,
+      "step": 1770
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 5.878212034383954e-05,
+      "loss": 0.0676,
+      "step": 1771
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.878143266475645e-05,
+      "loss": 0.093,
+      "step": 1772
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8780744985673353e-05,
+      "loss": 0.0777,
+      "step": 1773
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.878005730659026e-05,
+      "loss": 0.1128,
+      "step": 1774
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8779369627507164e-05,
+      "loss": 0.107,
+      "step": 1775
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.877868194842407e-05,
+      "loss": 0.0867,
+      "step": 1776
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8777994269340975e-05,
+      "loss": 0.0788,
+      "step": 1777
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8777306590257884e-05,
+      "loss": 0.0885,
+      "step": 1778
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8776618911174786e-05,
+      "loss": 0.0999,
+      "step": 1779
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8775931232091695e-05,
+      "loss": 0.0771,
+      "step": 1780
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.87752435530086e-05,
+      "loss": 0.0637,
+      "step": 1781
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.8774555873925506e-05,
+      "loss": 0.0975,
+      "step": 1782
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.877386819484241e-05,
+      "loss": 0.0488,
+      "step": 1783
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.877318051575931e-05,
+      "loss": 0.0679,
+      "step": 1784
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.877249283667622e-05,
+      "loss": 0.0672,
+      "step": 1785
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.877180515759312e-05,
+      "loss": 0.06,
+      "step": 1786
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.877111747851003e-05,
+      "loss": 0.0861,
+      "step": 1787
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 5.877042979942693e-05,
+      "loss": 0.0813,
+      "step": 1788
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.876974212034384e-05,
+      "loss": 0.0573,
+      "step": 1789
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8769054441260744e-05,
+      "loss": 0.1133,
+      "step": 1790
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.876836676217765e-05,
+      "loss": 0.0766,
+      "step": 1791
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8767679083094555e-05,
+      "loss": 0.072,
+      "step": 1792
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8766991404011463e-05,
+      "loss": 0.0716,
+      "step": 1793
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8766303724928366e-05,
+      "loss": 0.064,
+      "step": 1794
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8765616045845274e-05,
+      "loss": 0.1069,
+      "step": 1795
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8764928366762177e-05,
+      "loss": 0.0886,
+      "step": 1796
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8764240687679085e-05,
+      "loss": 0.0709,
+      "step": 1797
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.876355300859599e-05,
+      "loss": 0.0649,
+      "step": 1798
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.8762865329512896e-05,
+      "loss": 0.0888,
+      "step": 1799
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.87621776504298e-05,
+      "loss": 0.0681,
+      "step": 1800
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.876148997134671e-05,
+      "loss": 0.0892,
+      "step": 1801
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.876080229226361e-05,
+      "loss": 0.0667,
+      "step": 1802
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.876011461318052e-05,
+      "loss": 0.0478,
+      "step": 1803
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.875942693409742e-05,
+      "loss": 0.1122,
+      "step": 1804
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.875873925501433e-05,
+      "loss": 0.0464,
+      "step": 1805
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 5.875805157593123e-05,
+      "loss": 0.1111,
+      "step": 1806
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.875736389684814e-05,
+      "loss": 0.1309,
+      "step": 1807
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.875667621776504e-05,
+      "loss": 0.0815,
+      "step": 1808
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.875598853868195e-05,
+      "loss": 0.063,
+      "step": 1809
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.8755300859598854e-05,
+      "loss": 0.0869,
+      "step": 1810
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.875461318051576e-05,
+      "loss": 0.0726,
+      "step": 1811
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.8753925501432665e-05,
+      "loss": 0.1104,
+      "step": 1812
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.8753237822349573e-05,
+      "loss": 0.1267,
+      "step": 1813
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.8752550143266476e-05,
+      "loss": 0.0703,
+      "step": 1814
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.8751862464183384e-05,
+      "loss": 0.0711,
+      "step": 1815
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.8751174785100287e-05,
+      "loss": 0.0601,
+      "step": 1816
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.8750487106017195e-05,
+      "loss": 0.0895,
+      "step": 1817
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.87497994269341e-05,
+      "loss": 0.1152,
+      "step": 1818
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.8749111747851006e-05,
+      "loss": 0.0746,
+      "step": 1819
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.874842406876791e-05,
+      "loss": 0.0593,
+      "step": 1820
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.874773638968482e-05,
+      "loss": 0.0858,
+      "step": 1821
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.874704871060172e-05,
+      "loss": 0.1346,
+      "step": 1822
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 5.874636103151863e-05,
+      "loss": 0.1154,
+      "step": 1823
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.874567335243553e-05,
+      "loss": 0.1356,
+      "step": 1824
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.874498567335244e-05,
+      "loss": 0.082,
+      "step": 1825
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.874429799426934e-05,
+      "loss": 0.0519,
+      "step": 1826
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.874361031518625e-05,
+      "loss": 0.0459,
+      "step": 1827
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.874292263610315e-05,
+      "loss": 0.0595,
+      "step": 1828
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.874223495702006e-05,
+      "loss": 0.0895,
+      "step": 1829
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.8741547277936964e-05,
+      "loss": 0.0885,
+      "step": 1830
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.874085959885387e-05,
+      "loss": 0.1268,
+      "step": 1831
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.8740171919770775e-05,
+      "loss": 0.074,
+      "step": 1832
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.8739484240687683e-05,
+      "loss": 0.0785,
+      "step": 1833
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.8738796561604586e-05,
+      "loss": 0.0755,
+      "step": 1834
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.8738108882521494e-05,
+      "loss": 0.0968,
+      "step": 1835
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.8737421203438397e-05,
+      "loss": 0.0716,
+      "step": 1836
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.8736733524355305e-05,
+      "loss": 0.0824,
+      "step": 1837
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.873604584527221e-05,
+      "loss": 0.0904,
+      "step": 1838
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.8735358166189116e-05,
+      "loss": 0.1161,
+      "step": 1839
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 5.873467048710602e-05,
+      "loss": 0.131,
+      "step": 1840
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.873398280802293e-05,
+      "loss": 0.0861,
+      "step": 1841
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.873329512893983e-05,
+      "loss": 0.069,
+      "step": 1842
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.873260744985674e-05,
+      "loss": 0.0709,
+      "step": 1843
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.873191977077364e-05,
+      "loss": 0.1066,
+      "step": 1844
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.873123209169055e-05,
+      "loss": 0.059,
+      "step": 1845
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.873054441260745e-05,
+      "loss": 0.0803,
+      "step": 1846
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.872985673352436e-05,
+      "loss": 0.1562,
+      "step": 1847
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.872916905444126e-05,
+      "loss": 0.1097,
+      "step": 1848
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.872848137535817e-05,
+      "loss": 0.0711,
+      "step": 1849
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.8727793696275073e-05,
+      "loss": 0.101,
+      "step": 1850
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.872710601719198e-05,
+      "loss": 0.1152,
+      "step": 1851
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.8726418338108884e-05,
+      "loss": 0.1075,
+      "step": 1852
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.872573065902579e-05,
+      "loss": 0.0585,
+      "step": 1853
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.8725042979942695e-05,
+      "loss": 0.0815,
+      "step": 1854
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.8724355300859604e-05,
+      "loss": 0.083,
+      "step": 1855
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.8723667621776506e-05,
+      "loss": 0.0891,
+      "step": 1856
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.8722979942693415e-05,
+      "loss": 0.0947,
+      "step": 1857
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 5.872229226361032e-05,
+      "loss": 0.0719,
+      "step": 1858
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.8721604584527226e-05,
+      "loss": 0.0561,
+      "step": 1859
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.872091690544413e-05,
+      "loss": 0.0994,
+      "step": 1860
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.872022922636104e-05,
+      "loss": 0.0789,
+      "step": 1861
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.871954154727794e-05,
+      "loss": 0.1137,
+      "step": 1862
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.871885386819485e-05,
+      "loss": 0.0694,
+      "step": 1863
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.871816618911175e-05,
+      "loss": 0.0578,
+      "step": 1864
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.871747851002866e-05,
+      "loss": 0.0669,
+      "step": 1865
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.871679083094556e-05,
+      "loss": 0.0517,
+      "step": 1866
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.871610315186247e-05,
+      "loss": 0.0714,
+      "step": 1867
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.871541547277937e-05,
+      "loss": 0.0509,
+      "step": 1868
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.871472779369628e-05,
+      "loss": 0.0973,
+      "step": 1869
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.8714040114613183e-05,
+      "loss": 0.1257,
+      "step": 1870
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.871335243553009e-05,
+      "loss": 0.0576,
+      "step": 1871
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.8712664756446994e-05,
+      "loss": 0.0823,
+      "step": 1872
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.87119770773639e-05,
+      "loss": 0.0941,
+      "step": 1873
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.8711289398280805e-05,
+      "loss": 0.0866,
+      "step": 1874
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 5.8710601719197714e-05,
+      "loss": 0.0824,
+      "step": 1875
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.8709914040114616e-05,
+      "loss": 0.0722,
+      "step": 1876
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.8709226361031525e-05,
+      "loss": 0.0865,
+      "step": 1877
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.870853868194843e-05,
+      "loss": 0.0921,
+      "step": 1878
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.8707851002865336e-05,
+      "loss": 0.0861,
+      "step": 1879
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.870716332378224e-05,
+      "loss": 0.1235,
+      "step": 1880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.870647564469915e-05,
+      "loss": 0.0678,
+      "step": 1881
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.870578796561605e-05,
+      "loss": 0.0712,
+      "step": 1882
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.870510028653296e-05,
+      "loss": 0.0743,
+      "step": 1883
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.870441260744986e-05,
+      "loss": 0.0671,
+      "step": 1884
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.870372492836677e-05,
+      "loss": 0.0565,
+      "step": 1885
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.870303724928367e-05,
+      "loss": 0.0715,
+      "step": 1886
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.870234957020058e-05,
+      "loss": 0.1088,
+      "step": 1887
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.870166189111748e-05,
+      "loss": 0.0676,
+      "step": 1888
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.870097421203439e-05,
+      "loss": 0.063,
+      "step": 1889
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.8700286532951293e-05,
+      "loss": 0.1101,
+      "step": 1890
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.86995988538682e-05,
+      "loss": 0.0581,
+      "step": 1891
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.8698911174785104e-05,
+      "loss": 0.0963,
+      "step": 1892
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 5.869822349570201e-05,
+      "loss": 0.0665,
+      "step": 1893
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.8697535816618915e-05,
+      "loss": 0.0608,
+      "step": 1894
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.8696848137535824e-05,
+      "loss": 0.0729,
+      "step": 1895
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.8696160458452726e-05,
+      "loss": 0.1201,
+      "step": 1896
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.8695472779369635e-05,
+      "loss": 0.0832,
+      "step": 1897
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.869478510028654e-05,
+      "loss": 0.0573,
+      "step": 1898
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.8694097421203446e-05,
+      "loss": 0.0506,
+      "step": 1899
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.869340974212035e-05,
+      "loss": 0.0444,
+      "step": 1900
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.869272206303726e-05,
+      "loss": 0.0911,
+      "step": 1901
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.869203438395416e-05,
+      "loss": 0.0984,
+      "step": 1902
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.869134670487106e-05,
+      "loss": 0.0327,
+      "step": 1903
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.869065902578797e-05,
+      "loss": 0.1053,
+      "step": 1904
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.868997134670487e-05,
+      "loss": 0.0664,
+      "step": 1905
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.868928366762178e-05,
+      "loss": 0.0712,
+      "step": 1906
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.8688595988538684e-05,
+      "loss": 0.06,
+      "step": 1907
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.8687908309455586e-05,
+      "loss": 0.0897,
+      "step": 1908
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.8687220630372495e-05,
+      "loss": 0.0605,
+      "step": 1909
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 5.86865329512894e-05,
+      "loss": 0.1431,
+      "step": 1910
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.86858452722063e-05,
+      "loss": 0.0681,
+      "step": 1911
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.868515759312321e-05,
+      "loss": 0.102,
+      "step": 1912
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.868446991404011e-05,
+      "loss": 0.0819,
+      "step": 1913
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.868378223495702e-05,
+      "loss": 0.1009,
+      "step": 1914
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.868309455587392e-05,
+      "loss": 0.052,
+      "step": 1915
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.868240687679083e-05,
+      "loss": 0.065,
+      "step": 1916
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.868171919770773e-05,
+      "loss": 0.0525,
+      "step": 1917
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.868103151862464e-05,
+      "loss": 0.0859,
+      "step": 1918
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.868034383954154e-05,
+      "loss": 0.0617,
+      "step": 1919
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.867965616045845e-05,
+      "loss": 0.0629,
+      "step": 1920
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.8678968481375354e-05,
+      "loss": 0.0661,
+      "step": 1921
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.867828080229226e-05,
+      "loss": 0.0513,
+      "step": 1922
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.8677593123209165e-05,
+      "loss": 0.0672,
+      "step": 1923
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.8676905444126074e-05,
+      "loss": 0.051,
+      "step": 1924
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.8676217765042976e-05,
+      "loss": 0.0975,
+      "step": 1925
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.8675530085959885e-05,
+      "loss": 0.061,
+      "step": 1926
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.867484240687679e-05,
+      "loss": 0.0917,
+      "step": 1927
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 5.8674154727793696e-05,
+      "loss": 0.06,
+      "step": 1928
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.86734670487106e-05,
+      "loss": 0.0644,
+      "step": 1929
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.867277936962751e-05,
+      "loss": 0.0684,
+      "step": 1930
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.867209169054441e-05,
+      "loss": 0.0657,
+      "step": 1931
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.867140401146132e-05,
+      "loss": 0.1294,
+      "step": 1932
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.867071633237822e-05,
+      "loss": 0.0704,
+      "step": 1933
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.867002865329513e-05,
+      "loss": 0.1069,
+      "step": 1934
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.866934097421203e-05,
+      "loss": 0.0998,
+      "step": 1935
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.866865329512894e-05,
+      "loss": 0.0786,
+      "step": 1936
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.866796561604584e-05,
+      "loss": 0.0912,
+      "step": 1937
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.866727793696275e-05,
+      "loss": 0.0547,
+      "step": 1938
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.866659025787965e-05,
+      "loss": 0.0903,
+      "step": 1939
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.866590257879656e-05,
+      "loss": 0.0959,
+      "step": 1940
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.8665214899713464e-05,
+      "loss": 0.0782,
+      "step": 1941
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.866452722063037e-05,
+      "loss": 0.0665,
+      "step": 1942
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.8663839541547275e-05,
+      "loss": 0.0963,
+      "step": 1943
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.8663151862464184e-05,
+      "loss": 0.119,
+      "step": 1944
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 5.8662464183381086e-05,
+      "loss": 0.0984,
+      "step": 1945
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.8661776504297995e-05,
+      "loss": 0.0435,
+      "step": 1946
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.86610888252149e-05,
+      "loss": 0.092,
+      "step": 1947
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.8660401146131806e-05,
+      "loss": 0.0824,
+      "step": 1948
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865971346704871e-05,
+      "loss": 0.1104,
+      "step": 1949
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865902578796562e-05,
+      "loss": 0.0913,
+      "step": 1950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865833810888252e-05,
+      "loss": 0.0415,
+      "step": 1951
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865765042979943e-05,
+      "loss": 0.1021,
+      "step": 1952
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865696275071633e-05,
+      "loss": 0.0933,
+      "step": 1953
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865627507163324e-05,
+      "loss": 0.0798,
+      "step": 1954
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865558739255014e-05,
+      "loss": 0.0671,
+      "step": 1955
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865489971346705e-05,
+      "loss": 0.1153,
+      "step": 1956
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865421203438395e-05,
+      "loss": 0.0745,
+      "step": 1957
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865352435530086e-05,
+      "loss": 0.0827,
+      "step": 1958
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865283667621776e-05,
+      "loss": 0.0657,
+      "step": 1959
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865214899713467e-05,
+      "loss": 0.1454,
+      "step": 1960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.8651461318051574e-05,
+      "loss": 0.0721,
+      "step": 1961
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.865077363896848e-05,
+      "loss": 0.1237,
+      "step": 1962
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 5.8650085959885385e-05,
+      "loss": 0.0827,
+      "step": 1963
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.8649398280802294e-05,
+      "loss": 0.0445,
+      "step": 1964
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.8648710601719196e-05,
+      "loss": 0.0717,
+      "step": 1965
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.8648022922636105e-05,
+      "loss": 0.0911,
+      "step": 1966
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.864733524355301e-05,
+      "loss": 0.0837,
+      "step": 1967
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.8646647564469916e-05,
+      "loss": 0.0557,
+      "step": 1968
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.864595988538682e-05,
+      "loss": 0.0741,
+      "step": 1969
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.864527220630373e-05,
+      "loss": 0.0695,
+      "step": 1970
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.864458452722063e-05,
+      "loss": 0.1095,
+      "step": 1971
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.864389684813754e-05,
+      "loss": 0.0677,
+      "step": 1972
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.864320916905444e-05,
+      "loss": 0.0764,
+      "step": 1973
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.864252148997135e-05,
+      "loss": 0.0933,
+      "step": 1974
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.864183381088825e-05,
+      "loss": 0.0696,
+      "step": 1975
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.864114613180516e-05,
+      "loss": 0.0576,
+      "step": 1976
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.864045845272206e-05,
+      "loss": 0.0582,
+      "step": 1977
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.863977077363897e-05,
+      "loss": 0.1026,
+      "step": 1978
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.863908309455587e-05,
+      "loss": 0.0705,
+      "step": 1979
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 5.863839541547278e-05,
+      "loss": 0.0425,
+      "step": 1980
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.8637707736389684e-05,
+      "loss": 0.0832,
+      "step": 1981
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.863702005730659e-05,
+      "loss": 0.0811,
+      "step": 1982
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.8636332378223495e-05,
+      "loss": 0.1105,
+      "step": 1983
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.8635644699140404e-05,
+      "loss": 0.0701,
+      "step": 1984
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.8634957020057306e-05,
+      "loss": 0.0814,
+      "step": 1985
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.8634269340974215e-05,
+      "loss": 0.0713,
+      "step": 1986
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.863358166189112e-05,
+      "loss": 0.0833,
+      "step": 1987
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.8632893982808026e-05,
+      "loss": 0.0573,
+      "step": 1988
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.863220630372493e-05,
+      "loss": 0.0576,
+      "step": 1989
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.8631518624641837e-05,
+      "loss": 0.0471,
+      "step": 1990
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.863083094555874e-05,
+      "loss": 0.0532,
+      "step": 1991
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.863014326647565e-05,
+      "loss": 0.0528,
+      "step": 1992
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.862945558739255e-05,
+      "loss": 0.0673,
+      "step": 1993
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.862876790830946e-05,
+      "loss": 0.0956,
+      "step": 1994
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.862808022922636e-05,
+      "loss": 0.1285,
+      "step": 1995
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.862739255014327e-05,
+      "loss": 0.1008,
+      "step": 1996
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.862670487106017e-05,
+      "loss": 0.0648,
+      "step": 1997
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 5.862601719197708e-05,
+      "loss": 0.0536,
+      "step": 1998
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.862532951289398e-05,
+      "loss": 0.0849,
+      "step": 1999
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.862464183381089e-05,
+      "loss": 0.066,
+      "step": 2000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.8623954154727794e-05,
+      "loss": 0.1105,
+      "step": 2001
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.86232664756447e-05,
+      "loss": 0.0676,
+      "step": 2002
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.8622578796561605e-05,
+      "loss": 0.1191,
+      "step": 2003
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.8621891117478514e-05,
+      "loss": 0.0887,
+      "step": 2004
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.8621203438395416e-05,
+      "loss": 0.0586,
+      "step": 2005
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.8620515759312325e-05,
+      "loss": 0.0905,
+      "step": 2006
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.861982808022923e-05,
+      "loss": 0.0603,
+      "step": 2007
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.8619140401146136e-05,
+      "loss": 0.0919,
+      "step": 2008
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.861845272206304e-05,
+      "loss": 0.0718,
+      "step": 2009
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.8617765042979947e-05,
+      "loss": 0.0872,
+      "step": 2010
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.861707736389685e-05,
+      "loss": 0.0568,
+      "step": 2011
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.861638968481376e-05,
+      "loss": 0.0794,
+      "step": 2012
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.861570200573066e-05,
+      "loss": 0.118,
+      "step": 2013
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.861501432664757e-05,
+      "loss": 0.0555,
+      "step": 2014
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 5.861432664756447e-05,
+      "loss": 0.0802,
+      "step": 2015
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.861363896848138e-05,
+      "loss": 0.0755,
+      "step": 2016
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.861295128939828e-05,
+      "loss": 0.054,
+      "step": 2017
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.861226361031519e-05,
+      "loss": 0.089,
+      "step": 2018
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.861157593123209e-05,
+      "loss": 0.0918,
+      "step": 2019
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.8610888252149e-05,
+      "loss": 0.0522,
+      "step": 2020
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.8610200573065904e-05,
+      "loss": 0.0988,
+      "step": 2021
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.8609512893982806e-05,
+      "loss": 0.0603,
+      "step": 2022
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.8608825214899715e-05,
+      "loss": 0.0732,
+      "step": 2023
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.860813753581662e-05,
+      "loss": 0.0462,
+      "step": 2024
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.8607449856733526e-05,
+      "loss": 0.0714,
+      "step": 2025
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.860676217765043e-05,
+      "loss": 0.0633,
+      "step": 2026
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.860607449856734e-05,
+      "loss": 0.0761,
+      "step": 2027
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.860538681948424e-05,
+      "loss": 0.0795,
+      "step": 2028
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.860469914040115e-05,
+      "loss": 0.0785,
+      "step": 2029
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.860401146131805e-05,
+      "loss": 0.0732,
+      "step": 2030
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.860332378223496e-05,
+      "loss": 0.0626,
+      "step": 2031
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 5.860263610315186e-05,
+      "loss": 0.0766,
+      "step": 2032
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.860194842406877e-05,
+      "loss": 0.0757,
+      "step": 2033
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.860126074498567e-05,
+      "loss": 0.1168,
+      "step": 2034
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.860057306590258e-05,
+      "loss": 0.0938,
+      "step": 2035
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.859988538681948e-05,
+      "loss": 0.0743,
+      "step": 2036
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.859919770773639e-05,
+      "loss": 0.0637,
+      "step": 2037
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.8598510028653294e-05,
+      "loss": 0.0421,
+      "step": 2038
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.85978223495702e-05,
+      "loss": 0.0589,
+      "step": 2039
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.8597134670487105e-05,
+      "loss": 0.0725,
+      "step": 2040
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.8596446991404014e-05,
+      "loss": 0.0988,
+      "step": 2041
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.8595759312320916e-05,
+      "loss": 0.0651,
+      "step": 2042
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.8595071633237825e-05,
+      "loss": 0.0657,
+      "step": 2043
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.859438395415473e-05,
+      "loss": 0.0794,
+      "step": 2044
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.8593696275071636e-05,
+      "loss": 0.0878,
+      "step": 2045
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.859300859598854e-05,
+      "loss": 0.0791,
+      "step": 2046
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.859232091690545e-05,
+      "loss": 0.0429,
+      "step": 2047
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.859163323782235e-05,
+      "loss": 0.0814,
+      "step": 2048
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.859094555873926e-05,
+      "loss": 0.1217,
+      "step": 2049
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 5.859025787965616e-05,
+      "loss": 0.1605,
+      "step": 2050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.858957020057307e-05,
+      "loss": 0.0804,
+      "step": 2051
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.858888252148997e-05,
+      "loss": 0.0744,
+      "step": 2052
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.858819484240688e-05,
+      "loss": 0.0451,
+      "step": 2053
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.858750716332378e-05,
+      "loss": 0.038,
+      "step": 2054
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.858681948424069e-05,
+      "loss": 0.1329,
+      "step": 2055
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.858613180515759e-05,
+      "loss": 0.0672,
+      "step": 2056
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.85854441260745e-05,
+      "loss": 0.0868,
+      "step": 2057
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.8584756446991404e-05,
+      "loss": 0.091,
+      "step": 2058
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.858406876790831e-05,
+      "loss": 0.0499,
+      "step": 2059
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.8583381088825215e-05,
+      "loss": 0.0658,
+      "step": 2060
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.8582693409742124e-05,
+      "loss": 0.0918,
+      "step": 2061
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.8582005730659026e-05,
+      "loss": 0.0691,
+      "step": 2062
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.8581318051575935e-05,
+      "loss": 0.0913,
+      "step": 2063
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.858063037249284e-05,
+      "loss": 0.0581,
+      "step": 2064
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.8579942693409746e-05,
+      "loss": 0.0622,
+      "step": 2065
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.857925501432665e-05,
+      "loss": 0.0773,
+      "step": 2066
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 5.857856733524356e-05,
+      "loss": 0.0841,
+      "step": 2067
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.857787965616046e-05,
+      "loss": 0.0746,
+      "step": 2068
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.857719197707737e-05,
+      "loss": 0.0618,
+      "step": 2069
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.857650429799427e-05,
+      "loss": 0.096,
+      "step": 2070
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.857581661891118e-05,
+      "loss": 0.108,
+      "step": 2071
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.857512893982808e-05,
+      "loss": 0.0742,
+      "step": 2072
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.857444126074499e-05,
+      "loss": 0.0975,
+      "step": 2073
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.857375358166189e-05,
+      "loss": 0.0442,
+      "step": 2074
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.85730659025788e-05,
+      "loss": 0.0672,
+      "step": 2075
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.85723782234957e-05,
+      "loss": 0.0969,
+      "step": 2076
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.857169054441261e-05,
+      "loss": 0.0785,
+      "step": 2077
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.8571002865329514e-05,
+      "loss": 0.047,
+      "step": 2078
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.857031518624642e-05,
+      "loss": 0.0896,
+      "step": 2079
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.8569627507163325e-05,
+      "loss": 0.0664,
+      "step": 2080
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.8568939828080234e-05,
+      "loss": 0.0973,
+      "step": 2081
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.8568252148997136e-05,
+      "loss": 0.0707,
+      "step": 2082
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.8567564469914045e-05,
+      "loss": 0.1039,
+      "step": 2083
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.856687679083095e-05,
+      "loss": 0.0957,
+      "step": 2084
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 5.8566189111747856e-05,
+      "loss": 0.1073,
+      "step": 2085
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.856550143266476e-05,
+      "loss": 0.0831,
+      "step": 2086
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.856481375358167e-05,
+      "loss": 0.0557,
+      "step": 2087
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.856412607449857e-05,
+      "loss": 0.0513,
+      "step": 2088
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.856343839541548e-05,
+      "loss": 0.0717,
+      "step": 2089
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.856275071633238e-05,
+      "loss": 0.1565,
+      "step": 2090
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.856206303724929e-05,
+      "loss": 0.0489,
+      "step": 2091
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.856137535816619e-05,
+      "loss": 0.1737,
+      "step": 2092
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.85606876790831e-05,
+      "loss": 0.0501,
+      "step": 2093
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.856e-05,
+      "loss": 0.06,
+      "step": 2094
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.855931232091691e-05,
+      "loss": 0.0616,
+      "step": 2095
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.855862464183381e-05,
+      "loss": 0.0618,
+      "step": 2096
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.855793696275072e-05,
+      "loss": 0.0866,
+      "step": 2097
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.8557249283667624e-05,
+      "loss": 0.0782,
+      "step": 2098
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.855656160458453e-05,
+      "loss": 0.0516,
+      "step": 2099
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.8555873925501435e-05,
+      "loss": 0.0796,
+      "step": 2100
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.8555186246418344e-05,
+      "loss": 0.0524,
+      "step": 2101
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.8554498567335246e-05,
+      "loss": 0.1021,
+      "step": 2102
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.8553810888252155e-05,
+      "loss": 0.0848,
+      "step": 2103
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.855312320916906e-05,
+      "loss": 0.0589,
+      "step": 2104
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.8552435530085966e-05,
+      "loss": 0.0712,
+      "step": 2105
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.855174785100287e-05,
+      "loss": 0.0766,
+      "step": 2106
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.855106017191978e-05,
+      "loss": 0.0918,
+      "step": 2107
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.855037249283668e-05,
+      "loss": 0.0687,
+      "step": 2108
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.854968481375359e-05,
+      "loss": 0.1333,
+      "step": 2109
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.854899713467049e-05,
+      "loss": 0.0588,
+      "step": 2110
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.85483094555874e-05,
+      "loss": 0.0853,
+      "step": 2111
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.85476217765043e-05,
+      "loss": 0.0961,
+      "step": 2112
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.854693409742121e-05,
+      "loss": 0.118,
+      "step": 2113
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.854624641833811e-05,
+      "loss": 0.0996,
+      "step": 2114
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.854555873925502e-05,
+      "loss": 0.0858,
+      "step": 2115
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.854487106017192e-05,
+      "loss": 0.0542,
+      "step": 2116
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.854418338108883e-05,
+      "loss": 0.0801,
+      "step": 2117
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.8543495702005734e-05,
+      "loss": 0.0784,
+      "step": 2118
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.854280802292264e-05,
+      "loss": 0.0597,
+      "step": 2119
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.8542120343839545e-05,
+      "loss": 0.0621,
+      "step": 2120
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.8541432664756454e-05,
+      "loss": 0.1306,
+      "step": 2121
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.8540744985673356e-05,
+      "loss": 0.0593,
+      "step": 2122
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.8540057306590265e-05,
+      "loss": 0.0788,
+      "step": 2123
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.853936962750717e-05,
+      "loss": 0.0715,
+      "step": 2124
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.8538681948424076e-05,
+      "loss": 0.0702,
+      "step": 2125
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.853799426934098e-05,
+      "loss": 0.0732,
+      "step": 2126
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.853730659025789e-05,
+      "loss": 0.0803,
+      "step": 2127
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.853661891117479e-05,
+      "loss": 0.0847,
+      "step": 2128
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.85359312320917e-05,
+      "loss": 0.067,
+      "step": 2129
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.85352435530086e-05,
+      "loss": 0.0556,
+      "step": 2130
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.853455587392551e-05,
+      "loss": 0.075,
+      "step": 2131
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.853386819484241e-05,
+      "loss": 0.0665,
+      "step": 2132
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.853318051575932e-05,
+      "loss": 0.0626,
+      "step": 2133
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.853249283667622e-05,
+      "loss": 0.1258,
+      "step": 2134
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.853180515759313e-05,
+      "loss": 0.0704,
+      "step": 2135
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.853111747851003e-05,
+      "loss": 0.082,
+      "step": 2136
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.853042979942694e-05,
+      "loss": 0.0748,
+      "step": 2137
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.8529742120343844e-05,
+      "loss": 0.065,
+      "step": 2138
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.852905444126075e-05,
+      "loss": 0.095,
+      "step": 2139
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.8528366762177655e-05,
+      "loss": 0.0753,
+      "step": 2140
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.852767908309456e-05,
+      "loss": 0.0864,
+      "step": 2141
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.8526991404011466e-05,
+      "loss": 0.0679,
+      "step": 2142
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.852630372492837e-05,
+      "loss": 0.07,
+      "step": 2143
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.852561604584528e-05,
+      "loss": 0.0845,
+      "step": 2144
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.852492836676218e-05,
+      "loss": 0.0903,
+      "step": 2145
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.852424068767909e-05,
+      "loss": 0.1116,
+      "step": 2146
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.852355300859599e-05,
+      "loss": 0.0829,
+      "step": 2147
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.85228653295129e-05,
+      "loss": 0.0855,
+      "step": 2148
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.85221776504298e-05,
+      "loss": 0.083,
+      "step": 2149
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.852148997134671e-05,
+      "loss": 0.0806,
+      "step": 2150
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.852080229226361e-05,
+      "loss": 0.0957,
+      "step": 2151
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.852011461318052e-05,
+      "loss": 0.0657,
+      "step": 2152
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.851942693409742e-05,
+      "loss": 0.0679,
+      "step": 2153
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.851873925501433e-05,
+      "loss": 0.0894,
+      "step": 2154
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.8518051575931234e-05,
+      "loss": 0.0946,
+      "step": 2155
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.851736389684814e-05,
+      "loss": 0.1001,
+      "step": 2156
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.8516676217765045e-05,
+      "loss": 0.101,
+      "step": 2157
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.8515988538681954e-05,
+      "loss": 0.0539,
+      "step": 2158
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.8515300859598856e-05,
+      "loss": 0.1279,
+      "step": 2159
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.8514613180515765e-05,
+      "loss": 0.112,
+      "step": 2160
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.851392550143267e-05,
+      "loss": 0.1128,
+      "step": 2161
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.851323782234957e-05,
+      "loss": 0.1055,
+      "step": 2162
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.851255014326648e-05,
+      "loss": 0.0883,
+      "step": 2163
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.851186246418338e-05,
+      "loss": 0.0603,
+      "step": 2164
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.851117478510028e-05,
+      "loss": 0.0918,
+      "step": 2165
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.851048710601719e-05,
+      "loss": 0.0826,
+      "step": 2166
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.850979942693409e-05,
+      "loss": 0.0539,
+      "step": 2167
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.8509111747851e-05,
+      "loss": 0.0635,
+      "step": 2168
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.8508424068767904e-05,
+      "loss": 0.1361,
+      "step": 2169
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.850773638968481e-05,
+      "loss": 0.075,
+      "step": 2170
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.8507048710601715e-05,
+      "loss": 0.0673,
+      "step": 2171
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.8506361031518624e-05,
+      "loss": 0.0643,
+      "step": 2172
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.8505673352435526e-05,
+      "loss": 0.0612,
+      "step": 2173
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.8504985673352435e-05,
+      "loss": 0.0807,
+      "step": 2174
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.850429799426934e-05,
+      "loss": 0.0696,
+      "step": 2175
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.8503610315186246e-05,
+      "loss": 0.0449,
+      "step": 2176
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.850292263610315e-05,
+      "loss": 0.0951,
+      "step": 2177
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.850223495702006e-05,
+      "loss": 0.0697,
+      "step": 2178
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.850154727793696e-05,
+      "loss": 0.0771,
+      "step": 2179
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.850085959885387e-05,
+      "loss": 0.0729,
+      "step": 2180
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.850017191977077e-05,
+      "loss": 0.0454,
+      "step": 2181
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.849948424068768e-05,
+      "loss": 0.0515,
+      "step": 2182
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.849879656160458e-05,
+      "loss": 0.0786,
+      "step": 2183
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.849810888252149e-05,
+      "loss": 0.0542,
+      "step": 2184
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.849742120343839e-05,
+      "loss": 0.0648,
+      "step": 2185
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.84967335243553e-05,
+      "loss": 0.0606,
+      "step": 2186
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.84960458452722e-05,
+      "loss": 0.0732,
+      "step": 2187
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.849535816618911e-05,
+      "loss": 0.093,
+      "step": 2188
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.8494670487106014e-05,
+      "loss": 0.0752,
+      "step": 2189
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.849398280802292e-05,
+      "loss": 0.0612,
+      "step": 2190
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.8493295128939825e-05,
+      "loss": 0.0289,
+      "step": 2191
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.8492607449856734e-05,
+      "loss": 0.0543,
+      "step": 2192
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.8491919770773636e-05,
+      "loss": 0.0644,
+      "step": 2193
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.8491232091690545e-05,
+      "loss": 0.0622,
+      "step": 2194
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.849054441260745e-05,
+      "loss": 0.0669,
+      "step": 2195
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.8489856733524356e-05,
+      "loss": 0.0623,
+      "step": 2196
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.848916905444126e-05,
+      "loss": 0.1147,
+      "step": 2197
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.848848137535817e-05,
+      "loss": 0.0726,
+      "step": 2198
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.848779369627507e-05,
+      "loss": 0.0419,
+      "step": 2199
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.848710601719198e-05,
+      "loss": 0.0591,
+      "step": 2200
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.848641833810888e-05,
+      "loss": 0.0582,
+      "step": 2201
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.848573065902579e-05,
+      "loss": 0.0712,
+      "step": 2202
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.848504297994269e-05,
+      "loss": 0.0617,
+      "step": 2203
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.84843553008596e-05,
+      "loss": 0.0597,
+      "step": 2204
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.84836676217765e-05,
+      "loss": 0.1142,
+      "step": 2205
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.848297994269341e-05,
+      "loss": 0.0853,
+      "step": 2206
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.848229226361031e-05,
+      "loss": 0.055,
+      "step": 2207
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.848160458452722e-05,
+      "loss": 0.0519,
+      "step": 2208
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.8480916905444124e-05,
+      "loss": 0.0719,
+      "step": 2209
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.848022922636103e-05,
+      "loss": 0.1033,
+      "step": 2210
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.8479541547277935e-05,
+      "loss": 0.0546,
+      "step": 2211
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.8478853868194844e-05,
+      "loss": 0.0703,
+      "step": 2212
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.8478166189111746e-05,
+      "loss": 0.0861,
+      "step": 2213
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.8477478510028655e-05,
+      "loss": 0.0572,
+      "step": 2214
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.847679083094556e-05,
+      "loss": 0.0839,
+      "step": 2215
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.8476103151862466e-05,
+      "loss": 0.0549,
+      "step": 2216
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.847541547277937e-05,
+      "loss": 0.0648,
+      "step": 2217
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.847472779369628e-05,
+      "loss": 0.0466,
+      "step": 2218
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.847404011461318e-05,
+      "loss": 0.0956,
+      "step": 2219
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.847335243553009e-05,
+      "loss": 0.1244,
+      "step": 2220
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.847266475644699e-05,
+      "loss": 0.1148,
+      "step": 2221
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.84719770773639e-05,
+      "loss": 0.0695,
+      "step": 2222
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.84712893982808e-05,
+      "loss": 0.0545,
+      "step": 2223
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.847060171919771e-05,
+      "loss": 0.0427,
+      "step": 2224
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.846991404011461e-05,
+      "loss": 0.0792,
+      "step": 2225
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.846922636103152e-05,
+      "loss": 0.1241,
+      "step": 2226
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.846853868194842e-05,
+      "loss": 0.0876,
+      "step": 2227
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.846785100286533e-05,
+      "loss": 0.0466,
+      "step": 2228
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.8467163323782234e-05,
+      "loss": 0.0698,
+      "step": 2229
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.846647564469914e-05,
+      "loss": 0.104,
+      "step": 2230
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.8465787965616045e-05,
+      "loss": 0.0531,
+      "step": 2231
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.8465100286532954e-05,
+      "loss": 0.0934,
+      "step": 2232
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.8464412607449856e-05,
+      "loss": 0.0893,
+      "step": 2233
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.8463724928366765e-05,
+      "loss": 0.06,
+      "step": 2234
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.846303724928367e-05,
+      "loss": 0.055,
+      "step": 2235
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.8462349570200576e-05,
+      "loss": 0.0706,
+      "step": 2236
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.846166189111748e-05,
+      "loss": 0.0876,
+      "step": 2237
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.846097421203439e-05,
+      "loss": 0.063,
+      "step": 2238
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.846028653295129e-05,
+      "loss": 0.0973,
+      "step": 2239
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.84595988538682e-05,
+      "loss": 0.1012,
+      "step": 2240
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.84589111747851e-05,
+      "loss": 0.0957,
+      "step": 2241
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.845822349570201e-05,
+      "loss": 0.0714,
+      "step": 2242
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.845753581661891e-05,
+      "loss": 0.0301,
+      "step": 2243
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.845684813753582e-05,
+      "loss": 0.0787,
+      "step": 2244
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.845616045845272e-05,
+      "loss": 0.067,
+      "step": 2245
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.845547277936963e-05,
+      "loss": 0.0984,
+      "step": 2246
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.845478510028653e-05,
+      "loss": 0.0802,
+      "step": 2247
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.845409742120344e-05,
+      "loss": 0.0655,
+      "step": 2248
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8453409742120344e-05,
+      "loss": 0.0579,
+      "step": 2249
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.845272206303725e-05,
+      "loss": 0.0518,
+      "step": 2250
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8452034383954155e-05,
+      "loss": 0.0641,
+      "step": 2251
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8451346704871064e-05,
+      "loss": 0.039,
+      "step": 2252
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8450659025787966e-05,
+      "loss": 0.0745,
+      "step": 2253
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8449971346704875e-05,
+      "loss": 0.0726,
+      "step": 2254
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.844928366762178e-05,
+      "loss": 0.064,
+      "step": 2255
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.8448595988538686e-05,
+      "loss": 0.0777,
+      "step": 2256
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.844790830945559e-05,
+      "loss": 0.0805,
+      "step": 2257
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.84472206303725e-05,
+      "loss": 0.066,
+      "step": 2258
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.84465329512894e-05,
+      "loss": 0.0645,
+      "step": 2259
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.84458452722063e-05,
+      "loss": 0.1126,
+      "step": 2260
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.844515759312321e-05,
+      "loss": 0.1183,
+      "step": 2261
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.844446991404011e-05,
+      "loss": 0.0717,
+      "step": 2262
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.844378223495702e-05,
+      "loss": 0.0932,
+      "step": 2263
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.844309455587392e-05,
+      "loss": 0.0683,
+      "step": 2264
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.844240687679083e-05,
+      "loss": 0.0681,
+      "step": 2265
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8441719197707734e-05,
+      "loss": 0.0685,
+      "step": 2266
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.844103151862464e-05,
+      "loss": 0.167,
+      "step": 2267
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8440343839541545e-05,
+      "loss": 0.0797,
+      "step": 2268
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8439656160458454e-05,
+      "loss": 0.0582,
+      "step": 2269
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8438968481375356e-05,
+      "loss": 0.0698,
+      "step": 2270
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8438280802292265e-05,
+      "loss": 0.0637,
+      "step": 2271
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.843759312320917e-05,
+      "loss": 0.0439,
+      "step": 2272
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.8436905444126076e-05,
+      "loss": 0.0624,
+      "step": 2273
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.843621776504298e-05,
+      "loss": 0.0851,
+      "step": 2274
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.843553008595989e-05,
+      "loss": 0.0826,
+      "step": 2275
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.843484240687679e-05,
+      "loss": 0.0451,
+      "step": 2276
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.84341547277937e-05,
+      "loss": 0.055,
+      "step": 2277
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.84334670487106e-05,
+      "loss": 0.0425,
+      "step": 2278
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.843277936962751e-05,
+      "loss": 0.0531,
+      "step": 2279
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.843209169054441e-05,
+      "loss": 0.1262,
+      "step": 2280
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.843140401146132e-05,
+      "loss": 0.0958,
+      "step": 2281
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.843071633237822e-05,
+      "loss": 0.0524,
+      "step": 2282
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.843002865329513e-05,
+      "loss": 0.0315,
+      "step": 2283
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.842934097421203e-05,
+      "loss": 0.0702,
+      "step": 2284
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.842865329512894e-05,
+      "loss": 0.0859,
+      "step": 2285
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.8427965616045844e-05,
+      "loss": 0.0781,
+      "step": 2286
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.842727793696275e-05,
+      "loss": 0.1264,
+      "step": 2287
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.8426590257879655e-05,
+      "loss": 0.1184,
+      "step": 2288
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.8425902578796564e-05,
+      "loss": 0.0812,
+      "step": 2289
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.8425214899713466e-05,
+      "loss": 0.1408,
+      "step": 2290
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.8424527220630375e-05,
+      "loss": 0.1061,
+      "step": 2291
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.842383954154728e-05,
+      "loss": 0.0522,
+      "step": 2292
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.8423151862464186e-05,
+      "loss": 0.0484,
+      "step": 2293
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.842246418338109e-05,
+      "loss": 0.0608,
+      "step": 2294
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.8421776504298e-05,
+      "loss": 0.0647,
+      "step": 2295
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.84210888252149e-05,
+      "loss": 0.0763,
+      "step": 2296
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.842040114613181e-05,
+      "loss": 0.0787,
+      "step": 2297
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.841971346704871e-05,
+      "loss": 0.0619,
+      "step": 2298
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.841902578796562e-05,
+      "loss": 0.0741,
+      "step": 2299
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.841833810888252e-05,
+      "loss": 0.086,
+      "step": 2300
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.841765042979943e-05,
+      "loss": 0.0925,
+      "step": 2301
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.841696275071633e-05,
+      "loss": 0.0661,
+      "step": 2302
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.841627507163324e-05,
+      "loss": 0.1418,
+      "step": 2303
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.841558739255014e-05,
+      "loss": 0.0871,
+      "step": 2304
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.841489971346705e-05,
+      "loss": 0.082,
+      "step": 2305
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.8414212034383954e-05,
+      "loss": 0.052,
+      "step": 2306
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.841352435530086e-05,
+      "loss": 0.0514,
+      "step": 2307
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.8412836676217765e-05,
+      "loss": 0.0706,
+      "step": 2308
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.8412148997134674e-05,
+      "loss": 0.1114,
+      "step": 2309
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.8411461318051576e-05,
+      "loss": 0.1124,
+      "step": 2310
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.8410773638968485e-05,
+      "loss": 0.0679,
+      "step": 2311
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.841008595988539e-05,
+      "loss": 0.075,
+      "step": 2312
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.8409398280802296e-05,
+      "loss": 0.0873,
+      "step": 2313
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.84087106017192e-05,
+      "loss": 0.1048,
+      "step": 2314
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.840802292263611e-05,
+      "loss": 0.0694,
+      "step": 2315
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.840733524355301e-05,
+      "loss": 0.0711,
+      "step": 2316
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.840664756446992e-05,
+      "loss": 0.0999,
+      "step": 2317
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.840595988538682e-05,
+      "loss": 0.102,
+      "step": 2318
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.840527220630373e-05,
+      "loss": 0.1155,
+      "step": 2319
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.840458452722063e-05,
+      "loss": 0.0625,
+      "step": 2320
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.840389684813754e-05,
+      "loss": 0.1041,
+      "step": 2321
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.840320916905444e-05,
+      "loss": 0.058,
+      "step": 2322
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.840252148997135e-05,
+      "loss": 0.064,
+      "step": 2323
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.840183381088825e-05,
+      "loss": 0.0526,
+      "step": 2324
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.840114613180516e-05,
+      "loss": 0.0842,
+      "step": 2325
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.8400458452722064e-05,
+      "loss": 0.1043,
+      "step": 2326
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.839977077363897e-05,
+      "loss": 0.0961,
+      "step": 2327
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.8399083094555875e-05,
+      "loss": 0.0363,
+      "step": 2328
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.8398395415472784e-05,
+      "loss": 0.0797,
+      "step": 2329
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.8397707736389686e-05,
+      "loss": 0.0615,
+      "step": 2330
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.8397020057306595e-05,
+      "loss": 0.0701,
+      "step": 2331
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.83963323782235e-05,
+      "loss": 0.1002,
+      "step": 2332
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.8395644699140406e-05,
+      "loss": 0.0537,
+      "step": 2333
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.839495702005731e-05,
+      "loss": 0.0938,
+      "step": 2334
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.839426934097422e-05,
+      "loss": 0.1048,
+      "step": 2335
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.839358166189112e-05,
+      "loss": 0.0576,
+      "step": 2336
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.839289398280803e-05,
+      "loss": 0.0914,
+      "step": 2337
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.839220630372493e-05,
+      "loss": 0.0856,
+      "step": 2338
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.839151862464184e-05,
+      "loss": 0.0449,
+      "step": 2339
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.839083094555874e-05,
+      "loss": 0.0569,
+      "step": 2340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.839014326647565e-05,
+      "loss": 0.1039,
+      "step": 2341
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.838945558739255e-05,
+      "loss": 0.068,
+      "step": 2342
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.838876790830946e-05,
+      "loss": 0.1198,
+      "step": 2343
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.838808022922636e-05,
+      "loss": 0.0619,
+      "step": 2344
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.838739255014327e-05,
+      "loss": 0.0754,
+      "step": 2345
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.8386704871060174e-05,
+      "loss": 0.0687,
+      "step": 2346
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.838601719197708e-05,
+      "loss": 0.0824,
+      "step": 2347
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.8385329512893985e-05,
+      "loss": 0.1311,
+      "step": 2348
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.8384641833810894e-05,
+      "loss": 0.1428,
+      "step": 2349
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.8383954154727796e-05,
+      "loss": 0.0734,
+      "step": 2350
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.8383266475644705e-05,
+      "loss": 0.0879,
+      "step": 2351
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.838257879656161e-05,
+      "loss": 0.0857,
+      "step": 2352
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.8381891117478516e-05,
+      "loss": 0.1132,
+      "step": 2353
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.838120343839542e-05,
+      "loss": 0.0377,
+      "step": 2354
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.838051575931233e-05,
+      "loss": 0.1048,
+      "step": 2355
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.837982808022923e-05,
+      "loss": 0.0713,
+      "step": 2356
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.837914040114614e-05,
+      "loss": 0.0627,
+      "step": 2357
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.837845272206304e-05,
+      "loss": 0.0781,
+      "step": 2358
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.837776504297995e-05,
+      "loss": 0.0889,
+      "step": 2359
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.837707736389685e-05,
+      "loss": 0.0781,
+      "step": 2360
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.837638968481376e-05,
+      "loss": 0.0486,
+      "step": 2361
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.837570200573066e-05,
+      "loss": 0.0617,
+      "step": 2362
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.837501432664757e-05,
+      "loss": 0.0575,
+      "step": 2363
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.837432664756447e-05,
+      "loss": 0.0694,
+      "step": 2364
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.837363896848138e-05,
+      "loss": 0.077,
+      "step": 2365
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.8372951289398284e-05,
+      "loss": 0.0874,
+      "step": 2366
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.837226361031519e-05,
+      "loss": 0.0566,
+      "step": 2367
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.8371575931232095e-05,
+      "loss": 0.0658,
+      "step": 2368
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.8370888252149004e-05,
+      "loss": 0.0723,
+      "step": 2369
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.8370200573065906e-05,
+      "loss": 0.0746,
+      "step": 2370
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.8369512893982815e-05,
+      "loss": 0.0698,
+      "step": 2371
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.836882521489972e-05,
+      "loss": 0.0707,
+      "step": 2372
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.8368137535816626e-05,
+      "loss": 0.0721,
+      "step": 2373
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.836744985673353e-05,
+      "loss": 0.0547,
+      "step": 2374
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.836676217765044e-05,
+      "loss": 0.0445,
+      "step": 2375
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.836607449856734e-05,
+      "loss": 0.0447,
+      "step": 2376
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.836538681948425e-05,
+      "loss": 0.0532,
+      "step": 2377
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.836469914040115e-05,
+      "loss": 0.0879,
+      "step": 2378
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.836401146131805e-05,
+      "loss": 0.0737,
+      "step": 2379
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.836332378223496e-05,
+      "loss": 0.0831,
+      "step": 2380
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.836263610315186e-05,
+      "loss": 0.073,
+      "step": 2381
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.836194842406877e-05,
+      "loss": 0.0842,
+      "step": 2382
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.8361260744985674e-05,
+      "loss": 0.064,
+      "step": 2383
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.836057306590258e-05,
+      "loss": 0.0566,
+      "step": 2384
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.8359885386819485e-05,
+      "loss": 0.0804,
+      "step": 2385
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.8359197707736394e-05,
+      "loss": 0.0721,
+      "step": 2386
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.8358510028653296e-05,
+      "loss": 0.0937,
+      "step": 2387
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.8357822349570205e-05,
+      "loss": 0.0842,
+      "step": 2388
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.835713467048711e-05,
+      "loss": 0.0526,
+      "step": 2389
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.8356446991404016e-05,
+      "loss": 0.0631,
+      "step": 2390
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.835575931232092e-05,
+      "loss": 0.0754,
+      "step": 2391
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.835507163323783e-05,
+      "loss": 0.119,
+      "step": 2392
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.835438395415473e-05,
+      "loss": 0.0651,
+      "step": 2393
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.835369627507164e-05,
+      "loss": 0.063,
+      "step": 2394
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.835300859598854e-05,
+      "loss": 0.0709,
+      "step": 2395
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.835232091690545e-05,
+      "loss": 0.0583,
+      "step": 2396
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.835163323782235e-05,
+      "loss": 0.1009,
+      "step": 2397
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.835094555873926e-05,
+      "loss": 0.0863,
+      "step": 2398
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.835025787965616e-05,
+      "loss": 0.0864,
+      "step": 2399
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.834957020057307e-05,
+      "loss": 0.1146,
+      "step": 2400
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.834888252148997e-05,
+      "loss": 0.1163,
+      "step": 2401
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.834819484240688e-05,
+      "loss": 0.1115,
+      "step": 2402
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.8347507163323784e-05,
+      "loss": 0.0548,
+      "step": 2403
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.834681948424069e-05,
+      "loss": 0.0643,
+      "step": 2404
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.8346131805157595e-05,
+      "loss": 0.043,
+      "step": 2405
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.8345444126074504e-05,
+      "loss": 0.0828,
+      "step": 2406
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.8344756446991406e-05,
+      "loss": 0.1182,
+      "step": 2407
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.8344068767908315e-05,
+      "loss": 0.1094,
+      "step": 2408
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.834338108882522e-05,
+      "loss": 0.1555,
+      "step": 2409
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.8342693409742126e-05,
+      "loss": 0.108,
+      "step": 2410
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.834200573065903e-05,
+      "loss": 0.0825,
+      "step": 2411
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.834131805157594e-05,
+      "loss": 0.0837,
+      "step": 2412
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.834063037249284e-05,
+      "loss": 0.0507,
+      "step": 2413
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.833994269340975e-05,
+      "loss": 0.0719,
+      "step": 2414
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.833925501432665e-05,
+      "loss": 0.1021,
+      "step": 2415
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.833856733524355e-05,
+      "loss": 0.0443,
+      "step": 2416
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.833787965616046e-05,
+      "loss": 0.071,
+      "step": 2417
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.833719197707736e-05,
+      "loss": 0.0821,
+      "step": 2418
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.8336504297994265e-05,
+      "loss": 0.0669,
+      "step": 2419
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.8335816618911174e-05,
+      "loss": 0.079,
+      "step": 2420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.8335128939828076e-05,
+      "loss": 0.0634,
+      "step": 2421
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.8334441260744985e-05,
+      "loss": 0.0559,
+      "step": 2422
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.833375358166189e-05,
+      "loss": 0.0693,
+      "step": 2423
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.8333065902578796e-05,
+      "loss": 0.0825,
+      "step": 2424
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.83323782234957e-05,
+      "loss": 0.0883,
+      "step": 2425
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.833169054441261e-05,
+      "loss": 0.0456,
+      "step": 2426
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.833100286532951e-05,
+      "loss": 0.0987,
+      "step": 2427
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.833031518624642e-05,
+      "loss": 0.0833,
+      "step": 2428
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.832962750716332e-05,
+      "loss": 0.0966,
+      "step": 2429
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.832893982808023e-05,
+      "loss": 0.0451,
+      "step": 2430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.832825214899713e-05,
+      "loss": 0.0526,
+      "step": 2431
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.832756446991404e-05,
+      "loss": 0.0553,
+      "step": 2432
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.832687679083094e-05,
+      "loss": 0.0306,
+      "step": 2433
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.832618911174785e-05,
+      "loss": 0.0965,
+      "step": 2434
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.832550143266475e-05,
+      "loss": 0.0355,
+      "step": 2435
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.832481375358166e-05,
+      "loss": 0.0639,
+      "step": 2436
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8324126074498564e-05,
+      "loss": 0.0785,
+      "step": 2437
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.832343839541547e-05,
+      "loss": 0.0652,
+      "step": 2438
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8322750716332375e-05,
+      "loss": 0.0814,
+      "step": 2439
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8322063037249284e-05,
+      "loss": 0.0679,
+      "step": 2440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8321375358166186e-05,
+      "loss": 0.0753,
+      "step": 2441
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8320687679083095e-05,
+      "loss": 0.0601,
+      "step": 2442
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.832e-05,
+      "loss": 0.0331,
+      "step": 2443
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.8319312320916906e-05,
+      "loss": 0.1033,
+      "step": 2444
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.831862464183381e-05,
+      "loss": 0.0614,
+      "step": 2445
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.831793696275072e-05,
+      "loss": 0.0749,
+      "step": 2446
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.831724928366762e-05,
+      "loss": 0.0821,
+      "step": 2447
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.831656160458453e-05,
+      "loss": 0.039,
+      "step": 2448
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.831587392550143e-05,
+      "loss": 0.0529,
+      "step": 2449
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.831518624641834e-05,
+      "loss": 0.0547,
+      "step": 2450
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.831449856733524e-05,
+      "loss": 0.0665,
+      "step": 2451
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.831381088825215e-05,
+      "loss": 0.0373,
+      "step": 2452
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.831312320916905e-05,
+      "loss": 0.1095,
+      "step": 2453
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.831243553008596e-05,
+      "loss": 0.0517,
+      "step": 2454
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.831174785100286e-05,
+      "loss": 0.046,
+      "step": 2455
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.831106017191977e-05,
+      "loss": 0.0825,
+      "step": 2456
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8310372492836674e-05,
+      "loss": 0.0596,
+      "step": 2457
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.830968481375358e-05,
+      "loss": 0.1232,
+      "step": 2458
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8308997134670485e-05,
+      "loss": 0.1556,
+      "step": 2459
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8308309455587394e-05,
+      "loss": 0.0556,
+      "step": 2460
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8307621776504296e-05,
+      "loss": 0.0481,
+      "step": 2461
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8306934097421205e-05,
+      "loss": 0.0632,
+      "step": 2462
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.830624641833811e-05,
+      "loss": 0.0627,
+      "step": 2463
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.8305558739255016e-05,
+      "loss": 0.0372,
+      "step": 2464
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.830487106017192e-05,
+      "loss": 0.0592,
+      "step": 2465
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.830418338108883e-05,
+      "loss": 0.0535,
+      "step": 2466
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.830349570200573e-05,
+      "loss": 0.0709,
+      "step": 2467
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.830280802292264e-05,
+      "loss": 0.0842,
+      "step": 2468
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.830212034383954e-05,
+      "loss": 0.0727,
+      "step": 2469
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.830143266475645e-05,
+      "loss": 0.1101,
+      "step": 2470
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.830074498567335e-05,
+      "loss": 0.133,
+      "step": 2471
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.830005730659026e-05,
+      "loss": 0.0782,
+      "step": 2472
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.829936962750716e-05,
+      "loss": 0.0802,
+      "step": 2473
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.829868194842407e-05,
+      "loss": 0.0618,
+      "step": 2474
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.829799426934097e-05,
+      "loss": 0.0608,
+      "step": 2475
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.829730659025788e-05,
+      "loss": 0.0585,
+      "step": 2476
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8296618911174784e-05,
+      "loss": 0.0598,
+      "step": 2477
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.829593123209169e-05,
+      "loss": 0.0722,
+      "step": 2478
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8295243553008595e-05,
+      "loss": 0.069,
+      "step": 2479
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8294555873925504e-05,
+      "loss": 0.0637,
+      "step": 2480
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8293868194842406e-05,
+      "loss": 0.06,
+      "step": 2481
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8293180515759315e-05,
+      "loss": 0.0709,
+      "step": 2482
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.829249283667622e-05,
+      "loss": 0.1506,
+      "step": 2483
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.8291805157593126e-05,
+      "loss": 0.073,
+      "step": 2484
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.829111747851003e-05,
+      "loss": 0.1132,
+      "step": 2485
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.829042979942694e-05,
+      "loss": 0.073,
+      "step": 2486
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.828974212034384e-05,
+      "loss": 0.0493,
+      "step": 2487
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.828905444126075e-05,
+      "loss": 0.0886,
+      "step": 2488
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.828836676217765e-05,
+      "loss": 0.0552,
+      "step": 2489
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.828767908309456e-05,
+      "loss": 0.0773,
+      "step": 2490
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.828699140401146e-05,
+      "loss": 0.0595,
+      "step": 2491
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.828630372492837e-05,
+      "loss": 0.2183,
+      "step": 2492
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.828561604584527e-05,
+      "loss": 0.0811,
+      "step": 2493
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.828492836676218e-05,
+      "loss": 0.0473,
+      "step": 2494
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.828424068767908e-05,
+      "loss": 0.1812,
+      "step": 2495
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.828355300859599e-05,
+      "loss": 0.0602,
+      "step": 2496
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8282865329512894e-05,
+      "loss": 0.0688,
+      "step": 2497
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8282177650429796e-05,
+      "loss": 0.0359,
+      "step": 2498
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8281489971346705e-05,
+      "loss": 0.0536,
+      "step": 2499
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.828080229226361e-05,
+      "loss": 0.1436,
+      "step": 2500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.8280114613180516e-05,
+      "loss": 0.0645,
+      "step": 2501
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.827942693409742e-05,
+      "loss": 0.0327,
+      "step": 2502
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.827873925501433e-05,
+      "loss": 0.0646,
+      "step": 2503
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.827805157593123e-05,
+      "loss": 0.0907,
+      "step": 2504
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.827736389684814e-05,
+      "loss": 0.0705,
+      "step": 2505
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.827667621776504e-05,
+      "loss": 0.0667,
+      "step": 2506
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.827598853868195e-05,
+      "loss": 0.1115,
+      "step": 2507
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.827530085959885e-05,
+      "loss": 0.0451,
+      "step": 2508
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.827461318051576e-05,
+      "loss": 0.0783,
+      "step": 2509
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.827392550143266e-05,
+      "loss": 0.0859,
+      "step": 2510
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.827323782234957e-05,
+      "loss": 0.0776,
+      "step": 2511
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.827255014326647e-05,
+      "loss": 0.0334,
+      "step": 2512
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.827186246418338e-05,
+      "loss": 0.066,
+      "step": 2513
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8271174785100284e-05,
+      "loss": 0.0371,
+      "step": 2514
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.827048710601719e-05,
+      "loss": 0.0743,
+      "step": 2515
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8269799426934095e-05,
+      "loss": 0.0379,
+      "step": 2516
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8269111747851004e-05,
+      "loss": 0.0795,
+      "step": 2517
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8268424068767906e-05,
+      "loss": 0.0693,
+      "step": 2518
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8267736389684815e-05,
+      "loss": 0.0611,
+      "step": 2519
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.826704871060172e-05,
+      "loss": 0.156,
+      "step": 2520
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.8266361031518626e-05,
+      "loss": 0.076,
+      "step": 2521
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.826567335243553e-05,
+      "loss": 0.0695,
+      "step": 2522
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.826498567335244e-05,
+      "loss": 0.0569,
+      "step": 2523
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.826429799426934e-05,
+      "loss": 0.0731,
+      "step": 2524
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.826361031518625e-05,
+      "loss": 0.0652,
+      "step": 2525
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.826292263610315e-05,
+      "loss": 0.0544,
+      "step": 2526
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.826223495702006e-05,
+      "loss": 0.0627,
+      "step": 2527
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.826154727793696e-05,
+      "loss": 0.0425,
+      "step": 2528
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.826085959885387e-05,
+      "loss": 0.0766,
+      "step": 2529
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.826017191977077e-05,
+      "loss": 0.0788,
+      "step": 2530
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.825948424068768e-05,
+      "loss": 0.0835,
+      "step": 2531
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.825879656160458e-05,
+      "loss": 0.0531,
+      "step": 2532
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.825810888252149e-05,
+      "loss": 0.0851,
+      "step": 2533
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8257421203438394e-05,
+      "loss": 0.0801,
+      "step": 2534
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.82567335243553e-05,
+      "loss": 0.0738,
+      "step": 2535
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8256045845272205e-05,
+      "loss": 0.0937,
+      "step": 2536
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8255358166189114e-05,
+      "loss": 0.1108,
+      "step": 2537
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.8254670487106016e-05,
+      "loss": 0.0834,
+      "step": 2538
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.8253982808022925e-05,
+      "loss": 0.0558,
+      "step": 2539
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.825329512893983e-05,
+      "loss": 0.0502,
+      "step": 2540
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.8252607449856736e-05,
+      "loss": 0.0658,
+      "step": 2541
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.825191977077364e-05,
+      "loss": 0.0779,
+      "step": 2542
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.825123209169055e-05,
+      "loss": 0.0673,
+      "step": 2543
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.825054441260745e-05,
+      "loss": 0.0681,
+      "step": 2544
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.824985673352436e-05,
+      "loss": 0.0886,
+      "step": 2545
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.824916905444126e-05,
+      "loss": 0.0697,
+      "step": 2546
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.824848137535817e-05,
+      "loss": 0.0558,
+      "step": 2547
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.824779369627507e-05,
+      "loss": 0.0903,
+      "step": 2548
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.824710601719198e-05,
+      "loss": 0.0675,
+      "step": 2549
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.824641833810888e-05,
+      "loss": 0.053,
+      "step": 2550
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.824573065902579e-05,
+      "loss": 0.0849,
+      "step": 2551
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.824504297994269e-05,
+      "loss": 0.1271,
+      "step": 2552
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.82443553008596e-05,
+      "loss": 0.0995,
+      "step": 2553
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.8243667621776504e-05,
+      "loss": 0.0856,
+      "step": 2554
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.824297994269341e-05,
+      "loss": 0.1149,
+      "step": 2555
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.8242292263610315e-05,
+      "loss": 0.0621,
+      "step": 2556
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.8241604584527224e-05,
+      "loss": 0.0633,
+      "step": 2557
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.8240916905444126e-05,
+      "loss": 0.0422,
+      "step": 2558
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.8240229226361035e-05,
+      "loss": 0.0612,
+      "step": 2559
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.823954154727794e-05,
+      "loss": 0.0374,
+      "step": 2560
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.8238853868194846e-05,
+      "loss": 0.0544,
+      "step": 2561
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.823816618911175e-05,
+      "loss": 0.0769,
+      "step": 2562
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.823747851002866e-05,
+      "loss": 0.0747,
+      "step": 2563
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.823679083094556e-05,
+      "loss": 0.0376,
+      "step": 2564
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.823610315186247e-05,
+      "loss": 0.0735,
+      "step": 2565
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.823541547277937e-05,
+      "loss": 0.0769,
+      "step": 2566
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.823472779369628e-05,
+      "loss": 0.0502,
+      "step": 2567
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.823404011461318e-05,
+      "loss": 0.0613,
+      "step": 2568
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.823335243553009e-05,
+      "loss": 0.0617,
+      "step": 2569
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.823266475644699e-05,
+      "loss": 0.1211,
+      "step": 2570
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.82319770773639e-05,
+      "loss": 0.0612,
+      "step": 2571
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.82312893982808e-05,
+      "loss": 0.0852,
+      "step": 2572
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.823060171919771e-05,
+      "loss": 0.1041,
+      "step": 2573
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.8229914040114614e-05,
+      "loss": 0.0424,
+      "step": 2574
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.822922636103152e-05,
+      "loss": 0.063,
+      "step": 2575
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.8228538681948425e-05,
+      "loss": 0.1013,
+      "step": 2576
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.8227851002865334e-05,
+      "loss": 0.1143,
+      "step": 2577
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.8227163323782236e-05,
+      "loss": 0.0318,
+      "step": 2578
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.8226475644699145e-05,
+      "loss": 0.0712,
+      "step": 2579
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.822578796561605e-05,
+      "loss": 0.0534,
+      "step": 2580
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.8225100286532956e-05,
+      "loss": 0.0647,
+      "step": 2581
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.822441260744986e-05,
+      "loss": 0.121,
+      "step": 2582
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.822372492836677e-05,
+      "loss": 0.0869,
+      "step": 2583
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.822303724928367e-05,
+      "loss": 0.0515,
+      "step": 2584
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.822234957020058e-05,
+      "loss": 0.0905,
+      "step": 2585
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.822166189111748e-05,
+      "loss": 0.0662,
+      "step": 2586
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.822097421203439e-05,
+      "loss": 0.1316,
+      "step": 2587
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.822028653295129e-05,
+      "loss": 0.0782,
+      "step": 2588
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.82195988538682e-05,
+      "loss": 0.0689,
+      "step": 2589
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.82189111747851e-05,
+      "loss": 0.0987,
+      "step": 2590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.821822349570201e-05,
+      "loss": 0.0611,
+      "step": 2591
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.821753581661891e-05,
+      "loss": 0.0996,
+      "step": 2592
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.821684813753582e-05,
+      "loss": 0.0614,
+      "step": 2593
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.8216160458452724e-05,
+      "loss": 0.0748,
+      "step": 2594
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.821547277936963e-05,
+      "loss": 0.055,
+      "step": 2595
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.8214785100286535e-05,
+      "loss": 0.0752,
+      "step": 2596
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.8214097421203444e-05,
+      "loss": 0.0346,
+      "step": 2597
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.8213409742120346e-05,
+      "loss": 0.0695,
+      "step": 2598
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.8212722063037255e-05,
+      "loss": 0.0825,
+      "step": 2599
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.821203438395416e-05,
+      "loss": 0.0863,
+      "step": 2600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.8211346704871066e-05,
+      "loss": 0.0698,
+      "step": 2601
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.821065902578797e-05,
+      "loss": 0.1189,
+      "step": 2602
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.820997134670488e-05,
+      "loss": 0.0783,
+      "step": 2603
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.820928366762178e-05,
+      "loss": 0.0539,
+      "step": 2604
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.820859598853869e-05,
+      "loss": 0.0765,
+      "step": 2605
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.820790830945559e-05,
+      "loss": 0.0682,
+      "step": 2606
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.82072206303725e-05,
+      "loss": 0.0621,
+      "step": 2607
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.82065329512894e-05,
+      "loss": 0.0651,
+      "step": 2608
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.820584527220631e-05,
+      "loss": 0.045,
+      "step": 2609
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.820515759312321e-05,
+      "loss": 0.0817,
+      "step": 2610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.820446991404012e-05,
+      "loss": 0.0787,
+      "step": 2611
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.820378223495702e-05,
+      "loss": 0.0721,
+      "step": 2612
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.820309455587393e-05,
+      "loss": 0.0672,
+      "step": 2613
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.8202406876790834e-05,
+      "loss": 0.0812,
+      "step": 2614
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.820171919770774e-05,
+      "loss": 0.0616,
+      "step": 2615
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.8201031518624645e-05,
+      "loss": 0.0438,
+      "step": 2616
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.820034383954155e-05,
+      "loss": 0.0383,
+      "step": 2617
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.8199656160458456e-05,
+      "loss": 0.1195,
+      "step": 2618
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.819896848137536e-05,
+      "loss": 0.0562,
+      "step": 2619
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.819828080229227e-05,
+      "loss": 0.0791,
+      "step": 2620
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.819759312320917e-05,
+      "loss": 0.0884,
+      "step": 2621
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.819690544412608e-05,
+      "loss": 0.0717,
+      "step": 2622
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.819621776504298e-05,
+      "loss": 0.0711,
+      "step": 2623
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.819553008595989e-05,
+      "loss": 0.0874,
+      "step": 2624
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.819484240687679e-05,
+      "loss": 0.1135,
+      "step": 2625
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.81941547277937e-05,
+      "loss": 0.0494,
+      "step": 2626
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.81934670487106e-05,
+      "loss": 0.0906,
+      "step": 2627
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.819277936962751e-05,
+      "loss": 0.092,
+      "step": 2628
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.819209169054441e-05,
+      "loss": 0.083,
+      "step": 2629
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.819140401146132e-05,
+      "loss": 0.1066,
+      "step": 2630
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.8190716332378224e-05,
+      "loss": 0.0625,
+      "step": 2631
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.819002865329513e-05,
+      "loss": 0.0902,
+      "step": 2632
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.8189340974212035e-05,
+      "loss": 0.0589,
+      "step": 2633
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.8188653295128944e-05,
+      "loss": 0.1472,
+      "step": 2634
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.8187965616045846e-05,
+      "loss": 0.0882,
+      "step": 2635
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.8187277936962755e-05,
+      "loss": 0.048,
+      "step": 2636
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.818659025787966e-05,
+      "loss": 0.0958,
+      "step": 2637
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.8185902578796566e-05,
+      "loss": 0.0576,
+      "step": 2638
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.818521489971347e-05,
+      "loss": 0.0866,
+      "step": 2639
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.818452722063038e-05,
+      "loss": 0.0687,
+      "step": 2640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.818383954154728e-05,
+      "loss": 0.1112,
+      "step": 2641
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.818315186246419e-05,
+      "loss": 0.1029,
+      "step": 2642
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 5.818246418338109e-05,
+      "loss": 0.0564,
+      "step": 2643
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.8181776504298e-05,
+      "loss": 0.0488,
+      "step": 2644
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.81810888252149e-05,
+      "loss": 0.0527,
+      "step": 2645
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.818040114613181e-05,
+      "loss": 0.0716,
+      "step": 2646
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.817971346704871e-05,
+      "loss": 0.0708,
+      "step": 2647
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.817902578796562e-05,
+      "loss": 0.0467,
+      "step": 2648
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.817833810888252e-05,
+      "loss": 0.0766,
+      "step": 2649
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.817765042979943e-05,
+      "loss": 0.0418,
+      "step": 2650
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.8176962750716334e-05,
+      "loss": 0.063,
+      "step": 2651
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.817627507163324e-05,
+      "loss": 0.0795,
+      "step": 2652
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.8175587392550145e-05,
+      "loss": 0.0622,
+      "step": 2653
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.8174899713467054e-05,
+      "loss": 0.0765,
+      "step": 2654
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.8174212034383956e-05,
+      "loss": 0.0814,
+      "step": 2655
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.8173524355300865e-05,
+      "loss": 0.0523,
+      "step": 2656
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.817283667621777e-05,
+      "loss": 0.0733,
+      "step": 2657
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.8172148997134676e-05,
+      "loss": 0.1158,
+      "step": 2658
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.817146131805158e-05,
+      "loss": 0.0704,
+      "step": 2659
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.817077363896849e-05,
+      "loss": 0.059,
+      "step": 2660
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 5.817008595988539e-05,
+      "loss": 0.0671,
+      "step": 2661
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.81693982808023e-05,
+      "loss": 0.0671,
+      "step": 2662
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.81687106017192e-05,
+      "loss": 0.0689,
+      "step": 2663
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.816802292263611e-05,
+      "loss": 0.0593,
+      "step": 2664
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.816733524355301e-05,
+      "loss": 0.0838,
+      "step": 2665
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.816664756446992e-05,
+      "loss": 0.1686,
+      "step": 2666
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.816595988538682e-05,
+      "loss": 0.0473,
+      "step": 2667
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.816527220630373e-05,
+      "loss": 0.0565,
+      "step": 2668
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.816458452722063e-05,
+      "loss": 0.0409,
+      "step": 2669
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.8163896848137535e-05,
+      "loss": 0.0456,
+      "step": 2670
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.8163209169054444e-05,
+      "loss": 0.078,
+      "step": 2671
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.8162521489971346e-05,
+      "loss": 0.0516,
+      "step": 2672
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.816183381088825e-05,
+      "loss": 0.0745,
+      "step": 2673
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.816114613180516e-05,
+      "loss": 0.0385,
+      "step": 2674
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.816045845272206e-05,
+      "loss": 0.0865,
+      "step": 2675
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.815977077363897e-05,
+      "loss": 0.0714,
+      "step": 2676
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.815908309455587e-05,
+      "loss": 0.0393,
+      "step": 2677
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 5.815839541547278e-05,
+      "loss": 0.0681,
+      "step": 2678
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.815770773638968e-05,
+      "loss": 0.053,
+      "step": 2679
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.815702005730659e-05,
+      "loss": 0.0436,
+      "step": 2680
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.815633237822349e-05,
+      "loss": 0.0919,
+      "step": 2681
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.81556446991404e-05,
+      "loss": 0.0995,
+      "step": 2682
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.8154957020057304e-05,
+      "loss": 0.0713,
+      "step": 2683
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.815426934097421e-05,
+      "loss": 0.1189,
+      "step": 2684
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.8153581661891115e-05,
+      "loss": 0.0364,
+      "step": 2685
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.8152893982808023e-05,
+      "loss": 0.0742,
+      "step": 2686
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.8152206303724926e-05,
+      "loss": 0.0911,
+      "step": 2687
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.8151518624641834e-05,
+      "loss": 0.0536,
+      "step": 2688
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.8150830945558737e-05,
+      "loss": 0.1097,
+      "step": 2689
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.8150143266475645e-05,
+      "loss": 0.0482,
+      "step": 2690
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.814945558739255e-05,
+      "loss": 0.0652,
+      "step": 2691
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.8148767908309456e-05,
+      "loss": 0.0877,
+      "step": 2692
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.814808022922636e-05,
+      "loss": 0.0514,
+      "step": 2693
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.814739255014327e-05,
+      "loss": 0.0634,
+      "step": 2694
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.814670487106017e-05,
+      "loss": 0.0787,
+      "step": 2695
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.814601719197708e-05,
+      "loss": 0.0643,
+      "step": 2696
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.814532951289398e-05,
+      "loss": 0.0731,
+      "step": 2697
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.814464183381089e-05,
+      "loss": 0.0914,
+      "step": 2698
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.814395415472779e-05,
+      "loss": 0.0742,
+      "step": 2699
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.81432664756447e-05,
+      "loss": 0.1604,
+      "step": 2700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.81425787965616e-05,
+      "loss": 0.047,
+      "step": 2701
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.814189111747851e-05,
+      "loss": 0.0559,
+      "step": 2702
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.8141203438395414e-05,
+      "loss": 0.0534,
+      "step": 2703
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.814051575931232e-05,
+      "loss": 0.0589,
+      "step": 2704
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.8139828080229225e-05,
+      "loss": 0.063,
+      "step": 2705
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.8139140401146133e-05,
+      "loss": 0.0607,
+      "step": 2706
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.8138452722063036e-05,
+      "loss": 0.0592,
+      "step": 2707
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.8137765042979944e-05,
+      "loss": 0.1273,
+      "step": 2708
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.8137077363896847e-05,
+      "loss": 0.1981,
+      "step": 2709
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.8136389684813755e-05,
+      "loss": 0.0707,
+      "step": 2710
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.813570200573066e-05,
+      "loss": 0.0588,
+      "step": 2711
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.8135014326647566e-05,
+      "loss": 0.0681,
+      "step": 2712
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 5.813432664756447e-05,
+      "loss": 0.0577,
+      "step": 2713
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.813363896848138e-05,
+      "loss": 0.0461,
+      "step": 2714
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.813295128939828e-05,
+      "loss": 0.0769,
+      "step": 2715
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.813226361031519e-05,
+      "loss": 0.1223,
+      "step": 2716
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.813157593123209e-05,
+      "loss": 0.0764,
+      "step": 2717
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.8130888252149e-05,
+      "loss": 0.0951,
+      "step": 2718
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.81302005730659e-05,
+      "loss": 0.0859,
+      "step": 2719
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.812951289398281e-05,
+      "loss": 0.0642,
+      "step": 2720
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.812882521489971e-05,
+      "loss": 0.0936,
+      "step": 2721
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.812813753581662e-05,
+      "loss": 0.1043,
+      "step": 2722
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.8127449856733524e-05,
+      "loss": 0.0758,
+      "step": 2723
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.812676217765043e-05,
+      "loss": 0.0592,
+      "step": 2724
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.8126074498567335e-05,
+      "loss": 0.0366,
+      "step": 2725
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.8125386819484243e-05,
+      "loss": 0.1417,
+      "step": 2726
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.8124699140401146e-05,
+      "loss": 0.0518,
+      "step": 2727
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.8124011461318054e-05,
+      "loss": 0.0657,
+      "step": 2728
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.8123323782234957e-05,
+      "loss": 0.1092,
+      "step": 2729
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 5.8122636103151865e-05,
+      "loss": 0.1179,
+      "step": 2730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.812194842406877e-05,
+      "loss": 0.0565,
+      "step": 2731
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8121260744985676e-05,
+      "loss": 0.0427,
+      "step": 2732
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.812057306590258e-05,
+      "loss": 0.0751,
+      "step": 2733
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.811988538681948e-05,
+      "loss": 0.0929,
+      "step": 2734
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.811919770773639e-05,
+      "loss": 0.0601,
+      "step": 2735
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.811851002865329e-05,
+      "loss": 0.0889,
+      "step": 2736
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.81178223495702e-05,
+      "loss": 0.0622,
+      "step": 2737
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.81171346704871e-05,
+      "loss": 0.0678,
+      "step": 2738
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.811644699140401e-05,
+      "loss": 0.0513,
+      "step": 2739
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8115759312320914e-05,
+      "loss": 0.1094,
+      "step": 2740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.811507163323782e-05,
+      "loss": 0.0367,
+      "step": 2741
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8114383954154725e-05,
+      "loss": 0.0475,
+      "step": 2742
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8113696275071633e-05,
+      "loss": 0.0832,
+      "step": 2743
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8113008595988536e-05,
+      "loss": 0.0425,
+      "step": 2744
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8112320916905444e-05,
+      "loss": 0.0842,
+      "step": 2745
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8111633237822347e-05,
+      "loss": 0.0708,
+      "step": 2746
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.8110945558739255e-05,
+      "loss": 0.0888,
+      "step": 2747
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 5.811025787965616e-05,
+      "loss": 0.0597,
+      "step": 2748
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.8109570200573066e-05,
+      "loss": 0.0788,
+      "step": 2749
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.810888252148997e-05,
+      "loss": 0.0508,
+      "step": 2750
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.810819484240688e-05,
+      "loss": 0.058,
+      "step": 2751
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.810750716332378e-05,
+      "loss": 0.0534,
+      "step": 2752
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.810681948424069e-05,
+      "loss": 0.0561,
+      "step": 2753
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.810613180515759e-05,
+      "loss": 0.038,
+      "step": 2754
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.81054441260745e-05,
+      "loss": 0.075,
+      "step": 2755
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.81047564469914e-05,
+      "loss": 0.0544,
+      "step": 2756
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.810406876790831e-05,
+      "loss": 0.0541,
+      "step": 2757
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.810338108882521e-05,
+      "loss": 0.0608,
+      "step": 2758
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.810269340974212e-05,
+      "loss": 0.0669,
+      "step": 2759
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.8102005730659024e-05,
+      "loss": 0.0771,
+      "step": 2760
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.810131805157593e-05,
+      "loss": 0.0453,
+      "step": 2761
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.8100630372492835e-05,
+      "loss": 0.074,
+      "step": 2762
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.8099942693409743e-05,
+      "loss": 0.0954,
+      "step": 2763
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.8099255014326646e-05,
+      "loss": 0.0731,
+      "step": 2764
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 5.8098567335243554e-05,
+      "loss": 0.0962,
+      "step": 2765
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8097879656160457e-05,
+      "loss": 0.0821,
+      "step": 2766
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8097191977077365e-05,
+      "loss": 0.0672,
+      "step": 2767
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.809650429799427e-05,
+      "loss": 0.0504,
+      "step": 2768
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8095816618911176e-05,
+      "loss": 0.0599,
+      "step": 2769
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.809512893982808e-05,
+      "loss": 0.0472,
+      "step": 2770
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.809444126074499e-05,
+      "loss": 0.0641,
+      "step": 2771
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.809375358166189e-05,
+      "loss": 0.0543,
+      "step": 2772
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.80930659025788e-05,
+      "loss": 0.0603,
+      "step": 2773
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.80923782234957e-05,
+      "loss": 0.0643,
+      "step": 2774
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.809169054441261e-05,
+      "loss": 0.0346,
+      "step": 2775
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.809100286532951e-05,
+      "loss": 0.0544,
+      "step": 2776
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.809031518624642e-05,
+      "loss": 0.0625,
+      "step": 2777
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.808962750716332e-05,
+      "loss": 0.0745,
+      "step": 2778
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.808893982808023e-05,
+      "loss": 0.0434,
+      "step": 2779
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8088252148997134e-05,
+      "loss": 0.1072,
+      "step": 2780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.808756446991404e-05,
+      "loss": 0.054,
+      "step": 2781
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8086876790830945e-05,
+      "loss": 0.0741,
+      "step": 2782
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 5.8086189111747853e-05,
+      "loss": 0.0467,
+      "step": 2783
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.8085501432664756e-05,
+      "loss": 0.0512,
+      "step": 2784
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.8084813753581664e-05,
+      "loss": 0.1059,
+      "step": 2785
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.8084126074498567e-05,
+      "loss": 0.0405,
+      "step": 2786
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.8083438395415475e-05,
+      "loss": 0.0999,
+      "step": 2787
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.808275071633238e-05,
+      "loss": 0.0938,
+      "step": 2788
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.8082063037249286e-05,
+      "loss": 0.0839,
+      "step": 2789
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.808137535816619e-05,
+      "loss": 0.1153,
+      "step": 2790
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.80806876790831e-05,
+      "loss": 0.1182,
+      "step": 2791
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.808e-05,
+      "loss": 0.0449,
+      "step": 2792
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.807931232091691e-05,
+      "loss": 0.0604,
+      "step": 2793
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.807862464183381e-05,
+      "loss": 0.0708,
+      "step": 2794
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.807793696275072e-05,
+      "loss": 0.0827,
+      "step": 2795
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.807724928366762e-05,
+      "loss": 0.0865,
+      "step": 2796
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.807656160458453e-05,
+      "loss": 0.0404,
+      "step": 2797
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.807587392550143e-05,
+      "loss": 0.0848,
+      "step": 2798
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.807518624641834e-05,
+      "loss": 0.0645,
+      "step": 2799
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 5.8074498567335244e-05,
+      "loss": 0.1089,
+      "step": 2800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.807381088825215e-05,
+      "loss": 0.0638,
+      "step": 2801
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8073123209169055e-05,
+      "loss": 0.0629,
+      "step": 2802
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8072435530085963e-05,
+      "loss": 0.0715,
+      "step": 2803
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8071747851002866e-05,
+      "loss": 0.0466,
+      "step": 2804
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8071060171919774e-05,
+      "loss": 0.0619,
+      "step": 2805
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8070372492836677e-05,
+      "loss": 0.1089,
+      "step": 2806
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8069684813753585e-05,
+      "loss": 0.0783,
+      "step": 2807
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.806899713467049e-05,
+      "loss": 0.0506,
+      "step": 2808
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.8068309455587396e-05,
+      "loss": 0.0783,
+      "step": 2809
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.80676217765043e-05,
+      "loss": 0.055,
+      "step": 2810
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.806693409742121e-05,
+      "loss": 0.0521,
+      "step": 2811
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.806624641833811e-05,
+      "loss": 0.0765,
+      "step": 2812
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.806555873925502e-05,
+      "loss": 0.0586,
+      "step": 2813
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.806487106017192e-05,
+      "loss": 0.1035,
+      "step": 2814
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.806418338108883e-05,
+      "loss": 0.0512,
+      "step": 2815
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.806349570200573e-05,
+      "loss": 0.041,
+      "step": 2816
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.806280802292264e-05,
+      "loss": 0.0613,
+      "step": 2817
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 5.806212034383954e-05,
+      "loss": 0.0481,
+      "step": 2818
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.806143266475645e-05,
+      "loss": 0.0458,
+      "step": 2819
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.8060744985673354e-05,
+      "loss": 0.0628,
+      "step": 2820
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.806005730659026e-05,
+      "loss": 0.0617,
+      "step": 2821
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.8059369627507165e-05,
+      "loss": 0.0746,
+      "step": 2822
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.8058681948424073e-05,
+      "loss": 0.0687,
+      "step": 2823
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.8057994269340976e-05,
+      "loss": 0.1321,
+      "step": 2824
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.8057306590257884e-05,
+      "loss": 0.0623,
+      "step": 2825
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.8056618911174787e-05,
+      "loss": 0.0372,
+      "step": 2826
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.8055931232091695e-05,
+      "loss": 0.0291,
+      "step": 2827
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.80552435530086e-05,
+      "loss": 0.0516,
+      "step": 2828
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.8054555873925506e-05,
+      "loss": 0.168,
+      "step": 2829
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.805386819484241e-05,
+      "loss": 0.0648,
+      "step": 2830
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.805318051575932e-05,
+      "loss": 0.092,
+      "step": 2831
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.805249283667622e-05,
+      "loss": 0.1177,
+      "step": 2832
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.805180515759313e-05,
+      "loss": 0.0498,
+      "step": 2833
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.805111747851003e-05,
+      "loss": 0.087,
+      "step": 2834
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 5.805042979942694e-05,
+      "loss": 0.0368,
+      "step": 2835
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.804974212034384e-05,
+      "loss": 0.0583,
+      "step": 2836
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.804905444126075e-05,
+      "loss": 0.0638,
+      "step": 2837
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.804836676217765e-05,
+      "loss": 0.0545,
+      "step": 2838
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.804767908309456e-05,
+      "loss": 0.0333,
+      "step": 2839
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.8046991404011464e-05,
+      "loss": 0.0612,
+      "step": 2840
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.804630372492837e-05,
+      "loss": 0.0633,
+      "step": 2841
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.8045616045845275e-05,
+      "loss": 0.0382,
+      "step": 2842
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.8044928366762183e-05,
+      "loss": 0.0409,
+      "step": 2843
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.8044240687679086e-05,
+      "loss": 0.0628,
+      "step": 2844
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.8043553008595994e-05,
+      "loss": 0.0537,
+      "step": 2845
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.8042865329512897e-05,
+      "loss": 0.1208,
+      "step": 2846
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.8042177650429805e-05,
+      "loss": 0.0566,
+      "step": 2847
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.804148997134671e-05,
+      "loss": 0.1712,
+      "step": 2848
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.8040802292263616e-05,
+      "loss": 0.0828,
+      "step": 2849
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.804011461318052e-05,
+      "loss": 0.0731,
+      "step": 2850
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.803942693409743e-05,
+      "loss": 0.094,
+      "step": 2851
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.803873925501433e-05,
+      "loss": 0.0903,
+      "step": 2852
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 5.803805157593123e-05,
+      "loss": 0.0643,
+      "step": 2853
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.803736389684814e-05,
+      "loss": 0.1127,
+      "step": 2854
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.803667621776504e-05,
+      "loss": 0.0604,
+      "step": 2855
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.803598853868195e-05,
+      "loss": 0.0453,
+      "step": 2856
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.8035300859598854e-05,
+      "loss": 0.087,
+      "step": 2857
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.803461318051576e-05,
+      "loss": 0.095,
+      "step": 2858
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.8033925501432665e-05,
+      "loss": 0.0768,
+      "step": 2859
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.8033237822349574e-05,
+      "loss": 0.0851,
+      "step": 2860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.8032550143266476e-05,
+      "loss": 0.0681,
+      "step": 2861
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.8031862464183385e-05,
+      "loss": 0.0535,
+      "step": 2862
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.803117478510029e-05,
+      "loss": 0.0765,
+      "step": 2863
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.8030487106017196e-05,
+      "loss": 0.0525,
+      "step": 2864
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.80297994269341e-05,
+      "loss": 0.0564,
+      "step": 2865
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.8029111747851007e-05,
+      "loss": 0.0615,
+      "step": 2866
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.802842406876791e-05,
+      "loss": 0.0625,
+      "step": 2867
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.802773638968482e-05,
+      "loss": 0.1011,
+      "step": 2868
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.802704871060172e-05,
+      "loss": 0.0505,
+      "step": 2869
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 5.802636103151863e-05,
+      "loss": 0.0773,
+      "step": 2870
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.802567335243553e-05,
+      "loss": 0.0551,
+      "step": 2871
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.802498567335244e-05,
+      "loss": 0.0778,
+      "step": 2872
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.802429799426934e-05,
+      "loss": 0.0467,
+      "step": 2873
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.802361031518625e-05,
+      "loss": 0.0521,
+      "step": 2874
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.802292263610315e-05,
+      "loss": 0.0247,
+      "step": 2875
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.802223495702006e-05,
+      "loss": 0.0786,
+      "step": 2876
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8021547277936964e-05,
+      "loss": 0.1034,
+      "step": 2877
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.802085959885387e-05,
+      "loss": 0.1715,
+      "step": 2878
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8020171919770775e-05,
+      "loss": 0.0546,
+      "step": 2879
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8019484240687684e-05,
+      "loss": 0.0843,
+      "step": 2880
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8018796561604586e-05,
+      "loss": 0.0934,
+      "step": 2881
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8018108882521495e-05,
+      "loss": 0.0451,
+      "step": 2882
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.80174212034384e-05,
+      "loss": 0.1032,
+      "step": 2883
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8016733524355306e-05,
+      "loss": 0.0468,
+      "step": 2884
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.801604584527221e-05,
+      "loss": 0.0662,
+      "step": 2885
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.8015358166189117e-05,
+      "loss": 0.0673,
+      "step": 2886
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 5.801467048710602e-05,
+      "loss": 0.1206,
+      "step": 2887
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.801398280802293e-05,
+      "loss": 0.0685,
+      "step": 2888
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.801329512893983e-05,
+      "loss": 0.0596,
+      "step": 2889
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.801260744985674e-05,
+      "loss": 0.0588,
+      "step": 2890
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.801191977077364e-05,
+      "loss": 0.0555,
+      "step": 2891
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.801123209169055e-05,
+      "loss": 0.0929,
+      "step": 2892
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.801054441260745e-05,
+      "loss": 0.0701,
+      "step": 2893
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.800985673352436e-05,
+      "loss": 0.1049,
+      "step": 2894
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.800916905444126e-05,
+      "loss": 0.0544,
+      "step": 2895
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.800848137535817e-05,
+      "loss": 0.0593,
+      "step": 2896
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8007793696275074e-05,
+      "loss": 0.0531,
+      "step": 2897
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.800710601719198e-05,
+      "loss": 0.1185,
+      "step": 2898
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8006418338108885e-05,
+      "loss": 0.0798,
+      "step": 2899
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8005730659025793e-05,
+      "loss": 0.0935,
+      "step": 2900
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8005042979942696e-05,
+      "loss": 0.0602,
+      "step": 2901
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8004355300859604e-05,
+      "loss": 0.1159,
+      "step": 2902
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8003667621776507e-05,
+      "loss": 0.0422,
+      "step": 2903
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.8002979942693415e-05,
+      "loss": 0.0874,
+      "step": 2904
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 5.800229226361032e-05,
+      "loss": 0.0352,
+      "step": 2905
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.8001604584527226e-05,
+      "loss": 0.0831,
+      "step": 2906
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.800091690544413e-05,
+      "loss": 0.0608,
+      "step": 2907
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.800022922636104e-05,
+      "loss": 0.0816,
+      "step": 2908
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.799954154727794e-05,
+      "loss": 0.0741,
+      "step": 2909
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.799885386819485e-05,
+      "loss": 0.0888,
+      "step": 2910
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.799816618911175e-05,
+      "loss": 0.0855,
+      "step": 2911
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.799747851002866e-05,
+      "loss": 0.1291,
+      "step": 2912
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.799679083094556e-05,
+      "loss": 0.038,
+      "step": 2913
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.799610315186247e-05,
+      "loss": 0.0505,
+      "step": 2914
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.799541547277937e-05,
+      "loss": 0.0459,
+      "step": 2915
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.799472779369628e-05,
+      "loss": 0.0998,
+      "step": 2916
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7994040114613184e-05,
+      "loss": 0.0516,
+      "step": 2917
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.799335243553009e-05,
+      "loss": 0.0355,
+      "step": 2918
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7992664756446995e-05,
+      "loss": 0.0536,
+      "step": 2919
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7991977077363903e-05,
+      "loss": 0.081,
+      "step": 2920
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7991289398280806e-05,
+      "loss": 0.068,
+      "step": 2921
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 5.7990601719197714e-05,
+      "loss": 0.0998,
+      "step": 2922
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7989914040114617e-05,
+      "loss": 0.0519,
+      "step": 2923
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7989226361031525e-05,
+      "loss": 0.0714,
+      "step": 2924
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.798853868194843e-05,
+      "loss": 0.0284,
+      "step": 2925
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.798785100286533e-05,
+      "loss": 0.0962,
+      "step": 2926
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.798716332378223e-05,
+      "loss": 0.0571,
+      "step": 2927
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.798647564469914e-05,
+      "loss": 0.0542,
+      "step": 2928
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.798578796561604e-05,
+      "loss": 0.047,
+      "step": 2929
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.798510028653295e-05,
+      "loss": 0.0488,
+      "step": 2930
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7984412607449854e-05,
+      "loss": 0.0563,
+      "step": 2931
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.798372492836676e-05,
+      "loss": 0.0779,
+      "step": 2932
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7983037249283665e-05,
+      "loss": 0.0466,
+      "step": 2933
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7982349570200574e-05,
+      "loss": 0.0904,
+      "step": 2934
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7981661891117476e-05,
+      "loss": 0.0476,
+      "step": 2935
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7980974212034385e-05,
+      "loss": 0.0865,
+      "step": 2936
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.798028653295129e-05,
+      "loss": 0.0761,
+      "step": 2937
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.7979598853868196e-05,
+      "loss": 0.0579,
+      "step": 2938
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.79789111747851e-05,
+      "loss": 0.0976,
+      "step": 2939
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 5.797822349570201e-05,
+      "loss": 0.0531,
+      "step": 2940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.797753581661891e-05,
+      "loss": 0.0458,
+      "step": 2941
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.797684813753582e-05,
+      "loss": 0.0613,
+      "step": 2942
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.797616045845272e-05,
+      "loss": 0.1133,
+      "step": 2943
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.797547277936963e-05,
+      "loss": 0.0654,
+      "step": 2944
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.797478510028653e-05,
+      "loss": 0.0482,
+      "step": 2945
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.797409742120344e-05,
+      "loss": 0.0426,
+      "step": 2946
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.797340974212034e-05,
+      "loss": 0.0688,
+      "step": 2947
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.797272206303725e-05,
+      "loss": 0.061,
+      "step": 2948
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.797203438395415e-05,
+      "loss": 0.1063,
+      "step": 2949
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.797134670487106e-05,
+      "loss": 0.1148,
+      "step": 2950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.7970659025787964e-05,
+      "loss": 0.0565,
+      "step": 2951
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.796997134670487e-05,
+      "loss": 0.0943,
+      "step": 2952
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.7969283667621775e-05,
+      "loss": 0.109,
+      "step": 2953
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.7968595988538684e-05,
+      "loss": 0.0601,
+      "step": 2954
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.7967908309455586e-05,
+      "loss": 0.0551,
+      "step": 2955
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.7967220630372495e-05,
+      "loss": 0.1049,
+      "step": 2956
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 5.79665329512894e-05,
+      "loss": 0.0368,
+      "step": 2957
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.7965845272206306e-05,
+      "loss": 0.055,
+      "step": 2958
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.796515759312321e-05,
+      "loss": 0.1003,
+      "step": 2959
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.796446991404012e-05,
+      "loss": 0.0347,
+      "step": 2960
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.796378223495702e-05,
+      "loss": 0.1431,
+      "step": 2961
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.796309455587393e-05,
+      "loss": 0.0715,
+      "step": 2962
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.796240687679083e-05,
+      "loss": 0.0839,
+      "step": 2963
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.796171919770774e-05,
+      "loss": 0.0474,
+      "step": 2964
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.796103151862464e-05,
+      "loss": 0.0529,
+      "step": 2965
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.796034383954155e-05,
+      "loss": 0.0727,
+      "step": 2966
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.795965616045845e-05,
+      "loss": 0.0784,
+      "step": 2967
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.795896848137536e-05,
+      "loss": 0.0521,
+      "step": 2968
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.795828080229226e-05,
+      "loss": 0.0993,
+      "step": 2969
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.795759312320917e-05,
+      "loss": 0.0883,
+      "step": 2970
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.7956905444126074e-05,
+      "loss": 0.0889,
+      "step": 2971
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.7956217765042976e-05,
+      "loss": 0.0843,
+      "step": 2972
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.7955530085959885e-05,
+      "loss": 0.1414,
+      "step": 2973
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.795484240687679e-05,
+      "loss": 0.0763,
+      "step": 2974
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.7954154727793696e-05,
+      "loss": 0.0607,
+      "step": 2975
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.79534670487106e-05,
+      "loss": 0.0525,
+      "step": 2976
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.795277936962751e-05,
+      "loss": 0.2166,
+      "step": 2977
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.795209169054441e-05,
+      "loss": 0.0532,
+      "step": 2978
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.795140401146132e-05,
+      "loss": 0.083,
+      "step": 2979
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.795071633237822e-05,
+      "loss": 0.077,
+      "step": 2980
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.795002865329513e-05,
+      "loss": 0.0758,
+      "step": 2981
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.794934097421203e-05,
+      "loss": 0.0568,
+      "step": 2982
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.794865329512894e-05,
+      "loss": 0.0685,
+      "step": 2983
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.794796561604584e-05,
+      "loss": 0.0684,
+      "step": 2984
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.794727793696275e-05,
+      "loss": 0.0413,
+      "step": 2985
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.794659025787965e-05,
+      "loss": 0.0612,
+      "step": 2986
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.794590257879656e-05,
+      "loss": 0.0746,
+      "step": 2987
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.7945214899713464e-05,
+      "loss": 0.0558,
+      "step": 2988
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.794452722063037e-05,
+      "loss": 0.0631,
+      "step": 2989
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.7943839541547275e-05,
+      "loss": 0.0801,
+      "step": 2990
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.7943151862464184e-05,
+      "loss": 0.0444,
+      "step": 2991
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.7942464183381086e-05,
+      "loss": 0.0627,
+      "step": 2992
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.7941776504297995e-05,
+      "loss": 0.0977,
+      "step": 2993
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.79410888252149e-05,
+      "loss": 0.0501,
+      "step": 2994
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.7940401146131806e-05,
+      "loss": 0.1277,
+      "step": 2995
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793971346704871e-05,
+      "loss": 0.0812,
+      "step": 2996
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793902578796562e-05,
+      "loss": 0.0806,
+      "step": 2997
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793833810888252e-05,
+      "loss": 0.0757,
+      "step": 2998
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793765042979943e-05,
+      "loss": 0.0588,
+      "step": 2999
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793696275071633e-05,
+      "loss": 0.0504,
+      "step": 3000
+    },
+    {
+      "epoch": 1.72,
+      "eval_loss": 0.052984319627285004,
+      "eval_mean_accuracy": 0.7898752824289966,
+      "eval_mean_iou": 0.7030777943249182,
+      "eval_overall_accuracy": 0.9820018439822726,
+      "eval_per_category_accuracy": [
+        0.994965135511411,
+        0.7297640212863019,
+        0.8141417124860444,
+        0.6206302604322292
+      ],
+      "eval_per_category_iou": [
+        0.9828001146369459,
+        0.5928329085496412,
+        0.6981904195778292,
+        0.5384877345352564
+      ],
+      "eval_runtime": 263.8455,
+      "eval_samples_per_second": 13.644,
+      "eval_steps_per_second": 1.706,
+      "step": 3000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793627507163324e-05,
+      "loss": 0.0555,
+      "step": 3001
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793558739255014e-05,
+      "loss": 0.0697,
+      "step": 3002
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793489971346705e-05,
+      "loss": 0.0651,
+      "step": 3003
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793421203438395e-05,
+      "loss": 0.0514,
+      "step": 3004
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793352435530086e-05,
+      "loss": 0.0699,
+      "step": 3005
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793283667621776e-05,
+      "loss": 0.099,
+      "step": 3006
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793214899713467e-05,
+      "loss": 0.0664,
+      "step": 3007
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.7931461318051574e-05,
+      "loss": 0.0499,
+      "step": 3008
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.793077363896848e-05,
+      "loss": 0.0686,
+      "step": 3009
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.7930085959885385e-05,
+      "loss": 0.049,
+      "step": 3010
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.7929398280802294e-05,
+      "loss": 0.0678,
+      "step": 3011
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.7928710601719196e-05,
+      "loss": 0.0459,
+      "step": 3012
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.7928022922636105e-05,
+      "loss": 0.094,
+      "step": 3013
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.792733524355301e-05,
+      "loss": 0.0453,
+      "step": 3014
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.7926647564469916e-05,
+      "loss": 0.0519,
+      "step": 3015
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.792595988538682e-05,
+      "loss": 0.0587,
+      "step": 3016
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.792527220630373e-05,
+      "loss": 0.054,
+      "step": 3017
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.792458452722063e-05,
+      "loss": 0.0727,
+      "step": 3018
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.792389684813754e-05,
+      "loss": 0.0705,
+      "step": 3019
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.792320916905444e-05,
+      "loss": 0.127,
+      "step": 3020
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.792252148997135e-05,
+      "loss": 0.0679,
+      "step": 3021
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.792183381088825e-05,
+      "loss": 0.0893,
+      "step": 3022
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.792114613180516e-05,
+      "loss": 0.0427,
+      "step": 3023
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.792045845272206e-05,
+      "loss": 0.0748,
+      "step": 3024
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.791977077363897e-05,
+      "loss": 0.0718,
+      "step": 3025
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.791908309455587e-05,
+      "loss": 0.1167,
+      "step": 3026
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.791839541547278e-05,
+      "loss": 0.0784,
+      "step": 3027
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.7917707736389684e-05,
+      "loss": 0.1039,
+      "step": 3028
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.791702005730659e-05,
+      "loss": 0.1373,
+      "step": 3029
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.7916332378223495e-05,
+      "loss": 0.0481,
+      "step": 3030
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.7915644699140404e-05,
+      "loss": 0.0745,
+      "step": 3031
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.7914957020057306e-05,
+      "loss": 0.0521,
+      "step": 3032
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.7914269340974215e-05,
+      "loss": 0.0679,
+      "step": 3033
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.791358166189112e-05,
+      "loss": 0.0431,
+      "step": 3034
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.7912893982808026e-05,
+      "loss": 0.0697,
+      "step": 3035
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.791220630372493e-05,
+      "loss": 0.0449,
+      "step": 3036
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.791151862464184e-05,
+      "loss": 0.0883,
+      "step": 3037
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.791083094555874e-05,
+      "loss": 0.0701,
+      "step": 3038
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.791014326647565e-05,
+      "loss": 0.0494,
+      "step": 3039
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.790945558739255e-05,
+      "loss": 0.0592,
+      "step": 3040
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.790876790830946e-05,
+      "loss": 0.0831,
+      "step": 3041
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.790808022922636e-05,
+      "loss": 0.1009,
+      "step": 3042
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.790739255014327e-05,
+      "loss": 0.106,
+      "step": 3043
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.790670487106017e-05,
+      "loss": 0.0524,
+      "step": 3044
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.790601719197708e-05,
+      "loss": 0.0342,
+      "step": 3045
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.790532951289398e-05,
+      "loss": 0.0354,
+      "step": 3046
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.790464183381089e-05,
+      "loss": 0.1083,
+      "step": 3047
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.7903954154727794e-05,
+      "loss": 0.0512,
+      "step": 3048
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.79032664756447e-05,
+      "loss": 0.0632,
+      "step": 3049
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.7902578796561605e-05,
+      "loss": 0.0697,
+      "step": 3050
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.7901891117478514e-05,
+      "loss": 0.0349,
+      "step": 3051
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.7901203438395416e-05,
+      "loss": 0.0782,
+      "step": 3052
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.7900515759312325e-05,
+      "loss": 0.0762,
+      "step": 3053
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.789982808022923e-05,
+      "loss": 0.0606,
+      "step": 3054
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.7899140401146136e-05,
+      "loss": 0.0361,
+      "step": 3055
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.789845272206304e-05,
+      "loss": 0.0801,
+      "step": 3056
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.789776504297995e-05,
+      "loss": 0.0532,
+      "step": 3057
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.789707736389685e-05,
+      "loss": 0.0473,
+      "step": 3058
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.789638968481376e-05,
+      "loss": 0.0767,
+      "step": 3059
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.789570200573066e-05,
+      "loss": 0.0756,
+      "step": 3060
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.789501432664757e-05,
+      "loss": 0.0409,
+      "step": 3061
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.789432664756447e-05,
+      "loss": 0.0835,
+      "step": 3062
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.789363896848138e-05,
+      "loss": 0.0781,
+      "step": 3063
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.789295128939828e-05,
+      "loss": 0.073,
+      "step": 3064
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.789226361031519e-05,
+      "loss": 0.0503,
+      "step": 3065
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.789157593123209e-05,
+      "loss": 0.0775,
+      "step": 3066
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.7890888252149e-05,
+      "loss": 0.1086,
+      "step": 3067
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.7890200573065904e-05,
+      "loss": 0.074,
+      "step": 3068
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.788951289398281e-05,
+      "loss": 0.0516,
+      "step": 3069
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.7888825214899715e-05,
+      "loss": 0.0419,
+      "step": 3070
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.7888137535816624e-05,
+      "loss": 0.066,
+      "step": 3071
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.7887449856733526e-05,
+      "loss": 0.0932,
+      "step": 3072
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.7886762177650435e-05,
+      "loss": 0.0612,
+      "step": 3073
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.788607449856734e-05,
+      "loss": 0.0541,
+      "step": 3074
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.7885386819484246e-05,
+      "loss": 0.0838,
+      "step": 3075
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.788469914040115e-05,
+      "loss": 0.074,
+      "step": 3076
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.788401146131806e-05,
+      "loss": 0.0854,
+      "step": 3077
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.788332378223496e-05,
+      "loss": 0.0543,
+      "step": 3078
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 5.788263610315187e-05,
+      "loss": 0.0511,
+      "step": 3079
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.788194842406877e-05,
+      "loss": 0.0458,
+      "step": 3080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.788126074498568e-05,
+      "loss": 0.0537,
+      "step": 3081
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.788057306590258e-05,
+      "loss": 0.0558,
+      "step": 3082
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.787988538681949e-05,
+      "loss": 0.0509,
+      "step": 3083
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.787919770773639e-05,
+      "loss": 0.0631,
+      "step": 3084
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.78785100286533e-05,
+      "loss": 0.0912,
+      "step": 3085
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.78778223495702e-05,
+      "loss": 0.0973,
+      "step": 3086
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.787713467048711e-05,
+      "loss": 0.085,
+      "step": 3087
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.7876446991404014e-05,
+      "loss": 0.0297,
+      "step": 3088
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.787575931232092e-05,
+      "loss": 0.0773,
+      "step": 3089
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.7875071633237825e-05,
+      "loss": 0.1117,
+      "step": 3090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.787438395415473e-05,
+      "loss": 0.0317,
+      "step": 3091
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.7873696275071636e-05,
+      "loss": 0.0728,
+      "step": 3092
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.787300859598854e-05,
+      "loss": 0.0357,
+      "step": 3093
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.787232091690545e-05,
+      "loss": 0.0589,
+      "step": 3094
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.787163323782235e-05,
+      "loss": 0.0579,
+      "step": 3095
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.787094555873926e-05,
+      "loss": 0.0905,
+      "step": 3096
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 5.787025787965616e-05,
+      "loss": 0.0407,
+      "step": 3097
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.786957020057307e-05,
+      "loss": 0.0562,
+      "step": 3098
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.786888252148997e-05,
+      "loss": 0.0666,
+      "step": 3099
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.786819484240688e-05,
+      "loss": 0.0913,
+      "step": 3100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.786750716332378e-05,
+      "loss": 0.0724,
+      "step": 3101
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.786681948424069e-05,
+      "loss": 0.0745,
+      "step": 3102
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.786613180515759e-05,
+      "loss": 0.0636,
+      "step": 3103
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.78654441260745e-05,
+      "loss": 0.048,
+      "step": 3104
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.7864756446991404e-05,
+      "loss": 0.0564,
+      "step": 3105
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.786406876790831e-05,
+      "loss": 0.0472,
+      "step": 3106
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.7863381088825215e-05,
+      "loss": 0.0443,
+      "step": 3107
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.7862693409742124e-05,
+      "loss": 0.086,
+      "step": 3108
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.7862005730659026e-05,
+      "loss": 0.0783,
+      "step": 3109
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.7861318051575935e-05,
+      "loss": 0.1047,
+      "step": 3110
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.786063037249284e-05,
+      "loss": 0.0488,
+      "step": 3111
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.7859942693409746e-05,
+      "loss": 0.0583,
+      "step": 3112
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.785925501432665e-05,
+      "loss": 0.0764,
+      "step": 3113
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 5.785856733524356e-05,
+      "loss": 0.0806,
+      "step": 3114
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785787965616046e-05,
+      "loss": 0.0943,
+      "step": 3115
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785719197707737e-05,
+      "loss": 0.0556,
+      "step": 3116
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785650429799427e-05,
+      "loss": 0.0822,
+      "step": 3117
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785581661891118e-05,
+      "loss": 0.0779,
+      "step": 3118
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785512893982808e-05,
+      "loss": 0.0651,
+      "step": 3119
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785444126074499e-05,
+      "loss": 0.0682,
+      "step": 3120
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785375358166189e-05,
+      "loss": 0.0538,
+      "step": 3121
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.78530659025788e-05,
+      "loss": 0.1637,
+      "step": 3122
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.78523782234957e-05,
+      "loss": 0.1758,
+      "step": 3123
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785169054441261e-05,
+      "loss": 0.0366,
+      "step": 3124
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.7851002865329514e-05,
+      "loss": 0.051,
+      "step": 3125
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.785031518624642e-05,
+      "loss": 0.0483,
+      "step": 3126
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.7849627507163325e-05,
+      "loss": 0.0529,
+      "step": 3127
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.7848939828080234e-05,
+      "loss": 0.1138,
+      "step": 3128
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.7848252148997136e-05,
+      "loss": 0.0435,
+      "step": 3129
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.7847564469914045e-05,
+      "loss": 0.0488,
+      "step": 3130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.784687679083095e-05,
+      "loss": 0.063,
+      "step": 3131
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 5.7846189111747856e-05,
+      "loss": 0.1127,
+      "step": 3132
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.784550143266476e-05,
+      "loss": 0.0679,
+      "step": 3133
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.784481375358167e-05,
+      "loss": 0.0801,
+      "step": 3134
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.784412607449857e-05,
+      "loss": 0.0439,
+      "step": 3135
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.784343839541548e-05,
+      "loss": 0.0799,
+      "step": 3136
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.784275071633238e-05,
+      "loss": 0.0556,
+      "step": 3137
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.784206303724929e-05,
+      "loss": 0.0406,
+      "step": 3138
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.784137535816619e-05,
+      "loss": 0.2733,
+      "step": 3139
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.78406876790831e-05,
+      "loss": 0.0874,
+      "step": 3140
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.784e-05,
+      "loss": 0.0701,
+      "step": 3141
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.783931232091691e-05,
+      "loss": 0.0606,
+      "step": 3142
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.783862464183381e-05,
+      "loss": 0.078,
+      "step": 3143
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.783793696275072e-05,
+      "loss": 0.0335,
+      "step": 3144
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.7837249283667624e-05,
+      "loss": 0.1075,
+      "step": 3145
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.783656160458453e-05,
+      "loss": 0.0631,
+      "step": 3146
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.7835873925501435e-05,
+      "loss": 0.0794,
+      "step": 3147
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.7835186246418344e-05,
+      "loss": 0.0678,
+      "step": 3148
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 5.7834498567335246e-05,
+      "loss": 0.0453,
+      "step": 3149
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.7833810888252155e-05,
+      "loss": 0.0535,
+      "step": 3150
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.783312320916906e-05,
+      "loss": 0.0766,
+      "step": 3151
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.7832435530085966e-05,
+      "loss": 0.0793,
+      "step": 3152
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.783174785100287e-05,
+      "loss": 0.0595,
+      "step": 3153
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.783106017191978e-05,
+      "loss": 0.1118,
+      "step": 3154
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.783037249283668e-05,
+      "loss": 0.0394,
+      "step": 3155
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.782968481375359e-05,
+      "loss": 0.0453,
+      "step": 3156
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.782899713467049e-05,
+      "loss": 0.0684,
+      "step": 3157
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.78283094555874e-05,
+      "loss": 0.0882,
+      "step": 3158
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.78276217765043e-05,
+      "loss": 0.0992,
+      "step": 3159
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.782693409742121e-05,
+      "loss": 0.0397,
+      "step": 3160
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.782624641833811e-05,
+      "loss": 0.0674,
+      "step": 3161
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.782555873925502e-05,
+      "loss": 0.0614,
+      "step": 3162
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.782487106017192e-05,
+      "loss": 0.065,
+      "step": 3163
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.782418338108883e-05,
+      "loss": 0.0856,
+      "step": 3164
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.7823495702005734e-05,
+      "loss": 0.0339,
+      "step": 3165
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.782280802292264e-05,
+      "loss": 0.0668,
+      "step": 3166
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 5.7822120343839545e-05,
+      "loss": 0.0917,
+      "step": 3167
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.7821432664756454e-05,
+      "loss": 0.0511,
+      "step": 3168
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.7820744985673356e-05,
+      "loss": 0.0483,
+      "step": 3169
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.7820057306590265e-05,
+      "loss": 0.0775,
+      "step": 3170
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.781936962750717e-05,
+      "loss": 0.0887,
+      "step": 3171
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.7818681948424076e-05,
+      "loss": 0.0581,
+      "step": 3172
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.781799426934098e-05,
+      "loss": 0.0721,
+      "step": 3173
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.781730659025789e-05,
+      "loss": 0.0605,
+      "step": 3174
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.781661891117479e-05,
+      "loss": 0.0571,
+      "step": 3175
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.78159312320917e-05,
+      "loss": 0.0647,
+      "step": 3176
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.78152435530086e-05,
+      "loss": 0.0807,
+      "step": 3177
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.781455587392551e-05,
+      "loss": 0.0741,
+      "step": 3178
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.781386819484241e-05,
+      "loss": 0.0644,
+      "step": 3179
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.781318051575931e-05,
+      "loss": 0.0671,
+      "step": 3180
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.7812492836676215e-05,
+      "loss": 0.0382,
+      "step": 3181
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.7811805157593124e-05,
+      "loss": 0.0464,
+      "step": 3182
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.7811117478510026e-05,
+      "loss": 0.0483,
+      "step": 3183
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 5.7810429799426935e-05,
+      "loss": 0.0529,
+      "step": 3184
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.780974212034384e-05,
+      "loss": 0.082,
+      "step": 3185
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.7809054441260746e-05,
+      "loss": 0.0993,
+      "step": 3186
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.780836676217765e-05,
+      "loss": 0.0407,
+      "step": 3187
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.780767908309456e-05,
+      "loss": 0.0354,
+      "step": 3188
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.780699140401146e-05,
+      "loss": 0.092,
+      "step": 3189
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.780630372492837e-05,
+      "loss": 0.046,
+      "step": 3190
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.780561604584527e-05,
+      "loss": 0.0667,
+      "step": 3191
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.780492836676218e-05,
+      "loss": 0.0747,
+      "step": 3192
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.780424068767908e-05,
+      "loss": 0.0627,
+      "step": 3193
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.780355300859599e-05,
+      "loss": 0.0603,
+      "step": 3194
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.780286532951289e-05,
+      "loss": 0.054,
+      "step": 3195
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.78021776504298e-05,
+      "loss": 0.0558,
+      "step": 3196
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.78014899713467e-05,
+      "loss": 0.0898,
+      "step": 3197
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.780080229226361e-05,
+      "loss": 0.0781,
+      "step": 3198
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.7800114613180514e-05,
+      "loss": 0.0407,
+      "step": 3199
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.779942693409742e-05,
+      "loss": 0.0662,
+      "step": 3200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.7798739255014325e-05,
+      "loss": 0.0366,
+      "step": 3201
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 5.7798051575931234e-05,
+      "loss": 0.0576,
+      "step": 3202
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.7797363896848136e-05,
+      "loss": 0.0629,
+      "step": 3203
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.7796676217765045e-05,
+      "loss": 0.0901,
+      "step": 3204
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.779598853868195e-05,
+      "loss": 0.0648,
+      "step": 3205
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.7795300859598856e-05,
+      "loss": 0.0571,
+      "step": 3206
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.779461318051576e-05,
+      "loss": 0.0457,
+      "step": 3207
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.779392550143267e-05,
+      "loss": 0.0783,
+      "step": 3208
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.779323782234957e-05,
+      "loss": 0.0602,
+      "step": 3209
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.779255014326647e-05,
+      "loss": 0.0513,
+      "step": 3210
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.779186246418338e-05,
+      "loss": 0.0765,
+      "step": 3211
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.779117478510028e-05,
+      "loss": 0.0682,
+      "step": 3212
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.779048710601719e-05,
+      "loss": 0.0565,
+      "step": 3213
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.778979942693409e-05,
+      "loss": 0.1226,
+      "step": 3214
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.7789111747851e-05,
+      "loss": 0.0818,
+      "step": 3215
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.7788424068767904e-05,
+      "loss": 0.048,
+      "step": 3216
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.778773638968481e-05,
+      "loss": 0.0334,
+      "step": 3217
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.7787048710601715e-05,
+      "loss": 0.0551,
+      "step": 3218
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 5.7786361031518624e-05,
+      "loss": 0.0266,
+      "step": 3219
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.7785673352435526e-05,
+      "loss": 0.0662,
+      "step": 3220
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.7784985673352435e-05,
+      "loss": 0.0562,
+      "step": 3221
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.778429799426934e-05,
+      "loss": 0.0714,
+      "step": 3222
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.7783610315186246e-05,
+      "loss": 0.0656,
+      "step": 3223
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.778292263610315e-05,
+      "loss": 0.0647,
+      "step": 3224
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.778223495702006e-05,
+      "loss": 0.051,
+      "step": 3225
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.778154727793696e-05,
+      "loss": 0.074,
+      "step": 3226
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.778085959885387e-05,
+      "loss": 0.0497,
+      "step": 3227
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.778017191977077e-05,
+      "loss": 0.0669,
+      "step": 3228
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.777948424068768e-05,
+      "loss": 0.0569,
+      "step": 3229
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.777879656160458e-05,
+      "loss": 0.0866,
+      "step": 3230
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.777810888252149e-05,
+      "loss": 0.0447,
+      "step": 3231
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.777742120343839e-05,
+      "loss": 0.0911,
+      "step": 3232
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.77767335243553e-05,
+      "loss": 0.07,
+      "step": 3233
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.77760458452722e-05,
+      "loss": 0.1351,
+      "step": 3234
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.777535816618911e-05,
+      "loss": 0.0714,
+      "step": 3235
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 5.7774670487106014e-05,
+      "loss": 0.0828,
+      "step": 3236
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.777398280802292e-05,
+      "loss": 0.0774,
+      "step": 3237
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.7773295128939825e-05,
+      "loss": 0.0627,
+      "step": 3238
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.7772607449856734e-05,
+      "loss": 0.0427,
+      "step": 3239
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.7771919770773636e-05,
+      "loss": 0.0664,
+      "step": 3240
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.7771232091690545e-05,
+      "loss": 0.0479,
+      "step": 3241
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.777054441260745e-05,
+      "loss": 0.0347,
+      "step": 3242
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.7769856733524356e-05,
+      "loss": 0.0799,
+      "step": 3243
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.776916905444126e-05,
+      "loss": 0.0476,
+      "step": 3244
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.776848137535817e-05,
+      "loss": 0.0417,
+      "step": 3245
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.776779369627507e-05,
+      "loss": 0.0371,
+      "step": 3246
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.776710601719198e-05,
+      "loss": 0.0554,
+      "step": 3247
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.776641833810888e-05,
+      "loss": 0.0656,
+      "step": 3248
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.776573065902579e-05,
+      "loss": 0.0558,
+      "step": 3249
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.776504297994269e-05,
+      "loss": 0.0699,
+      "step": 3250
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.77643553008596e-05,
+      "loss": 0.0603,
+      "step": 3251
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.77636676217765e-05,
+      "loss": 0.0699,
+      "step": 3252
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.776297994269341e-05,
+      "loss": 0.0544,
+      "step": 3253
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 5.776229226361031e-05,
+      "loss": 0.0626,
+      "step": 3254
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.776160458452722e-05,
+      "loss": 0.0541,
+      "step": 3255
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.7760916905444124e-05,
+      "loss": 0.0519,
+      "step": 3256
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.776022922636103e-05,
+      "loss": 0.0894,
+      "step": 3257
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.7759541547277935e-05,
+      "loss": 0.0547,
+      "step": 3258
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.7758853868194844e-05,
+      "loss": 0.0482,
+      "step": 3259
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.7758166189111746e-05,
+      "loss": 0.0614,
+      "step": 3260
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.7757478510028655e-05,
+      "loss": 0.0676,
+      "step": 3261
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.775679083094556e-05,
+      "loss": 0.0648,
+      "step": 3262
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.7756103151862466e-05,
+      "loss": 0.0461,
+      "step": 3263
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.775541547277937e-05,
+      "loss": 0.0841,
+      "step": 3264
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.775472779369628e-05,
+      "loss": 0.0584,
+      "step": 3265
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.775404011461318e-05,
+      "loss": 0.0743,
+      "step": 3266
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.775335243553009e-05,
+      "loss": 0.0804,
+      "step": 3267
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.775266475644699e-05,
+      "loss": 0.0404,
+      "step": 3268
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.77519770773639e-05,
+      "loss": 0.0449,
+      "step": 3269
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.77512893982808e-05,
+      "loss": 0.0727,
+      "step": 3270
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 5.775060171919771e-05,
+      "loss": 0.0926,
+      "step": 3271
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.774991404011461e-05,
+      "loss": 0.0785,
+      "step": 3272
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.774922636103152e-05,
+      "loss": 0.0618,
+      "step": 3273
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.774853868194842e-05,
+      "loss": 0.0831,
+      "step": 3274
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.774785100286533e-05,
+      "loss": 0.0415,
+      "step": 3275
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.7747163323782234e-05,
+      "loss": 0.0689,
+      "step": 3276
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.774647564469914e-05,
+      "loss": 0.0507,
+      "step": 3277
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.7745787965616045e-05,
+      "loss": 0.0538,
+      "step": 3278
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.7745100286532954e-05,
+      "loss": 0.0549,
+      "step": 3279
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.7744412607449856e-05,
+      "loss": 0.0809,
+      "step": 3280
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.7743724928366765e-05,
+      "loss": 0.0546,
+      "step": 3281
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.774303724928367e-05,
+      "loss": 0.1076,
+      "step": 3282
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.7742349570200576e-05,
+      "loss": 0.0413,
+      "step": 3283
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.774166189111748e-05,
+      "loss": 0.0655,
+      "step": 3284
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.774097421203439e-05,
+      "loss": 0.0818,
+      "step": 3285
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.774028653295129e-05,
+      "loss": 0.0806,
+      "step": 3286
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.77395988538682e-05,
+      "loss": 0.0944,
+      "step": 3287
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.77389111747851e-05,
+      "loss": 0.0577,
+      "step": 3288
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 5.773822349570201e-05,
+      "loss": 0.069,
+      "step": 3289
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.773753581661891e-05,
+      "loss": 0.0347,
+      "step": 3290
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.773684813753582e-05,
+      "loss": 0.04,
+      "step": 3291
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.773616045845272e-05,
+      "loss": 0.0265,
+      "step": 3292
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.773547277936963e-05,
+      "loss": 0.0337,
+      "step": 3293
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.773478510028653e-05,
+      "loss": 0.0767,
+      "step": 3294
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.773409742120344e-05,
+      "loss": 0.0743,
+      "step": 3295
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7733409742120344e-05,
+      "loss": 0.0893,
+      "step": 3296
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.773272206303725e-05,
+      "loss": 0.053,
+      "step": 3297
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7732034383954155e-05,
+      "loss": 0.0338,
+      "step": 3298
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7731346704871064e-05,
+      "loss": 0.0758,
+      "step": 3299
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7730659025787966e-05,
+      "loss": 0.0571,
+      "step": 3300
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7729971346704875e-05,
+      "loss": 0.081,
+      "step": 3301
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.772928366762178e-05,
+      "loss": 0.0662,
+      "step": 3302
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.7728595988538686e-05,
+      "loss": 0.0574,
+      "step": 3303
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.772790830945559e-05,
+      "loss": 0.0637,
+      "step": 3304
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.77272206303725e-05,
+      "loss": 0.0337,
+      "step": 3305
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 5.77265329512894e-05,
+      "loss": 0.1,
+      "step": 3306
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.772584527220631e-05,
+      "loss": 0.0388,
+      "step": 3307
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.772515759312321e-05,
+      "loss": 0.0374,
+      "step": 3308
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.772446991404012e-05,
+      "loss": 0.0287,
+      "step": 3309
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.772378223495702e-05,
+      "loss": 0.044,
+      "step": 3310
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.772309455587393e-05,
+      "loss": 0.0694,
+      "step": 3311
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.772240687679083e-05,
+      "loss": 0.0443,
+      "step": 3312
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.772171919770774e-05,
+      "loss": 0.0922,
+      "step": 3313
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.772103151862464e-05,
+      "loss": 0.0607,
+      "step": 3314
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.772034383954155e-05,
+      "loss": 0.1257,
+      "step": 3315
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7719656160458454e-05,
+      "loss": 0.1106,
+      "step": 3316
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.771896848137536e-05,
+      "loss": 0.0735,
+      "step": 3317
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7718280802292265e-05,
+      "loss": 0.06,
+      "step": 3318
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7717593123209174e-05,
+      "loss": 0.0442,
+      "step": 3319
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7716905444126076e-05,
+      "loss": 0.0455,
+      "step": 3320
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7716217765042985e-05,
+      "loss": 0.0946,
+      "step": 3321
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.771553008595989e-05,
+      "loss": 0.0455,
+      "step": 3322
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.7714842406876796e-05,
+      "loss": 0.0481,
+      "step": 3323
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 5.77141547277937e-05,
+      "loss": 0.0628,
+      "step": 3324
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.771346704871061e-05,
+      "loss": 0.0934,
+      "step": 3325
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.771277936962751e-05,
+      "loss": 0.0616,
+      "step": 3326
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.771209169054442e-05,
+      "loss": 0.0485,
+      "step": 3327
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.771140401146132e-05,
+      "loss": 0.0833,
+      "step": 3328
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.771071633237822e-05,
+      "loss": 0.049,
+      "step": 3329
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.771002865329513e-05,
+      "loss": 0.0588,
+      "step": 3330
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.770934097421203e-05,
+      "loss": 0.0916,
+      "step": 3331
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.770865329512894e-05,
+      "loss": 0.0763,
+      "step": 3332
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.7707965616045844e-05,
+      "loss": 0.0644,
+      "step": 3333
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.770727793696275e-05,
+      "loss": 0.0502,
+      "step": 3334
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.7706590257879655e-05,
+      "loss": 0.0628,
+      "step": 3335
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.7705902578796564e-05,
+      "loss": 0.0511,
+      "step": 3336
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.7705214899713466e-05,
+      "loss": 0.0772,
+      "step": 3337
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.7704527220630375e-05,
+      "loss": 0.0434,
+      "step": 3338
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.770383954154728e-05,
+      "loss": 0.0546,
+      "step": 3339
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.7703151862464186e-05,
+      "loss": 0.1087,
+      "step": 3340
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 5.770246418338109e-05,
+      "loss": 0.0419,
+      "step": 3341
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.7701776504298e-05,
+      "loss": 0.0525,
+      "step": 3342
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.77010888252149e-05,
+      "loss": 0.0625,
+      "step": 3343
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.770040114613181e-05,
+      "loss": 0.0643,
+      "step": 3344
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.769971346704871e-05,
+      "loss": 0.0439,
+      "step": 3345
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.769902578796562e-05,
+      "loss": 0.0773,
+      "step": 3346
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.769833810888252e-05,
+      "loss": 0.0486,
+      "step": 3347
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.769765042979943e-05,
+      "loss": 0.0452,
+      "step": 3348
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.769696275071633e-05,
+      "loss": 0.067,
+      "step": 3349
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.769627507163324e-05,
+      "loss": 0.0564,
+      "step": 3350
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.769558739255014e-05,
+      "loss": 0.0615,
+      "step": 3351
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.769489971346705e-05,
+      "loss": 0.0278,
+      "step": 3352
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.7694212034383954e-05,
+      "loss": 0.1192,
+      "step": 3353
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.769352435530086e-05,
+      "loss": 0.0254,
+      "step": 3354
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.7692836676217765e-05,
+      "loss": 0.0658,
+      "step": 3355
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.7692148997134674e-05,
+      "loss": 0.0551,
+      "step": 3356
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.7691461318051576e-05,
+      "loss": 0.0701,
+      "step": 3357
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.7690773638968485e-05,
+      "loss": 0.0859,
+      "step": 3358
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 5.769008595988539e-05,
+      "loss": 0.0832,
+      "step": 3359
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7689398280802296e-05,
+      "loss": 0.0512,
+      "step": 3360
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.76887106017192e-05,
+      "loss": 0.0747,
+      "step": 3361
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.768802292263611e-05,
+      "loss": 0.0522,
+      "step": 3362
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.768733524355301e-05,
+      "loss": 0.0567,
+      "step": 3363
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.768664756446992e-05,
+      "loss": 0.0542,
+      "step": 3364
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.768595988538682e-05,
+      "loss": 0.0822,
+      "step": 3365
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.768527220630373e-05,
+      "loss": 0.1218,
+      "step": 3366
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.768458452722063e-05,
+      "loss": 0.0825,
+      "step": 3367
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.768389684813754e-05,
+      "loss": 0.0789,
+      "step": 3368
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.768320916905444e-05,
+      "loss": 0.057,
+      "step": 3369
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.768252148997135e-05,
+      "loss": 0.0706,
+      "step": 3370
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.768183381088825e-05,
+      "loss": 0.095,
+      "step": 3371
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.768114613180516e-05,
+      "loss": 0.0935,
+      "step": 3372
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7680458452722064e-05,
+      "loss": 0.0664,
+      "step": 3373
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.767977077363897e-05,
+      "loss": 0.0365,
+      "step": 3374
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7679083094555875e-05,
+      "loss": 0.0504,
+      "step": 3375
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.7678395415472784e-05,
+      "loss": 0.0518,
+      "step": 3376
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.7677707736389686e-05,
+      "loss": 0.0364,
+      "step": 3377
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.7677020057306595e-05,
+      "loss": 0.0701,
+      "step": 3378
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.76763323782235e-05,
+      "loss": 0.0851,
+      "step": 3379
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.7675644699140406e-05,
+      "loss": 0.0751,
+      "step": 3380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.767495702005731e-05,
+      "loss": 0.0848,
+      "step": 3381
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.767426934097422e-05,
+      "loss": 0.0353,
+      "step": 3382
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.767358166189112e-05,
+      "loss": 0.0561,
+      "step": 3383
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.767289398280803e-05,
+      "loss": 0.0627,
+      "step": 3384
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.767220630372493e-05,
+      "loss": 0.0641,
+      "step": 3385
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.767151862464184e-05,
+      "loss": 0.0966,
+      "step": 3386
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.767083094555874e-05,
+      "loss": 0.102,
+      "step": 3387
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.767014326647565e-05,
+      "loss": 0.0797,
+      "step": 3388
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.766945558739255e-05,
+      "loss": 0.0649,
+      "step": 3389
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.766876790830946e-05,
+      "loss": 0.0855,
+      "step": 3390
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.766808022922636e-05,
+      "loss": 0.07,
+      "step": 3391
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.766739255014327e-05,
+      "loss": 0.0897,
+      "step": 3392
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.7666704871060174e-05,
+      "loss": 0.0749,
+      "step": 3393
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.766601719197708e-05,
+      "loss": 0.0592,
+      "step": 3394
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7665329512893985e-05,
+      "loss": 0.0801,
+      "step": 3395
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7664641833810894e-05,
+      "loss": 0.0983,
+      "step": 3396
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7663954154727796e-05,
+      "loss": 0.0558,
+      "step": 3397
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7663266475644705e-05,
+      "loss": 0.0695,
+      "step": 3398
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.766257879656161e-05,
+      "loss": 0.0477,
+      "step": 3399
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7661891117478516e-05,
+      "loss": 0.0506,
+      "step": 3400
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.766120343839542e-05,
+      "loss": 0.0662,
+      "step": 3401
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.766051575931233e-05,
+      "loss": 0.0892,
+      "step": 3402
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765982808022923e-05,
+      "loss": 0.0631,
+      "step": 3403
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765914040114614e-05,
+      "loss": 0.0801,
+      "step": 3404
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765845272206304e-05,
+      "loss": 0.0554,
+      "step": 3405
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765776504297995e-05,
+      "loss": 0.1081,
+      "step": 3406
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765707736389685e-05,
+      "loss": 0.0806,
+      "step": 3407
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765638968481376e-05,
+      "loss": 0.1056,
+      "step": 3408
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765570200573066e-05,
+      "loss": 0.0468,
+      "step": 3409
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765501432664757e-05,
+      "loss": 0.0345,
+      "step": 3410
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.765432664756447e-05,
+      "loss": 0.0776,
+      "step": 3411
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.765363896848138e-05,
+      "loss": 0.0508,
+      "step": 3412
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7652951289398284e-05,
+      "loss": 0.0967,
+      "step": 3413
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.765226361031519e-05,
+      "loss": 0.0492,
+      "step": 3414
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7651575931232095e-05,
+      "loss": 0.0568,
+      "step": 3415
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7650888252149004e-05,
+      "loss": 0.0777,
+      "step": 3416
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7650200573065906e-05,
+      "loss": 0.0367,
+      "step": 3417
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7649512893982815e-05,
+      "loss": 0.0818,
+      "step": 3418
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.764882521489972e-05,
+      "loss": 0.0664,
+      "step": 3419
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7648137535816626e-05,
+      "loss": 0.0632,
+      "step": 3420
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.764744985673353e-05,
+      "loss": 0.0504,
+      "step": 3421
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.764676217765044e-05,
+      "loss": 0.0756,
+      "step": 3422
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.764607449856734e-05,
+      "loss": 0.0722,
+      "step": 3423
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.764538681948425e-05,
+      "loss": 0.07,
+      "step": 3424
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.764469914040115e-05,
+      "loss": 0.0755,
+      "step": 3425
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.764401146131806e-05,
+      "loss": 0.0291,
+      "step": 3426
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.764332378223496e-05,
+      "loss": 0.0742,
+      "step": 3427
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.764263610315187e-05,
+      "loss": 0.0844,
+      "step": 3428
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.764194842406877e-05,
+      "loss": 0.1966,
+      "step": 3429
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.764126074498568e-05,
+      "loss": 0.1485,
+      "step": 3430
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.764057306590258e-05,
+      "loss": 0.0723,
+      "step": 3431
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763988538681949e-05,
+      "loss": 0.0701,
+      "step": 3432
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.7639197707736394e-05,
+      "loss": 0.0561,
+      "step": 3433
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.7638510028653296e-05,
+      "loss": 0.1159,
+      "step": 3434
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.76378223495702e-05,
+      "loss": 0.0622,
+      "step": 3435
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763713467048711e-05,
+      "loss": 0.0617,
+      "step": 3436
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763644699140401e-05,
+      "loss": 0.0493,
+      "step": 3437
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763575931232092e-05,
+      "loss": 0.0629,
+      "step": 3438
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763507163323782e-05,
+      "loss": 0.0639,
+      "step": 3439
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763438395415473e-05,
+      "loss": 0.0406,
+      "step": 3440
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763369627507163e-05,
+      "loss": 0.0935,
+      "step": 3441
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763300859598854e-05,
+      "loss": 0.1217,
+      "step": 3442
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763232091690544e-05,
+      "loss": 0.0516,
+      "step": 3443
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763163323782235e-05,
+      "loss": 0.101,
+      "step": 3444
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763094555873925e-05,
+      "loss": 0.1184,
+      "step": 3445
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.763025787965616e-05,
+      "loss": 0.0477,
+      "step": 3446
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.7629570200573064e-05,
+      "loss": 0.089,
+      "step": 3447
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.7628882521489966e-05,
+      "loss": 0.0743,
+      "step": 3448
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.7628194842406875e-05,
+      "loss": 0.0749,
+      "step": 3449
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.762750716332378e-05,
+      "loss": 0.0684,
+      "step": 3450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.7626819484240686e-05,
+      "loss": 0.0724,
+      "step": 3451
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.762613180515759e-05,
+      "loss": 0.0536,
+      "step": 3452
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.76254441260745e-05,
+      "loss": 0.1043,
+      "step": 3453
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.76247564469914e-05,
+      "loss": 0.0512,
+      "step": 3454
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.762406876790831e-05,
+      "loss": 0.0554,
+      "step": 3455
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.762338108882521e-05,
+      "loss": 0.0636,
+      "step": 3456
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.762269340974212e-05,
+      "loss": 0.0532,
+      "step": 3457
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.762200573065902e-05,
+      "loss": 0.0912,
+      "step": 3458
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.762131805157593e-05,
+      "loss": 0.0665,
+      "step": 3459
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.762063037249283e-05,
+      "loss": 0.0765,
+      "step": 3460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.761994269340974e-05,
+      "loss": 0.0782,
+      "step": 3461
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.7619255014326643e-05,
+      "loss": 0.0752,
+      "step": 3462
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.761856733524355e-05,
+      "loss": 0.0595,
+      "step": 3463
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.7617879656160454e-05,
+      "loss": 0.0728,
+      "step": 3464
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.761719197707736e-05,
+      "loss": 0.0467,
+      "step": 3465
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.7616504297994265e-05,
+      "loss": 0.0693,
+      "step": 3466
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.7615816618911174e-05,
+      "loss": 0.0458,
+      "step": 3467
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.7615128939828076e-05,
+      "loss": 0.0372,
+      "step": 3468
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.7614441260744985e-05,
+      "loss": 0.0384,
+      "step": 3469
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.761375358166189e-05,
+      "loss": 0.0704,
+      "step": 3470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.7613065902578796e-05,
+      "loss": 0.0488,
+      "step": 3471
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.76123782234957e-05,
+      "loss": 0.0712,
+      "step": 3472
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.761169054441261e-05,
+      "loss": 0.0513,
+      "step": 3473
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.761100286532951e-05,
+      "loss": 0.1339,
+      "step": 3474
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.761031518624642e-05,
+      "loss": 0.0899,
+      "step": 3475
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.760962750716332e-05,
+      "loss": 0.1181,
+      "step": 3476
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.760893982808023e-05,
+      "loss": 0.0764,
+      "step": 3477
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.760825214899713e-05,
+      "loss": 0.0499,
+      "step": 3478
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.760756446991404e-05,
+      "loss": 0.0805,
+      "step": 3479
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.760687679083094e-05,
+      "loss": 0.0416,
+      "step": 3480
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.760618911174785e-05,
+      "loss": 0.0585,
+      "step": 3481
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.7605501432664753e-05,
+      "loss": 0.052,
+      "step": 3482
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.760481375358166e-05,
+      "loss": 0.0799,
+      "step": 3483
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.7604126074498564e-05,
+      "loss": 0.0654,
+      "step": 3484
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.760343839541547e-05,
+      "loss": 0.0399,
+      "step": 3485
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.7602750716332375e-05,
+      "loss": 0.1044,
+      "step": 3486
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.7602063037249284e-05,
+      "loss": 0.0466,
+      "step": 3487
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.7601375358166186e-05,
+      "loss": 0.0484,
+      "step": 3488
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.7600687679083095e-05,
+      "loss": 0.1077,
+      "step": 3489
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.76e-05,
+      "loss": 0.0398,
+      "step": 3490
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.7599312320916906e-05,
+      "loss": 0.0804,
+      "step": 3491
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.759862464183381e-05,
+      "loss": 0.052,
+      "step": 3492
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.759793696275072e-05,
+      "loss": 0.0448,
+      "step": 3493
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.759724928366762e-05,
+      "loss": 0.0293,
+      "step": 3494
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.759656160458453e-05,
+      "loss": 0.0826,
+      "step": 3495
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.759587392550143e-05,
+      "loss": 0.0473,
+      "step": 3496
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.759518624641834e-05,
+      "loss": 0.0445,
+      "step": 3497
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.759449856733524e-05,
+      "loss": 0.0709,
+      "step": 3498
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.759381088825215e-05,
+      "loss": 0.0438,
+      "step": 3499
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.759312320916905e-05,
+      "loss": 0.0415,
+      "step": 3500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.759243553008596e-05,
+      "loss": 0.0722,
+      "step": 3501
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.759174785100286e-05,
+      "loss": 0.0272,
+      "step": 3502
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.759106017191977e-05,
+      "loss": 0.034,
+      "step": 3503
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.7590372492836674e-05,
+      "loss": 0.0726,
+      "step": 3504
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.758968481375358e-05,
+      "loss": 0.0393,
+      "step": 3505
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.7588997134670485e-05,
+      "loss": 0.0372,
+      "step": 3506
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.7588309455587394e-05,
+      "loss": 0.0426,
+      "step": 3507
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.7587621776504296e-05,
+      "loss": 0.0456,
+      "step": 3508
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.7586934097421205e-05,
+      "loss": 0.0543,
+      "step": 3509
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.758624641833811e-05,
+      "loss": 0.0675,
+      "step": 3510
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.7585558739255016e-05,
+      "loss": 0.0431,
+      "step": 3511
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.758487106017192e-05,
+      "loss": 0.0547,
+      "step": 3512
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.758418338108883e-05,
+      "loss": 0.0666,
+      "step": 3513
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.758349570200573e-05,
+      "loss": 0.0696,
+      "step": 3514
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.758280802292264e-05,
+      "loss": 0.054,
+      "step": 3515
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.758212034383954e-05,
+      "loss": 0.0306,
+      "step": 3516
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.758143266475645e-05,
+      "loss": 0.0502,
+      "step": 3517
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.758074498567335e-05,
+      "loss": 0.0637,
+      "step": 3518
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.758005730659026e-05,
+      "loss": 0.0537,
+      "step": 3519
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.757936962750716e-05,
+      "loss": 0.0735,
+      "step": 3520
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.757868194842407e-05,
+      "loss": 0.0318,
+      "step": 3521
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.757799426934097e-05,
+      "loss": 0.0715,
+      "step": 3522
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.757730659025788e-05,
+      "loss": 0.06,
+      "step": 3523
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.7576618911174784e-05,
+      "loss": 0.0713,
+      "step": 3524
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.757593123209169e-05,
+      "loss": 0.0244,
+      "step": 3525
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.7575243553008595e-05,
+      "loss": 0.032,
+      "step": 3526
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.7574555873925504e-05,
+      "loss": 0.0478,
+      "step": 3527
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.7573868194842406e-05,
+      "loss": 0.0711,
+      "step": 3528
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.7573180515759315e-05,
+      "loss": 0.0452,
+      "step": 3529
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.757249283667622e-05,
+      "loss": 0.0488,
+      "step": 3530
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.7571805157593126e-05,
+      "loss": 0.0552,
+      "step": 3531
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.757111747851003e-05,
+      "loss": 0.0515,
+      "step": 3532
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.757042979942694e-05,
+      "loss": 0.0406,
+      "step": 3533
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.756974212034384e-05,
+      "loss": 0.0321,
+      "step": 3534
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.756905444126075e-05,
+      "loss": 0.0518,
+      "step": 3535
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.756836676217765e-05,
+      "loss": 0.0487,
+      "step": 3536
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.756767908309456e-05,
+      "loss": 0.0379,
+      "step": 3537
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.756699140401146e-05,
+      "loss": 0.0798,
+      "step": 3538
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.756630372492837e-05,
+      "loss": 0.0349,
+      "step": 3539
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.756561604584527e-05,
+      "loss": 0.0341,
+      "step": 3540
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.756492836676218e-05,
+      "loss": 0.0512,
+      "step": 3541
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.756424068767908e-05,
+      "loss": 0.0598,
+      "step": 3542
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.756355300859599e-05,
+      "loss": 0.0528,
+      "step": 3543
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.7562865329512894e-05,
+      "loss": 0.0383,
+      "step": 3544
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.75621776504298e-05,
+      "loss": 0.0467,
+      "step": 3545
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.7561489971346705e-05,
+      "loss": 0.0538,
+      "step": 3546
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.7560802292263614e-05,
+      "loss": 0.0516,
+      "step": 3547
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.7560114613180516e-05,
+      "loss": 0.0755,
+      "step": 3548
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.7559426934097425e-05,
+      "loss": 0.0672,
+      "step": 3549
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.755873925501433e-05,
+      "loss": 0.0319,
+      "step": 3550
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.7558051575931236e-05,
+      "loss": 0.0632,
+      "step": 3551
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755736389684814e-05,
+      "loss": 0.045,
+      "step": 3552
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755667621776505e-05,
+      "loss": 0.0604,
+      "step": 3553
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755598853868195e-05,
+      "loss": 0.0424,
+      "step": 3554
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755530085959886e-05,
+      "loss": 0.0471,
+      "step": 3555
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755461318051576e-05,
+      "loss": 0.0676,
+      "step": 3556
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755392550143267e-05,
+      "loss": 0.0695,
+      "step": 3557
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755323782234957e-05,
+      "loss": 0.0659,
+      "step": 3558
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755255014326648e-05,
+      "loss": 0.0457,
+      "step": 3559
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755186246418338e-05,
+      "loss": 0.0508,
+      "step": 3560
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755117478510029e-05,
+      "loss": 0.0535,
+      "step": 3561
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.755048710601719e-05,
+      "loss": 0.0354,
+      "step": 3562
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.75497994269341e-05,
+      "loss": 0.0408,
+      "step": 3563
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.7549111747851004e-05,
+      "loss": 0.0404,
+      "step": 3564
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.754842406876791e-05,
+      "loss": 0.0641,
+      "step": 3565
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.7547736389684815e-05,
+      "loss": 0.0469,
+      "step": 3566
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.754704871060172e-05,
+      "loss": 0.0747,
+      "step": 3567
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 5.7546361031518626e-05,
+      "loss": 0.0654,
+      "step": 3568
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.754567335243553e-05,
+      "loss": 0.0419,
+      "step": 3569
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.754498567335244e-05,
+      "loss": 0.0759,
+      "step": 3570
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.754429799426934e-05,
+      "loss": 0.0608,
+      "step": 3571
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.754361031518625e-05,
+      "loss": 0.0494,
+      "step": 3572
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.754292263610315e-05,
+      "loss": 0.0339,
+      "step": 3573
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.754223495702006e-05,
+      "loss": 0.0292,
+      "step": 3574
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.754154727793696e-05,
+      "loss": 0.0792,
+      "step": 3575
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.754085959885387e-05,
+      "loss": 0.0456,
+      "step": 3576
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.754017191977077e-05,
+      "loss": 0.0455,
+      "step": 3577
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.753948424068768e-05,
+      "loss": 0.0392,
+      "step": 3578
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.7538796561604583e-05,
+      "loss": 0.0418,
+      "step": 3579
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.753810888252149e-05,
+      "loss": 0.0763,
+      "step": 3580
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.7537421203438394e-05,
+      "loss": 0.0729,
+      "step": 3581
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.75367335243553e-05,
+      "loss": 0.0525,
+      "step": 3582
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.7536045845272205e-05,
+      "loss": 0.0362,
+      "step": 3583
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.7535358166189114e-05,
+      "loss": 0.0663,
+      "step": 3584
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 5.7534670487106016e-05,
+      "loss": 0.0581,
+      "step": 3585
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.7533982808022925e-05,
+      "loss": 0.066,
+      "step": 3586
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.753329512893983e-05,
+      "loss": 0.0416,
+      "step": 3587
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.7532607449856736e-05,
+      "loss": 0.0595,
+      "step": 3588
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.753191977077364e-05,
+      "loss": 0.0493,
+      "step": 3589
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.753123209169055e-05,
+      "loss": 0.0492,
+      "step": 3590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.753054441260745e-05,
+      "loss": 0.0788,
+      "step": 3591
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.752985673352436e-05,
+      "loss": 0.053,
+      "step": 3592
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.752916905444126e-05,
+      "loss": 0.0348,
+      "step": 3593
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.752848137535817e-05,
+      "loss": 0.0367,
+      "step": 3594
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.752779369627507e-05,
+      "loss": 0.0191,
+      "step": 3595
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.752710601719198e-05,
+      "loss": 0.0527,
+      "step": 3596
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.752641833810888e-05,
+      "loss": 0.0317,
+      "step": 3597
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.752573065902579e-05,
+      "loss": 0.0604,
+      "step": 3598
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.7525042979942693e-05,
+      "loss": 0.0617,
+      "step": 3599
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.75243553008596e-05,
+      "loss": 0.0647,
+      "step": 3600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.7523667621776504e-05,
+      "loss": 0.061,
+      "step": 3601
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.752297994269341e-05,
+      "loss": 0.0613,
+      "step": 3602
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 5.7522292263610315e-05,
+      "loss": 0.0505,
+      "step": 3603
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.7521604584527224e-05,
+      "loss": 0.0715,
+      "step": 3604
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.7520916905444126e-05,
+      "loss": 0.0416,
+      "step": 3605
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.7520229226361035e-05,
+      "loss": 0.0577,
+      "step": 3606
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.751954154727794e-05,
+      "loss": 0.0572,
+      "step": 3607
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.7518853868194846e-05,
+      "loss": 0.0713,
+      "step": 3608
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.751816618911175e-05,
+      "loss": 0.0412,
+      "step": 3609
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.751747851002866e-05,
+      "loss": 0.0495,
+      "step": 3610
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.751679083094556e-05,
+      "loss": 0.0807,
+      "step": 3611
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.751610315186247e-05,
+      "loss": 0.0358,
+      "step": 3612
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.751541547277937e-05,
+      "loss": 0.0473,
+      "step": 3613
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.751472779369628e-05,
+      "loss": 0.0571,
+      "step": 3614
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.751404011461318e-05,
+      "loss": 0.0645,
+      "step": 3615
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.751335243553009e-05,
+      "loss": 0.0315,
+      "step": 3616
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.751266475644699e-05,
+      "loss": 0.0482,
+      "step": 3617
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.75119770773639e-05,
+      "loss": 0.0504,
+      "step": 3618
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.7511289398280803e-05,
+      "loss": 0.0659,
+      "step": 3619
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 5.751060171919771e-05,
+      "loss": 0.0718,
+      "step": 3620
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.7509914040114614e-05,
+      "loss": 0.044,
+      "step": 3621
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.750922636103152e-05,
+      "loss": 0.0361,
+      "step": 3622
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.7508538681948425e-05,
+      "loss": 0.0467,
+      "step": 3623
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.7507851002865334e-05,
+      "loss": 0.0454,
+      "step": 3624
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.7507163323782236e-05,
+      "loss": 0.0594,
+      "step": 3625
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.7506475644699145e-05,
+      "loss": 0.0499,
+      "step": 3626
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.750578796561605e-05,
+      "loss": 0.0423,
+      "step": 3627
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.7505100286532956e-05,
+      "loss": 0.0416,
+      "step": 3628
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.750441260744986e-05,
+      "loss": 0.0444,
+      "step": 3629
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.750372492836677e-05,
+      "loss": 0.0881,
+      "step": 3630
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.750303724928367e-05,
+      "loss": 0.0769,
+      "step": 3631
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.750234957020058e-05,
+      "loss": 0.048,
+      "step": 3632
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.750166189111748e-05,
+      "loss": 0.0576,
+      "step": 3633
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.750097421203439e-05,
+      "loss": 0.0991,
+      "step": 3634
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.750028653295129e-05,
+      "loss": 0.0282,
+      "step": 3635
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.74995988538682e-05,
+      "loss": 0.0511,
+      "step": 3636
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.74989111747851e-05,
+      "loss": 0.0411,
+      "step": 3637
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 5.749822349570201e-05,
+      "loss": 0.061,
+      "step": 3638
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.7497535816618913e-05,
+      "loss": 0.0441,
+      "step": 3639
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.749684813753582e-05,
+      "loss": 0.0463,
+      "step": 3640
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.7496160458452724e-05,
+      "loss": 0.025,
+      "step": 3641
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.749547277936963e-05,
+      "loss": 0.0618,
+      "step": 3642
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.7494785100286535e-05,
+      "loss": 0.0482,
+      "step": 3643
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.7494097421203444e-05,
+      "loss": 0.0662,
+      "step": 3644
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.7493409742120346e-05,
+      "loss": 0.0257,
+      "step": 3645
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.7492722063037255e-05,
+      "loss": 0.0554,
+      "step": 3646
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.749203438395416e-05,
+      "loss": 0.0671,
+      "step": 3647
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.7491346704871066e-05,
+      "loss": 0.0617,
+      "step": 3648
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.749065902578797e-05,
+      "loss": 0.049,
+      "step": 3649
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.748997134670488e-05,
+      "loss": 0.0263,
+      "step": 3650
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.748928366762178e-05,
+      "loss": 0.0217,
+      "step": 3651
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.748859598853869e-05,
+      "loss": 0.0538,
+      "step": 3652
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.748790830945559e-05,
+      "loss": 0.0384,
+      "step": 3653
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.74872206303725e-05,
+      "loss": 0.0724,
+      "step": 3654
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 5.74865329512894e-05,
+      "loss": 0.0325,
+      "step": 3655
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.748584527220631e-05,
+      "loss": 0.0644,
+      "step": 3656
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.748515759312321e-05,
+      "loss": 0.0668,
+      "step": 3657
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.748446991404012e-05,
+      "loss": 0.1114,
+      "step": 3658
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.748378223495702e-05,
+      "loss": 0.0418,
+      "step": 3659
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.748309455587393e-05,
+      "loss": 0.0638,
+      "step": 3660
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.7482406876790834e-05,
+      "loss": 0.0781,
+      "step": 3661
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.748171919770774e-05,
+      "loss": 0.0572,
+      "step": 3662
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.7481031518624645e-05,
+      "loss": 0.0482,
+      "step": 3663
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.7480343839541554e-05,
+      "loss": 0.0632,
+      "step": 3664
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.7479656160458456e-05,
+      "loss": 0.0463,
+      "step": 3665
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.7478968481375365e-05,
+      "loss": 0.1418,
+      "step": 3666
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.747828080229227e-05,
+      "loss": 0.0374,
+      "step": 3667
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.7477593123209176e-05,
+      "loss": 0.0476,
+      "step": 3668
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.747690544412608e-05,
+      "loss": 0.0499,
+      "step": 3669
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.747621776504299e-05,
+      "loss": 0.0337,
+      "step": 3670
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.747553008595989e-05,
+      "loss": 0.0569,
+      "step": 3671
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.74748424068768e-05,
+      "loss": 0.0887,
+      "step": 3672
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 5.74741547277937e-05,
+      "loss": 0.0421,
+      "step": 3673
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.747346704871061e-05,
+      "loss": 0.0575,
+      "step": 3674
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.747277936962751e-05,
+      "loss": 0.0399,
+      "step": 3675
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.747209169054442e-05,
+      "loss": 0.0403,
+      "step": 3676
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.747140401146132e-05,
+      "loss": 0.0553,
+      "step": 3677
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.747071633237823e-05,
+      "loss": 0.0391,
+      "step": 3678
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.747002865329513e-05,
+      "loss": 0.0444,
+      "step": 3679
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.746934097421204e-05,
+      "loss": 0.0239,
+      "step": 3680
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.7468653295128944e-05,
+      "loss": 0.0636,
+      "step": 3681
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.746796561604585e-05,
+      "loss": 0.0485,
+      "step": 3682
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.7467277936962755e-05,
+      "loss": 0.0592,
+      "step": 3683
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.7466590257879664e-05,
+      "loss": 0.0691,
+      "step": 3684
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.7465902578796566e-05,
+      "loss": 0.0382,
+      "step": 3685
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.746521489971347e-05,
+      "loss": 0.0669,
+      "step": 3686
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.746452722063038e-05,
+      "loss": 0.0569,
+      "step": 3687
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.746383954154728e-05,
+      "loss": 0.0597,
+      "step": 3688
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.746315186246418e-05,
+      "loss": 0.1017,
+      "step": 3689
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 5.746246418338109e-05,
+      "loss": 0.055,
+      "step": 3690
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.746177650429799e-05,
+      "loss": 0.0357,
+      "step": 3691
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.7461088825214895e-05,
+      "loss": 0.0323,
+      "step": 3692
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.7460401146131804e-05,
+      "loss": 0.0554,
+      "step": 3693
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.7459713467048706e-05,
+      "loss": 0.0275,
+      "step": 3694
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.7459025787965615e-05,
+      "loss": 0.0508,
+      "step": 3695
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745833810888252e-05,
+      "loss": 0.0542,
+      "step": 3696
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.7457650429799426e-05,
+      "loss": 0.0383,
+      "step": 3697
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745696275071633e-05,
+      "loss": 0.0493,
+      "step": 3698
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.7456275071633237e-05,
+      "loss": 0.0408,
+      "step": 3699
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745558739255014e-05,
+      "loss": 0.0369,
+      "step": 3700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745489971346705e-05,
+      "loss": 0.0266,
+      "step": 3701
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745421203438395e-05,
+      "loss": 0.0382,
+      "step": 3702
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745352435530086e-05,
+      "loss": 0.0468,
+      "step": 3703
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745283667621776e-05,
+      "loss": 0.0349,
+      "step": 3704
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745214899713467e-05,
+      "loss": 0.0415,
+      "step": 3705
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745146131805157e-05,
+      "loss": 0.0793,
+      "step": 3706
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745077363896848e-05,
+      "loss": 0.0433,
+      "step": 3707
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 5.745008595988538e-05,
+      "loss": 0.0533,
+      "step": 3708
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.744939828080229e-05,
+      "loss": 0.0475,
+      "step": 3709
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7448710601719194e-05,
+      "loss": 0.0641,
+      "step": 3710
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.74480229226361e-05,
+      "loss": 0.0568,
+      "step": 3711
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7447335243553005e-05,
+      "loss": 0.0453,
+      "step": 3712
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7446647564469914e-05,
+      "loss": 0.0587,
+      "step": 3713
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7445959885386816e-05,
+      "loss": 0.0533,
+      "step": 3714
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7445272206303725e-05,
+      "loss": 0.0373,
+      "step": 3715
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.744458452722063e-05,
+      "loss": 0.0413,
+      "step": 3716
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7443896848137536e-05,
+      "loss": 0.0583,
+      "step": 3717
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.744320916905444e-05,
+      "loss": 0.0426,
+      "step": 3718
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.7442521489971347e-05,
+      "loss": 0.0593,
+      "step": 3719
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.744183381088825e-05,
+      "loss": 0.0774,
+      "step": 3720
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.744114613180516e-05,
+      "loss": 0.0306,
+      "step": 3721
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.744045845272206e-05,
+      "loss": 0.0401,
+      "step": 3722
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.743977077363897e-05,
+      "loss": 0.037,
+      "step": 3723
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.743908309455587e-05,
+      "loss": 0.0783,
+      "step": 3724
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 5.743839541547278e-05,
+      "loss": 0.0662,
+      "step": 3725
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.743770773638968e-05,
+      "loss": 0.0383,
+      "step": 3726
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.743702005730659e-05,
+      "loss": 0.0543,
+      "step": 3727
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.743633237822349e-05,
+      "loss": 0.0268,
+      "step": 3728
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.74356446991404e-05,
+      "loss": 0.0485,
+      "step": 3729
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7434957020057304e-05,
+      "loss": 0.0411,
+      "step": 3730
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.743426934097421e-05,
+      "loss": 0.0574,
+      "step": 3731
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7433581661891115e-05,
+      "loss": 0.0643,
+      "step": 3732
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7432893982808024e-05,
+      "loss": 0.0699,
+      "step": 3733
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7432206303724926e-05,
+      "loss": 0.0238,
+      "step": 3734
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7431518624641835e-05,
+      "loss": 0.0749,
+      "step": 3735
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.743083094555874e-05,
+      "loss": 0.0477,
+      "step": 3736
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7430143266475646e-05,
+      "loss": 0.0478,
+      "step": 3737
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.742945558739255e-05,
+      "loss": 0.0497,
+      "step": 3738
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.7428767908309457e-05,
+      "loss": 0.0492,
+      "step": 3739
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.742808022922636e-05,
+      "loss": 0.0617,
+      "step": 3740
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.742739255014327e-05,
+      "loss": 0.0669,
+      "step": 3741
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.742670487106017e-05,
+      "loss": 0.0347,
+      "step": 3742
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 5.742601719197708e-05,
+      "loss": 0.032,
+      "step": 3743
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.742532951289398e-05,
+      "loss": 0.0655,
+      "step": 3744
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.742464183381089e-05,
+      "loss": 0.0586,
+      "step": 3745
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.742395415472779e-05,
+      "loss": 0.0649,
+      "step": 3746
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.74232664756447e-05,
+      "loss": 0.0857,
+      "step": 3747
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.74225787965616e-05,
+      "loss": 0.0497,
+      "step": 3748
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.742189111747851e-05,
+      "loss": 0.0465,
+      "step": 3749
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.7421203438395414e-05,
+      "loss": 0.0672,
+      "step": 3750
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.742051575931232e-05,
+      "loss": 0.0553,
+      "step": 3751
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.7419828080229225e-05,
+      "loss": 0.0444,
+      "step": 3752
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.7419140401146134e-05,
+      "loss": 0.03,
+      "step": 3753
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.7418452722063036e-05,
+      "loss": 0.0463,
+      "step": 3754
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.7417765042979945e-05,
+      "loss": 0.0341,
+      "step": 3755
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.741707736389685e-05,
+      "loss": 0.0367,
+      "step": 3756
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.7416389684813756e-05,
+      "loss": 0.0361,
+      "step": 3757
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.741570200573066e-05,
+      "loss": 0.0546,
+      "step": 3758
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.7415014326647567e-05,
+      "loss": 0.0983,
+      "step": 3759
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 5.741432664756447e-05,
+      "loss": 0.0603,
+      "step": 3760
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.741363896848138e-05,
+      "loss": 0.0421,
+      "step": 3761
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.741295128939828e-05,
+      "loss": 0.0378,
+      "step": 3762
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.741226361031519e-05,
+      "loss": 0.0565,
+      "step": 3763
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.741157593123209e-05,
+      "loss": 0.0576,
+      "step": 3764
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7410888252149e-05,
+      "loss": 0.045,
+      "step": 3765
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.74102005730659e-05,
+      "loss": 0.0434,
+      "step": 3766
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.740951289398281e-05,
+      "loss": 0.0704,
+      "step": 3767
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.740882521489971e-05,
+      "loss": 0.0509,
+      "step": 3768
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.740813753581662e-05,
+      "loss": 0.0408,
+      "step": 3769
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7407449856733524e-05,
+      "loss": 0.0807,
+      "step": 3770
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.740676217765043e-05,
+      "loss": 0.0472,
+      "step": 3771
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7406074498567335e-05,
+      "loss": 0.0416,
+      "step": 3772
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7405386819484244e-05,
+      "loss": 0.0386,
+      "step": 3773
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7404699140401146e-05,
+      "loss": 0.0388,
+      "step": 3774
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7404011461318055e-05,
+      "loss": 0.0614,
+      "step": 3775
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.740332378223496e-05,
+      "loss": 0.041,
+      "step": 3776
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 5.7402636103151866e-05,
+      "loss": 0.0604,
+      "step": 3777
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.740194842406877e-05,
+      "loss": 0.1176,
+      "step": 3778
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7401260744985677e-05,
+      "loss": 0.0523,
+      "step": 3779
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.740057306590258e-05,
+      "loss": 0.0409,
+      "step": 3780
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.739988538681949e-05,
+      "loss": 0.0364,
+      "step": 3781
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.739919770773639e-05,
+      "loss": 0.0489,
+      "step": 3782
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.73985100286533e-05,
+      "loss": 0.0384,
+      "step": 3783
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.73978223495702e-05,
+      "loss": 0.0567,
+      "step": 3784
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.739713467048711e-05,
+      "loss": 0.0796,
+      "step": 3785
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.739644699140401e-05,
+      "loss": 0.0452,
+      "step": 3786
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.739575931232092e-05,
+      "loss": 0.0443,
+      "step": 3787
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.739507163323782e-05,
+      "loss": 0.0485,
+      "step": 3788
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.739438395415473e-05,
+      "loss": 0.055,
+      "step": 3789
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7393696275071634e-05,
+      "loss": 0.0516,
+      "step": 3790
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.739300859598854e-05,
+      "loss": 0.0482,
+      "step": 3791
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7392320916905445e-05,
+      "loss": 0.075,
+      "step": 3792
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7391633237822354e-05,
+      "loss": 0.0317,
+      "step": 3793
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7390945558739256e-05,
+      "loss": 0.0527,
+      "step": 3794
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 5.7390257879656165e-05,
+      "loss": 0.069,
+      "step": 3795
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.7389570200573067e-05,
+      "loss": 0.0739,
+      "step": 3796
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.7388882521489975e-05,
+      "loss": 0.0583,
+      "step": 3797
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.738819484240688e-05,
+      "loss": 0.0716,
+      "step": 3798
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.7387507163323786e-05,
+      "loss": 0.0713,
+      "step": 3799
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.738681948424069e-05,
+      "loss": 0.0632,
+      "step": 3800
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.73861318051576e-05,
+      "loss": 0.1049,
+      "step": 3801
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.73854441260745e-05,
+      "loss": 0.0278,
+      "step": 3802
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.738475644699141e-05,
+      "loss": 0.0574,
+      "step": 3803
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.738406876790831e-05,
+      "loss": 0.0455,
+      "step": 3804
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.738338108882521e-05,
+      "loss": 0.0509,
+      "step": 3805
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.738269340974212e-05,
+      "loss": 0.0415,
+      "step": 3806
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.7382005730659024e-05,
+      "loss": 0.0463,
+      "step": 3807
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.738131805157593e-05,
+      "loss": 0.0488,
+      "step": 3808
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.7380630372492835e-05,
+      "loss": 0.0894,
+      "step": 3809
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.7379942693409744e-05,
+      "loss": 0.0711,
+      "step": 3810
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.7379255014326646e-05,
+      "loss": 0.0576,
+      "step": 3811
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 5.7378567335243555e-05,
+      "loss": 0.0697,
+      "step": 3812
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.737787965616046e-05,
+      "loss": 0.0763,
+      "step": 3813
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.7377191977077366e-05,
+      "loss": 0.0386,
+      "step": 3814
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.737650429799427e-05,
+      "loss": 0.041,
+      "step": 3815
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.7375816618911177e-05,
+      "loss": 0.0395,
+      "step": 3816
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.737512893982808e-05,
+      "loss": 0.0587,
+      "step": 3817
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.737444126074499e-05,
+      "loss": 0.0851,
+      "step": 3818
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.737375358166189e-05,
+      "loss": 0.0762,
+      "step": 3819
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.73730659025788e-05,
+      "loss": 0.075,
+      "step": 3820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.73723782234957e-05,
+      "loss": 0.1092,
+      "step": 3821
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.737169054441261e-05,
+      "loss": 0.0299,
+      "step": 3822
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.737100286532951e-05,
+      "loss": 0.0342,
+      "step": 3823
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.737031518624642e-05,
+      "loss": 0.0419,
+      "step": 3824
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.736962750716332e-05,
+      "loss": 0.0501,
+      "step": 3825
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.736893982808023e-05,
+      "loss": 0.0481,
+      "step": 3826
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.7368252148997134e-05,
+      "loss": 0.0346,
+      "step": 3827
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.736756446991404e-05,
+      "loss": 0.06,
+      "step": 3828
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.7366876790830945e-05,
+      "loss": 0.0558,
+      "step": 3829
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 5.7366189111747854e-05,
+      "loss": 0.0569,
+      "step": 3830
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.7365501432664756e-05,
+      "loss": 0.0305,
+      "step": 3831
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.7364813753581665e-05,
+      "loss": 0.0491,
+      "step": 3832
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.736412607449857e-05,
+      "loss": 0.0508,
+      "step": 3833
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.7363438395415476e-05,
+      "loss": 0.0613,
+      "step": 3834
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.736275071633238e-05,
+      "loss": 0.0465,
+      "step": 3835
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.7362063037249287e-05,
+      "loss": 0.034,
+      "step": 3836
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.736137535816619e-05,
+      "loss": 0.0663,
+      "step": 3837
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.73606876790831e-05,
+      "loss": 0.0539,
+      "step": 3838
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.736e-05,
+      "loss": 0.0297,
+      "step": 3839
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.735931232091691e-05,
+      "loss": 0.0414,
+      "step": 3840
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.735862464183381e-05,
+      "loss": 0.0504,
+      "step": 3841
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.735793696275072e-05,
+      "loss": 0.0454,
+      "step": 3842
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.735724928366762e-05,
+      "loss": 0.0365,
+      "step": 3843
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.735656160458453e-05,
+      "loss": 0.0281,
+      "step": 3844
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.735587392550143e-05,
+      "loss": 0.0284,
+      "step": 3845
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.735518624641834e-05,
+      "loss": 0.0154,
+      "step": 3846
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 5.7354498567335244e-05,
+      "loss": 0.0612,
+      "step": 3847
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.735381088825215e-05,
+      "loss": 0.0606,
+      "step": 3848
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.7353123209169055e-05,
+      "loss": 0.1308,
+      "step": 3849
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.7352435530085964e-05,
+      "loss": 0.0553,
+      "step": 3850
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.7351747851002866e-05,
+      "loss": 0.0434,
+      "step": 3851
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.7351060171919775e-05,
+      "loss": 0.0704,
+      "step": 3852
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.735037249283668e-05,
+      "loss": 0.0277,
+      "step": 3853
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.7349684813753586e-05,
+      "loss": 0.0641,
+      "step": 3854
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.734899713467049e-05,
+      "loss": 0.0547,
+      "step": 3855
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.7348309455587397e-05,
+      "loss": 0.0531,
+      "step": 3856
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.73476217765043e-05,
+      "loss": 0.0718,
+      "step": 3857
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.734693409742121e-05,
+      "loss": 0.0273,
+      "step": 3858
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.734624641833811e-05,
+      "loss": 0.0602,
+      "step": 3859
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.734555873925502e-05,
+      "loss": 0.0525,
+      "step": 3860
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.734487106017192e-05,
+      "loss": 0.0762,
+      "step": 3861
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.734418338108883e-05,
+      "loss": 0.0494,
+      "step": 3862
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.734349570200573e-05,
+      "loss": 0.0397,
+      "step": 3863
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.734280802292264e-05,
+      "loss": 0.0536,
+      "step": 3864
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 5.734212034383954e-05,
+      "loss": 0.0663,
+      "step": 3865
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.734143266475645e-05,
+      "loss": 0.0657,
+      "step": 3866
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.7340744985673354e-05,
+      "loss": 0.0331,
+      "step": 3867
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.734005730659026e-05,
+      "loss": 0.0531,
+      "step": 3868
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.7339369627507165e-05,
+      "loss": 0.0623,
+      "step": 3869
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.7338681948424074e-05,
+      "loss": 0.0541,
+      "step": 3870
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.7337994269340976e-05,
+      "loss": 0.0456,
+      "step": 3871
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.7337306590257885e-05,
+      "loss": 0.0695,
+      "step": 3872
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.733661891117479e-05,
+      "loss": 0.0415,
+      "step": 3873
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.7335931232091696e-05,
+      "loss": 0.0591,
+      "step": 3874
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.73352435530086e-05,
+      "loss": 0.0511,
+      "step": 3875
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.7334555873925507e-05,
+      "loss": 0.0527,
+      "step": 3876
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.733386819484241e-05,
+      "loss": 0.0296,
+      "step": 3877
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.733318051575932e-05,
+      "loss": 0.0654,
+      "step": 3878
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.733249283667622e-05,
+      "loss": 0.0482,
+      "step": 3879
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.733180515759313e-05,
+      "loss": 0.0404,
+      "step": 3880
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.733111747851003e-05,
+      "loss": 0.0425,
+      "step": 3881
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 5.733042979942694e-05,
+      "loss": 0.0389,
+      "step": 3882
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.732974212034384e-05,
+      "loss": 0.0525,
+      "step": 3883
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.732905444126075e-05,
+      "loss": 0.0527,
+      "step": 3884
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.732836676217765e-05,
+      "loss": 0.0462,
+      "step": 3885
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.732767908309456e-05,
+      "loss": 0.0547,
+      "step": 3886
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.7326991404011464e-05,
+      "loss": 0.0438,
+      "step": 3887
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.732630372492837e-05,
+      "loss": 0.048,
+      "step": 3888
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.7325616045845275e-05,
+      "loss": 0.0443,
+      "step": 3889
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.7324928366762184e-05,
+      "loss": 0.0374,
+      "step": 3890
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.7324240687679086e-05,
+      "loss": 0.0923,
+      "step": 3891
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.7323553008595995e-05,
+      "loss": 0.0665,
+      "step": 3892
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.73228653295129e-05,
+      "loss": 0.0349,
+      "step": 3893
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.7322177650429806e-05,
+      "loss": 0.0563,
+      "step": 3894
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.732148997134671e-05,
+      "loss": 0.0437,
+      "step": 3895
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.7320802292263617e-05,
+      "loss": 0.0559,
+      "step": 3896
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.732011461318052e-05,
+      "loss": 0.0354,
+      "step": 3897
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.731942693409743e-05,
+      "loss": 0.0386,
+      "step": 3898
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.731873925501433e-05,
+      "loss": 0.0637,
+      "step": 3899
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 5.731805157593124e-05,
+      "loss": 0.0297,
+      "step": 3900
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.731736389684814e-05,
+      "loss": 0.0201,
+      "step": 3901
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.731667621776505e-05,
+      "loss": 0.0451,
+      "step": 3902
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.731598853868195e-05,
+      "loss": 0.0417,
+      "step": 3903
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.731530085959886e-05,
+      "loss": 0.0316,
+      "step": 3904
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.731461318051576e-05,
+      "loss": 0.0408,
+      "step": 3905
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.731392550143267e-05,
+      "loss": 0.0674,
+      "step": 3906
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.7313237822349574e-05,
+      "loss": 0.0369,
+      "step": 3907
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.731255014326648e-05,
+      "loss": 0.0464,
+      "step": 3908
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.7311862464183385e-05,
+      "loss": 0.0564,
+      "step": 3909
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.7311174785100294e-05,
+      "loss": 0.0717,
+      "step": 3910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.7310487106017196e-05,
+      "loss": 0.0848,
+      "step": 3911
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.7309799426934105e-05,
+      "loss": 0.0372,
+      "step": 3912
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.730911174785101e-05,
+      "loss": 0.0691,
+      "step": 3913
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.7308424068767916e-05,
+      "loss": 0.06,
+      "step": 3914
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.730773638968482e-05,
+      "loss": 0.0626,
+      "step": 3915
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.7307048710601727e-05,
+      "loss": 0.0583,
+      "step": 3916
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 5.730636103151863e-05,
+      "loss": 0.0381,
+      "step": 3917
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.730567335243554e-05,
+      "loss": 0.0502,
+      "step": 3918
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.730498567335244e-05,
+      "loss": 0.0525,
+      "step": 3919
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.730429799426935e-05,
+      "loss": 0.0386,
+      "step": 3920
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.730361031518625e-05,
+      "loss": 0.043,
+      "step": 3921
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.730292263610316e-05,
+      "loss": 0.0476,
+      "step": 3922
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.730223495702006e-05,
+      "loss": 0.0227,
+      "step": 3923
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.7301547277936964e-05,
+      "loss": 0.0596,
+      "step": 3924
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.730085959885387e-05,
+      "loss": 0.0823,
+      "step": 3925
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.7300171919770775e-05,
+      "loss": 0.0274,
+      "step": 3926
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.7299484240687684e-05,
+      "loss": 0.0518,
+      "step": 3927
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.7298796561604586e-05,
+      "loss": 0.0342,
+      "step": 3928
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.7298108882521495e-05,
+      "loss": 0.0301,
+      "step": 3929
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.72974212034384e-05,
+      "loss": 0.0584,
+      "step": 3930
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.7296733524355306e-05,
+      "loss": 0.0522,
+      "step": 3931
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.729604584527221e-05,
+      "loss": 0.0353,
+      "step": 3932
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.729535816618912e-05,
+      "loss": 0.0867,
+      "step": 3933
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 5.729467048710602e-05,
+      "loss": 0.0409,
+      "step": 3934
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.729398280802293e-05,
+      "loss": 0.0752,
+      "step": 3935
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.729329512893983e-05,
+      "loss": 0.0599,
+      "step": 3936
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.729260744985674e-05,
+      "loss": 0.0652,
+      "step": 3937
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.729191977077364e-05,
+      "loss": 0.0628,
+      "step": 3938
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.729123209169055e-05,
+      "loss": 0.035,
+      "step": 3939
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.729054441260745e-05,
+      "loss": 0.0338,
+      "step": 3940
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.728985673352436e-05,
+      "loss": 0.0514,
+      "step": 3941
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.728916905444126e-05,
+      "loss": 0.0448,
+      "step": 3942
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.7288481375358165e-05,
+      "loss": 0.0566,
+      "step": 3943
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.7287793696275074e-05,
+      "loss": 0.0367,
+      "step": 3944
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.7287106017191976e-05,
+      "loss": 0.0618,
+      "step": 3945
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.728641833810888e-05,
+      "loss": 0.0621,
+      "step": 3946
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.728573065902579e-05,
+      "loss": 0.0351,
+      "step": 3947
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.728504297994269e-05,
+      "loss": 0.0746,
+      "step": 3948
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.72843553008596e-05,
+      "loss": 0.125,
+      "step": 3949
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.72836676217765e-05,
+      "loss": 0.0416,
+      "step": 3950
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.728297994269341e-05,
+      "loss": 0.0442,
+      "step": 3951
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 5.728229226361031e-05,
+      "loss": 0.0409,
+      "step": 3952
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.728160458452722e-05,
+      "loss": 0.0442,
+      "step": 3953
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.728091690544412e-05,
+      "loss": 0.0427,
+      "step": 3954
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.728022922636103e-05,
+      "loss": 0.0407,
+      "step": 3955
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.727954154727793e-05,
+      "loss": 0.0609,
+      "step": 3956
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.727885386819484e-05,
+      "loss": 0.0533,
+      "step": 3957
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.7278166189111744e-05,
+      "loss": 0.0579,
+      "step": 3958
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.727747851002865e-05,
+      "loss": 0.0398,
+      "step": 3959
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.7276790830945555e-05,
+      "loss": 0.0597,
+      "step": 3960
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.7276103151862464e-05,
+      "loss": 0.0587,
+      "step": 3961
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.7275415472779366e-05,
+      "loss": 0.0413,
+      "step": 3962
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.7274727793696275e-05,
+      "loss": 0.0676,
+      "step": 3963
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.727404011461318e-05,
+      "loss": 0.0352,
+      "step": 3964
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.7273352435530086e-05,
+      "loss": 0.1451,
+      "step": 3965
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.727266475644699e-05,
+      "loss": 0.036,
+      "step": 3966
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.72719770773639e-05,
+      "loss": 0.0501,
+      "step": 3967
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.72712893982808e-05,
+      "loss": 0.066,
+      "step": 3968
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.727060171919771e-05,
+      "loss": 0.0679,
+      "step": 3969
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.726991404011461e-05,
+      "loss": 0.0335,
+      "step": 3970
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.726922636103152e-05,
+      "loss": 0.046,
+      "step": 3971
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.726853868194842e-05,
+      "loss": 0.0556,
+      "step": 3972
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.726785100286533e-05,
+      "loss": 0.0606,
+      "step": 3973
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.726716332378223e-05,
+      "loss": 0.066,
+      "step": 3974
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.726647564469914e-05,
+      "loss": 0.0464,
+      "step": 3975
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.726578796561604e-05,
+      "loss": 0.0423,
+      "step": 3976
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.726510028653295e-05,
+      "loss": 0.0607,
+      "step": 3977
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7264412607449854e-05,
+      "loss": 0.0399,
+      "step": 3978
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.726372492836676e-05,
+      "loss": 0.0732,
+      "step": 3979
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7263037249283665e-05,
+      "loss": 0.0382,
+      "step": 3980
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7262349570200574e-05,
+      "loss": 0.0405,
+      "step": 3981
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7261661891117476e-05,
+      "loss": 0.0443,
+      "step": 3982
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7260974212034385e-05,
+      "loss": 0.0312,
+      "step": 3983
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.726028653295129e-05,
+      "loss": 0.0392,
+      "step": 3984
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.7259598853868196e-05,
+      "loss": 0.0604,
+      "step": 3985
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.72589111747851e-05,
+      "loss": 0.0991,
+      "step": 3986
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.725822349570201e-05,
+      "loss": 0.0716,
+      "step": 3987
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725753581661891e-05,
+      "loss": 0.0555,
+      "step": 3988
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725684813753582e-05,
+      "loss": 0.0451,
+      "step": 3989
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725616045845272e-05,
+      "loss": 0.04,
+      "step": 3990
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725547277936963e-05,
+      "loss": 0.0283,
+      "step": 3991
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725478510028653e-05,
+      "loss": 0.0565,
+      "step": 3992
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725409742120344e-05,
+      "loss": 0.0535,
+      "step": 3993
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725340974212034e-05,
+      "loss": 0.0694,
+      "step": 3994
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725272206303725e-05,
+      "loss": 0.0468,
+      "step": 3995
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725203438395415e-05,
+      "loss": 0.0723,
+      "step": 3996
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.725134670487106e-05,
+      "loss": 0.0494,
+      "step": 3997
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.7250659025787964e-05,
+      "loss": 0.1003,
+      "step": 3998
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.724997134670487e-05,
+      "loss": 0.0327,
+      "step": 3999
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.7249283667621775e-05,
+      "loss": 0.0381,
+      "step": 4000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.7248595988538684e-05,
+      "loss": 0.0378,
+      "step": 4001
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.7247908309455586e-05,
+      "loss": 0.0506,
+      "step": 4002
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.7247220630372495e-05,
+      "loss": 0.0607,
+      "step": 4003
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.72465329512894e-05,
+      "loss": 0.0494,
+      "step": 4004
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.7245845272206306e-05,
+      "loss": 0.0367,
+      "step": 4005
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.724515759312321e-05,
+      "loss": 0.0408,
+      "step": 4006
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.724446991404012e-05,
+      "loss": 0.0383,
+      "step": 4007
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.724378223495702e-05,
+      "loss": 0.0879,
+      "step": 4008
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.724309455587393e-05,
+      "loss": 0.0464,
+      "step": 4009
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.724240687679083e-05,
+      "loss": 0.0878,
+      "step": 4010
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.724171919770774e-05,
+      "loss": 0.0515,
+      "step": 4011
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.724103151862464e-05,
+      "loss": 0.0302,
+      "step": 4012
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.724034383954155e-05,
+      "loss": 0.0574,
+      "step": 4013
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.723965616045845e-05,
+      "loss": 0.0339,
+      "step": 4014
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.723896848137536e-05,
+      "loss": 0.0669,
+      "step": 4015
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.723828080229226e-05,
+      "loss": 0.092,
+      "step": 4016
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.723759312320917e-05,
+      "loss": 0.0424,
+      "step": 4017
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.7236905444126074e-05,
+      "loss": 0.0679,
+      "step": 4018
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.723621776504298e-05,
+      "loss": 0.0563,
+      "step": 4019
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.7235530085959885e-05,
+      "loss": 0.0353,
+      "step": 4020
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.7234842406876794e-05,
+      "loss": 0.0367,
+      "step": 4021
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.7234154727793696e-05,
+      "loss": 0.0444,
+      "step": 4022
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7233467048710605e-05,
+      "loss": 0.0687,
+      "step": 4023
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.723277936962751e-05,
+      "loss": 0.0506,
+      "step": 4024
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7232091690544416e-05,
+      "loss": 0.0356,
+      "step": 4025
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.723140401146132e-05,
+      "loss": 0.0665,
+      "step": 4026
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.723071633237823e-05,
+      "loss": 0.057,
+      "step": 4027
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.723002865329513e-05,
+      "loss": 0.0349,
+      "step": 4028
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.722934097421204e-05,
+      "loss": 0.0247,
+      "step": 4029
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.722865329512894e-05,
+      "loss": 0.0505,
+      "step": 4030
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.722796561604585e-05,
+      "loss": 0.0405,
+      "step": 4031
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.722727793696275e-05,
+      "loss": 0.049,
+      "step": 4032
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.722659025787966e-05,
+      "loss": 0.0532,
+      "step": 4033
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.722590257879656e-05,
+      "loss": 0.051,
+      "step": 4034
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.722521489971347e-05,
+      "loss": 0.0516,
+      "step": 4035
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.722452722063037e-05,
+      "loss": 0.036,
+      "step": 4036
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.722383954154728e-05,
+      "loss": 0.0652,
+      "step": 4037
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.7223151862464184e-05,
+      "loss": 0.043,
+      "step": 4038
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.722246418338109e-05,
+      "loss": 0.0306,
+      "step": 4039
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.7221776504297995e-05,
+      "loss": 0.0427,
+      "step": 4040
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.7221088825214904e-05,
+      "loss": 0.0739,
+      "step": 4041
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.7220401146131806e-05,
+      "loss": 0.0464,
+      "step": 4042
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721971346704871e-05,
+      "loss": 0.0588,
+      "step": 4043
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721902578796562e-05,
+      "loss": 0.048,
+      "step": 4044
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721833810888252e-05,
+      "loss": 0.0728,
+      "step": 4045
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721765042979943e-05,
+      "loss": 0.0451,
+      "step": 4046
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721696275071633e-05,
+      "loss": 0.0483,
+      "step": 4047
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721627507163324e-05,
+      "loss": 0.033,
+      "step": 4048
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721558739255014e-05,
+      "loss": 0.0725,
+      "step": 4049
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721489971346705e-05,
+      "loss": 0.0291,
+      "step": 4050
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721421203438395e-05,
+      "loss": 0.0456,
+      "step": 4051
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721352435530086e-05,
+      "loss": 0.0309,
+      "step": 4052
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721283667621776e-05,
+      "loss": 0.0615,
+      "step": 4053
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721214899713467e-05,
+      "loss": 0.088,
+      "step": 4054
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.7211461318051574e-05,
+      "loss": 0.0565,
+      "step": 4055
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.721077363896848e-05,
+      "loss": 0.0315,
+      "step": 4056
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.7210085959885385e-05,
+      "loss": 0.0287,
+      "step": 4057
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7209398280802294e-05,
+      "loss": 0.0336,
+      "step": 4058
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7208710601719196e-05,
+      "loss": 0.0665,
+      "step": 4059
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7208022922636105e-05,
+      "loss": 0.0447,
+      "step": 4060
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720733524355301e-05,
+      "loss": 0.052,
+      "step": 4061
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7206647564469916e-05,
+      "loss": 0.0405,
+      "step": 4062
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720595988538682e-05,
+      "loss": 0.0714,
+      "step": 4063
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720527220630373e-05,
+      "loss": 0.0467,
+      "step": 4064
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720458452722063e-05,
+      "loss": 0.0307,
+      "step": 4065
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720389684813754e-05,
+      "loss": 0.0695,
+      "step": 4066
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720320916905444e-05,
+      "loss": 0.0762,
+      "step": 4067
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720252148997135e-05,
+      "loss": 0.0971,
+      "step": 4068
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720183381088825e-05,
+      "loss": 0.0511,
+      "step": 4069
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720114613180516e-05,
+      "loss": 0.0335,
+      "step": 4070
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.720045845272206e-05,
+      "loss": 0.0511,
+      "step": 4071
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.719977077363897e-05,
+      "loss": 0.0474,
+      "step": 4072
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.719908309455587e-05,
+      "loss": 0.0321,
+      "step": 4073
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.719839541547278e-05,
+      "loss": 0.0526,
+      "step": 4074
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7197707736389684e-05,
+      "loss": 0.0361,
+      "step": 4075
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.719702005730659e-05,
+      "loss": 0.0524,
+      "step": 4076
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7196332378223495e-05,
+      "loss": 0.0477,
+      "step": 4077
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7195644699140404e-05,
+      "loss": 0.0573,
+      "step": 4078
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7194957020057306e-05,
+      "loss": 0.0333,
+      "step": 4079
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7194269340974215e-05,
+      "loss": 0.0838,
+      "step": 4080
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.719358166189112e-05,
+      "loss": 0.0665,
+      "step": 4081
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.7192893982808026e-05,
+      "loss": 0.0415,
+      "step": 4082
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.719220630372493e-05,
+      "loss": 0.0693,
+      "step": 4083
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.719151862464184e-05,
+      "loss": 0.0607,
+      "step": 4084
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.719083094555874e-05,
+      "loss": 0.0422,
+      "step": 4085
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.719014326647565e-05,
+      "loss": 0.0547,
+      "step": 4086
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.718945558739255e-05,
+      "loss": 0.0908,
+      "step": 4087
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.718876790830946e-05,
+      "loss": 0.0272,
+      "step": 4088
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.718808022922636e-05,
+      "loss": 0.0644,
+      "step": 4089
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.718739255014327e-05,
+      "loss": 0.0936,
+      "step": 4090
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.718670487106017e-05,
+      "loss": 0.0601,
+      "step": 4091
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.718601719197708e-05,
+      "loss": 0.0282,
+      "step": 4092
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.718532951289398e-05,
+      "loss": 0.071,
+      "step": 4093
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.718464183381089e-05,
+      "loss": 0.0613,
+      "step": 4094
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.7183954154727794e-05,
+      "loss": 0.0633,
+      "step": 4095
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.71832664756447e-05,
+      "loss": 0.045,
+      "step": 4096
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.7182578796561605e-05,
+      "loss": 0.045,
+      "step": 4097
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.7181891117478514e-05,
+      "loss": 0.0359,
+      "step": 4098
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.7181203438395416e-05,
+      "loss": 0.0691,
+      "step": 4099
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.7180515759312325e-05,
+      "loss": 0.0476,
+      "step": 4100
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.717982808022923e-05,
+      "loss": 0.0554,
+      "step": 4101
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.7179140401146136e-05,
+      "loss": 0.0661,
+      "step": 4102
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.717845272206304e-05,
+      "loss": 0.0463,
+      "step": 4103
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.717776504297995e-05,
+      "loss": 0.0847,
+      "step": 4104
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.717707736389685e-05,
+      "loss": 0.0627,
+      "step": 4105
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.717638968481376e-05,
+      "loss": 0.0321,
+      "step": 4106
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.717570200573066e-05,
+      "loss": 0.0693,
+      "step": 4107
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.717501432664757e-05,
+      "loss": 0.0506,
+      "step": 4108
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.717432664756447e-05,
+      "loss": 0.0656,
+      "step": 4109
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.717363896848138e-05,
+      "loss": 0.0378,
+      "step": 4110
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.717295128939828e-05,
+      "loss": 0.0489,
+      "step": 4111
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.717226361031519e-05,
+      "loss": 0.1015,
+      "step": 4112
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.717157593123209e-05,
+      "loss": 0.044,
+      "step": 4113
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.7170888252149e-05,
+      "loss": 0.0493,
+      "step": 4114
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.7170200573065904e-05,
+      "loss": 0.0378,
+      "step": 4115
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.716951289398281e-05,
+      "loss": 0.0985,
+      "step": 4116
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.7168825214899715e-05,
+      "loss": 0.0425,
+      "step": 4117
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.7168137535816624e-05,
+      "loss": 0.0607,
+      "step": 4118
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.7167449856733526e-05,
+      "loss": 0.0442,
+      "step": 4119
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.7166762177650435e-05,
+      "loss": 0.0439,
+      "step": 4120
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.716607449856734e-05,
+      "loss": 0.0485,
+      "step": 4121
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.7165386819484246e-05,
+      "loss": 0.0424,
+      "step": 4122
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.716469914040115e-05,
+      "loss": 0.0441,
+      "step": 4123
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.716401146131806e-05,
+      "loss": 0.0491,
+      "step": 4124
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.716332378223496e-05,
+      "loss": 0.085,
+      "step": 4125
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.716263610315187e-05,
+      "loss": 0.0424,
+      "step": 4126
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.716194842406877e-05,
+      "loss": 0.0426,
+      "step": 4127
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.716126074498568e-05,
+      "loss": 0.044,
+      "step": 4128
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.716057306590258e-05,
+      "loss": 0.0419,
+      "step": 4129
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.715988538681949e-05,
+      "loss": 0.0531,
+      "step": 4130
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.715919770773639e-05,
+      "loss": 0.057,
+      "step": 4131
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.71585100286533e-05,
+      "loss": 0.0458,
+      "step": 4132
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.71578223495702e-05,
+      "loss": 0.0402,
+      "step": 4133
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.715713467048711e-05,
+      "loss": 0.0382,
+      "step": 4134
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.7156446991404014e-05,
+      "loss": 0.0483,
+      "step": 4135
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.715575931232092e-05,
+      "loss": 0.0295,
+      "step": 4136
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.7155071633237825e-05,
+      "loss": 0.0619,
+      "step": 4137
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.7154383954154734e-05,
+      "loss": 0.0643,
+      "step": 4138
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.7153696275071636e-05,
+      "loss": 0.0672,
+      "step": 4139
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.7153008595988545e-05,
+      "loss": 0.0635,
+      "step": 4140
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.715232091690545e-05,
+      "loss": 0.0388,
+      "step": 4141
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.7151633237822356e-05,
+      "loss": 0.0721,
+      "step": 4142
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.715094555873926e-05,
+      "loss": 0.0385,
+      "step": 4143
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.715025787965617e-05,
+      "loss": 0.0724,
+      "step": 4144
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.714957020057307e-05,
+      "loss": 0.0441,
+      "step": 4145
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.714888252148998e-05,
+      "loss": 0.0543,
+      "step": 4146
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.714819484240688e-05,
+      "loss": 0.0805,
+      "step": 4147
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.714750716332379e-05,
+      "loss": 0.0421,
+      "step": 4148
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.714681948424069e-05,
+      "loss": 0.0908,
+      "step": 4149
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.71461318051576e-05,
+      "loss": 0.0585,
+      "step": 4150
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.71454441260745e-05,
+      "loss": 0.0642,
+      "step": 4151
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.714475644699141e-05,
+      "loss": 0.0235,
+      "step": 4152
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.714406876790831e-05,
+      "loss": 0.0415,
+      "step": 4153
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.714338108882522e-05,
+      "loss": 0.0442,
+      "step": 4154
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.7142693409742124e-05,
+      "loss": 0.0405,
+      "step": 4155
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.714200573065903e-05,
+      "loss": 0.0541,
+      "step": 4156
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.7141318051575935e-05,
+      "loss": 0.0413,
+      "step": 4157
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.7140630372492844e-05,
+      "loss": 0.034,
+      "step": 4158
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.7139942693409746e-05,
+      "loss": 0.1165,
+      "step": 4159
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.7139255014326655e-05,
+      "loss": 0.0638,
+      "step": 4160
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.713856733524356e-05,
+      "loss": 0.0499,
+      "step": 4161
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.713787965616046e-05,
+      "loss": 0.0529,
+      "step": 4162
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.713719197707737e-05,
+      "loss": 0.0295,
+      "step": 4163
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.713650429799427e-05,
+      "loss": 0.0916,
+      "step": 4164
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.713581661891118e-05,
+      "loss": 0.0411,
+      "step": 4165
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.713512893982808e-05,
+      "loss": 0.0589,
+      "step": 4166
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.713444126074499e-05,
+      "loss": 0.0874,
+      "step": 4167
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.713375358166189e-05,
+      "loss": 0.0506,
+      "step": 4168
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.71330659025788e-05,
+      "loss": 0.0311,
+      "step": 4169
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.71323782234957e-05,
+      "loss": 0.0359,
+      "step": 4170
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.713169054441261e-05,
+      "loss": 0.0517,
+      "step": 4171
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.7131002865329514e-05,
+      "loss": 0.0648,
+      "step": 4172
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.713031518624642e-05,
+      "loss": 0.0673,
+      "step": 4173
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.7129627507163325e-05,
+      "loss": 0.0494,
+      "step": 4174
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.7128939828080234e-05,
+      "loss": 0.0935,
+      "step": 4175
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.7128252148997136e-05,
+      "loss": 0.0398,
+      "step": 4176
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.7127564469914045e-05,
+      "loss": 0.0174,
+      "step": 4177
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.712687679083095e-05,
+      "loss": 0.058,
+      "step": 4178
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.7126189111747856e-05,
+      "loss": 0.0325,
+      "step": 4179
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.712550143266476e-05,
+      "loss": 0.0663,
+      "step": 4180
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.712481375358167e-05,
+      "loss": 0.0621,
+      "step": 4181
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.712412607449857e-05,
+      "loss": 0.0562,
+      "step": 4182
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.712343839541548e-05,
+      "loss": 0.0425,
+      "step": 4183
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.712275071633238e-05,
+      "loss": 0.0701,
+      "step": 4184
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.712206303724929e-05,
+      "loss": 0.0513,
+      "step": 4185
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.712137535816619e-05,
+      "loss": 0.0601,
+      "step": 4186
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.71206876790831e-05,
+      "loss": 0.0671,
+      "step": 4187
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.712e-05,
+      "loss": 0.0678,
+      "step": 4188
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.711931232091691e-05,
+      "loss": 0.0719,
+      "step": 4189
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.711862464183381e-05,
+      "loss": 0.0694,
+      "step": 4190
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.711793696275072e-05,
+      "loss": 0.0669,
+      "step": 4191
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.7117249283667624e-05,
+      "loss": 0.0343,
+      "step": 4192
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.711656160458453e-05,
+      "loss": 0.0475,
+      "step": 4193
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.7115873925501435e-05,
+      "loss": 0.0284,
+      "step": 4194
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.7115186246418344e-05,
+      "loss": 0.0535,
+      "step": 4195
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.7114498567335246e-05,
+      "loss": 0.045,
+      "step": 4196
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.711381088825215e-05,
+      "loss": 0.0458,
+      "step": 4197
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.711312320916906e-05,
+      "loss": 0.0304,
+      "step": 4198
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.711243553008596e-05,
+      "loss": 0.0345,
+      "step": 4199
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.711174785100286e-05,
+      "loss": 0.0411,
+      "step": 4200
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.711106017191977e-05,
+      "loss": 0.0378,
+      "step": 4201
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.711037249283667e-05,
+      "loss": 0.0907,
+      "step": 4202
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.710968481375358e-05,
+      "loss": 0.0624,
+      "step": 4203
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.710899713467048e-05,
+      "loss": 0.0894,
+      "step": 4204
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.710830945558739e-05,
+      "loss": 0.0491,
+      "step": 4205
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.7107621776504294e-05,
+      "loss": 0.0387,
+      "step": 4206
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.71069340974212e-05,
+      "loss": 0.0326,
+      "step": 4207
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.7106246418338105e-05,
+      "loss": 0.0368,
+      "step": 4208
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.7105558739255014e-05,
+      "loss": 0.0646,
+      "step": 4209
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.7104871060171916e-05,
+      "loss": 0.0463,
+      "step": 4210
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.7104183381088825e-05,
+      "loss": 0.0275,
+      "step": 4211
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.710349570200573e-05,
+      "loss": 0.0346,
+      "step": 4212
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.7102808022922636e-05,
+      "loss": 0.051,
+      "step": 4213
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.710212034383954e-05,
+      "loss": 0.0491,
+      "step": 4214
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.710143266475645e-05,
+      "loss": 0.0306,
+      "step": 4215
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.710074498567335e-05,
+      "loss": 0.0528,
+      "step": 4216
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.710005730659026e-05,
+      "loss": 0.0638,
+      "step": 4217
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.709936962750716e-05,
+      "loss": 0.0258,
+      "step": 4218
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.709868194842407e-05,
+      "loss": 0.073,
+      "step": 4219
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.709799426934097e-05,
+      "loss": 0.0368,
+      "step": 4220
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.709730659025788e-05,
+      "loss": 0.043,
+      "step": 4221
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.709661891117478e-05,
+      "loss": 0.0595,
+      "step": 4222
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.709593123209169e-05,
+      "loss": 0.0319,
+      "step": 4223
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.709524355300859e-05,
+      "loss": 0.055,
+      "step": 4224
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.70945558739255e-05,
+      "loss": 0.0583,
+      "step": 4225
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.7093868194842404e-05,
+      "loss": 0.0453,
+      "step": 4226
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.709318051575931e-05,
+      "loss": 0.0402,
+      "step": 4227
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.7092492836676215e-05,
+      "loss": 0.0661,
+      "step": 4228
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.7091805157593124e-05,
+      "loss": 0.089,
+      "step": 4229
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.7091117478510026e-05,
+      "loss": 0.0439,
+      "step": 4230
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.7090429799426935e-05,
+      "loss": 0.0617,
+      "step": 4231
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708974212034384e-05,
+      "loss": 0.0735,
+      "step": 4232
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.7089054441260746e-05,
+      "loss": 0.095,
+      "step": 4233
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708836676217765e-05,
+      "loss": 0.0415,
+      "step": 4234
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708767908309456e-05,
+      "loss": 0.0536,
+      "step": 4235
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708699140401146e-05,
+      "loss": 0.0534,
+      "step": 4236
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708630372492837e-05,
+      "loss": 0.0382,
+      "step": 4237
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708561604584527e-05,
+      "loss": 0.0443,
+      "step": 4238
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708492836676218e-05,
+      "loss": 0.0516,
+      "step": 4239
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708424068767908e-05,
+      "loss": 0.0457,
+      "step": 4240
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708355300859599e-05,
+      "loss": 0.0732,
+      "step": 4241
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708286532951289e-05,
+      "loss": 0.0633,
+      "step": 4242
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.70821776504298e-05,
+      "loss": 0.0591,
+      "step": 4243
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.70814899713467e-05,
+      "loss": 0.0746,
+      "step": 4244
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.708080229226361e-05,
+      "loss": 0.0319,
+      "step": 4245
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.7080114613180514e-05,
+      "loss": 0.0509,
+      "step": 4246
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.707942693409742e-05,
+      "loss": 0.0432,
+      "step": 4247
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.7078739255014325e-05,
+      "loss": 0.0299,
+      "step": 4248
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.7078051575931234e-05,
+      "loss": 0.0494,
+      "step": 4249
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.7077363896848136e-05,
+      "loss": 0.0482,
+      "step": 4250
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.7076676217765045e-05,
+      "loss": 0.0503,
+      "step": 4251
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.707598853868195e-05,
+      "loss": 0.0386,
+      "step": 4252
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.7075300859598856e-05,
+      "loss": 0.053,
+      "step": 4253
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.707461318051576e-05,
+      "loss": 0.0522,
+      "step": 4254
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.707392550143267e-05,
+      "loss": 0.0309,
+      "step": 4255
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.707323782234957e-05,
+      "loss": 0.0438,
+      "step": 4256
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.707255014326648e-05,
+      "loss": 0.0306,
+      "step": 4257
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.707186246418338e-05,
+      "loss": 0.0301,
+      "step": 4258
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.707117478510029e-05,
+      "loss": 0.0528,
+      "step": 4259
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.707048710601719e-05,
+      "loss": 0.0443,
+      "step": 4260
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.70697994269341e-05,
+      "loss": 0.0606,
+      "step": 4261
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.7069111747851e-05,
+      "loss": 0.0508,
+      "step": 4262
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.706842406876791e-05,
+      "loss": 0.0361,
+      "step": 4263
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.706773638968481e-05,
+      "loss": 0.0455,
+      "step": 4264
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.706704871060172e-05,
+      "loss": 0.0407,
+      "step": 4265
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.7066361031518624e-05,
+      "loss": 0.0484,
+      "step": 4266
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.706567335243553e-05,
+      "loss": 0.0974,
+      "step": 4267
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.7064985673352435e-05,
+      "loss": 0.0236,
+      "step": 4268
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.7064297994269344e-05,
+      "loss": 0.0796,
+      "step": 4269
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.7063610315186246e-05,
+      "loss": 0.0794,
+      "step": 4270
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.7062922636103155e-05,
+      "loss": 0.0439,
+      "step": 4271
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.706223495702006e-05,
+      "loss": 0.0576,
+      "step": 4272
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.7061547277936966e-05,
+      "loss": 0.0489,
+      "step": 4273
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.706085959885387e-05,
+      "loss": 0.049,
+      "step": 4274
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.706017191977078e-05,
+      "loss": 0.0464,
+      "step": 4275
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.705948424068768e-05,
+      "loss": 0.0555,
+      "step": 4276
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.705879656160459e-05,
+      "loss": 0.095,
+      "step": 4277
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.705810888252149e-05,
+      "loss": 0.0426,
+      "step": 4278
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.70574212034384e-05,
+      "loss": 0.0703,
+      "step": 4279
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.70567335243553e-05,
+      "loss": 0.055,
+      "step": 4280
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.70560458452722e-05,
+      "loss": 0.0572,
+      "step": 4281
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.705535816618911e-05,
+      "loss": 0.0536,
+      "step": 4282
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.7054670487106014e-05,
+      "loss": 0.0472,
+      "step": 4283
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.705398280802292e-05,
+      "loss": 0.0984,
+      "step": 4284
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.7053295128939825e-05,
+      "loss": 0.0685,
+      "step": 4285
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.7052607449856734e-05,
+      "loss": 0.0417,
+      "step": 4286
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.7051919770773636e-05,
+      "loss": 0.0927,
+      "step": 4287
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.7051232091690545e-05,
+      "loss": 0.0511,
+      "step": 4288
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.705054441260745e-05,
+      "loss": 0.06,
+      "step": 4289
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.7049856733524356e-05,
+      "loss": 0.0535,
+      "step": 4290
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.704916905444126e-05,
+      "loss": 0.0454,
+      "step": 4291
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.704848137535817e-05,
+      "loss": 0.0618,
+      "step": 4292
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.704779369627507e-05,
+      "loss": 0.054,
+      "step": 4293
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.704710601719198e-05,
+      "loss": 0.0639,
+      "step": 4294
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.704641833810888e-05,
+      "loss": 0.0503,
+      "step": 4295
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.704573065902579e-05,
+      "loss": 0.0769,
+      "step": 4296
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.704504297994269e-05,
+      "loss": 0.0608,
+      "step": 4297
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.70443553008596e-05,
+      "loss": 0.0562,
+      "step": 4298
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.70436676217765e-05,
+      "loss": 0.0587,
+      "step": 4299
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.704297994269341e-05,
+      "loss": 0.0443,
+      "step": 4300
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.704229226361031e-05,
+      "loss": 0.0498,
+      "step": 4301
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.704160458452722e-05,
+      "loss": 0.0425,
+      "step": 4302
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.7040916905444124e-05,
+      "loss": 0.1045,
+      "step": 4303
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.704022922636103e-05,
+      "loss": 0.046,
+      "step": 4304
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.7039541547277935e-05,
+      "loss": 0.044,
+      "step": 4305
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.7038853868194844e-05,
+      "loss": 0.0402,
+      "step": 4306
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.7038166189111746e-05,
+      "loss": 0.0581,
+      "step": 4307
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.7037478510028655e-05,
+      "loss": 0.0465,
+      "step": 4308
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.703679083094556e-05,
+      "loss": 0.06,
+      "step": 4309
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.7036103151862466e-05,
+      "loss": 0.044,
+      "step": 4310
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.703541547277937e-05,
+      "loss": 0.0709,
+      "step": 4311
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.703472779369628e-05,
+      "loss": 0.0919,
+      "step": 4312
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.703404011461318e-05,
+      "loss": 0.0435,
+      "step": 4313
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.703335243553009e-05,
+      "loss": 0.0577,
+      "step": 4314
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.703266475644699e-05,
+      "loss": 0.0504,
+      "step": 4315
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.70319770773639e-05,
+      "loss": 0.0273,
+      "step": 4316
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.70312893982808e-05,
+      "loss": 0.0517,
+      "step": 4317
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.703060171919771e-05,
+      "loss": 0.0892,
+      "step": 4318
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.702991404011461e-05,
+      "loss": 0.0779,
+      "step": 4319
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.702922636103152e-05,
+      "loss": 0.0575,
+      "step": 4320
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.702853868194842e-05,
+      "loss": 0.0788,
+      "step": 4321
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.702785100286533e-05,
+      "loss": 0.0464,
+      "step": 4322
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7027163323782234e-05,
+      "loss": 0.0413,
+      "step": 4323
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.702647564469914e-05,
+      "loss": 0.0568,
+      "step": 4324
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7025787965616045e-05,
+      "loss": 0.0289,
+      "step": 4325
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7025100286532954e-05,
+      "loss": 0.0478,
+      "step": 4326
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7024412607449856e-05,
+      "loss": 0.0692,
+      "step": 4327
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7023724928366765e-05,
+      "loss": 0.0568,
+      "step": 4328
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.702303724928367e-05,
+      "loss": 0.0467,
+      "step": 4329
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7022349570200576e-05,
+      "loss": 0.0455,
+      "step": 4330
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.702166189111748e-05,
+      "loss": 0.0455,
+      "step": 4331
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.702097421203439e-05,
+      "loss": 0.0715,
+      "step": 4332
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.702028653295129e-05,
+      "loss": 0.0332,
+      "step": 4333
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.70195988538682e-05,
+      "loss": 0.0247,
+      "step": 4334
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.70189111747851e-05,
+      "loss": 0.0443,
+      "step": 4335
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.701822349570201e-05,
+      "loss": 0.0373,
+      "step": 4336
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.701753581661891e-05,
+      "loss": 0.0373,
+      "step": 4337
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.701684813753582e-05,
+      "loss": 0.0445,
+      "step": 4338
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.701616045845272e-05,
+      "loss": 0.0514,
+      "step": 4339
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.701547277936963e-05,
+      "loss": 0.0386,
+      "step": 4340
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.701478510028653e-05,
+      "loss": 0.0337,
+      "step": 4341
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.701409742120344e-05,
+      "loss": 0.0412,
+      "step": 4342
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7013409742120344e-05,
+      "loss": 0.0602,
+      "step": 4343
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.701272206303725e-05,
+      "loss": 0.0498,
+      "step": 4344
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7012034383954155e-05,
+      "loss": 0.0523,
+      "step": 4345
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7011346704871064e-05,
+      "loss": 0.0637,
+      "step": 4346
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7010659025787966e-05,
+      "loss": 0.0457,
+      "step": 4347
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7009971346704875e-05,
+      "loss": 0.0433,
+      "step": 4348
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.700928366762178e-05,
+      "loss": 0.0305,
+      "step": 4349
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.7008595988538686e-05,
+      "loss": 0.0818,
+      "step": 4350
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.700790830945559e-05,
+      "loss": 0.0603,
+      "step": 4351
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.70072206303725e-05,
+      "loss": 0.0457,
+      "step": 4352
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.70065329512894e-05,
+      "loss": 0.0587,
+      "step": 4353
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.700584527220631e-05,
+      "loss": 0.0452,
+      "step": 4354
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.700515759312321e-05,
+      "loss": 0.0558,
+      "step": 4355
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.700446991404012e-05,
+      "loss": 0.043,
+      "step": 4356
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.700378223495702e-05,
+      "loss": 0.0386,
+      "step": 4357
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.700309455587393e-05,
+      "loss": 0.0389,
+      "step": 4358
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.700240687679083e-05,
+      "loss": 0.0443,
+      "step": 4359
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.700171919770774e-05,
+      "loss": 0.0243,
+      "step": 4360
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.700103151862464e-05,
+      "loss": 0.0413,
+      "step": 4361
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.700034383954155e-05,
+      "loss": 0.0454,
+      "step": 4362
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6999656160458454e-05,
+      "loss": 0.0379,
+      "step": 4363
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.699896848137536e-05,
+      "loss": 0.0512,
+      "step": 4364
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6998280802292265e-05,
+      "loss": 0.0368,
+      "step": 4365
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6997593123209174e-05,
+      "loss": 0.0658,
+      "step": 4366
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6996905444126076e-05,
+      "loss": 0.1034,
+      "step": 4367
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6996217765042985e-05,
+      "loss": 0.0952,
+      "step": 4368
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.699553008595989e-05,
+      "loss": 0.0547,
+      "step": 4369
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.6994842406876796e-05,
+      "loss": 0.0526,
+      "step": 4370
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.69941547277937e-05,
+      "loss": 0.0466,
+      "step": 4371
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.699346704871061e-05,
+      "loss": 0.0537,
+      "step": 4372
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.699277936962751e-05,
+      "loss": 0.0413,
+      "step": 4373
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.699209169054442e-05,
+      "loss": 0.051,
+      "step": 4374
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.699140401146132e-05,
+      "loss": 0.0543,
+      "step": 4375
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.699071633237823e-05,
+      "loss": 0.081,
+      "step": 4376
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.699002865329513e-05,
+      "loss": 0.0655,
+      "step": 4377
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.698934097421204e-05,
+      "loss": 0.0635,
+      "step": 4378
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.698865329512894e-05,
+      "loss": 0.0413,
+      "step": 4379
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.698796561604585e-05,
+      "loss": 0.0446,
+      "step": 4380
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.698727793696275e-05,
+      "loss": 0.0594,
+      "step": 4381
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.698659025787966e-05,
+      "loss": 0.0365,
+      "step": 4382
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.6985902578796564e-05,
+      "loss": 0.0326,
+      "step": 4383
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.698521489971347e-05,
+      "loss": 0.0834,
+      "step": 4384
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.6984527220630375e-05,
+      "loss": 0.0684,
+      "step": 4385
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.6983839541547284e-05,
+      "loss": 0.0665,
+      "step": 4386
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.6983151862464186e-05,
+      "loss": 0.0548,
+      "step": 4387
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.6982464183381095e-05,
+      "loss": 0.0553,
+      "step": 4388
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.6981776504298e-05,
+      "loss": 0.049,
+      "step": 4389
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.6981088825214906e-05,
+      "loss": 0.0678,
+      "step": 4390
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.698040114613181e-05,
+      "loss": 0.0435,
+      "step": 4391
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.697971346704872e-05,
+      "loss": 0.0448,
+      "step": 4392
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.697902578796562e-05,
+      "loss": 0.0381,
+      "step": 4393
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.697833810888253e-05,
+      "loss": 0.0353,
+      "step": 4394
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.697765042979943e-05,
+      "loss": 0.0626,
+      "step": 4395
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.697696275071634e-05,
+      "loss": 0.0688,
+      "step": 4396
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.697627507163324e-05,
+      "loss": 0.0309,
+      "step": 4397
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.697558739255015e-05,
+      "loss": 0.0647,
+      "step": 4398
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.697489971346705e-05,
+      "loss": 0.0443,
+      "step": 4399
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.6974212034383954e-05,
+      "loss": 0.0556,
+      "step": 4400
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.697352435530086e-05,
+      "loss": 0.062,
+      "step": 4401
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.6972836676217765e-05,
+      "loss": 0.0488,
+      "step": 4402
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.6972148997134674e-05,
+      "loss": 0.0502,
+      "step": 4403
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.6971461318051576e-05,
+      "loss": 0.0625,
+      "step": 4404
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.6970773638968485e-05,
+      "loss": 0.0654,
+      "step": 4405
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.697008595988539e-05,
+      "loss": 0.04,
+      "step": 4406
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.6969398280802296e-05,
+      "loss": 0.0498,
+      "step": 4407
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.69687106017192e-05,
+      "loss": 0.0414,
+      "step": 4408
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.696802292263611e-05,
+      "loss": 0.0645,
+      "step": 4409
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.696733524355301e-05,
+      "loss": 0.0459,
+      "step": 4410
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.696664756446992e-05,
+      "loss": 0.0505,
+      "step": 4411
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.696595988538682e-05,
+      "loss": 0.0404,
+      "step": 4412
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.696527220630373e-05,
+      "loss": 0.0374,
+      "step": 4413
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.696458452722063e-05,
+      "loss": 0.0475,
+      "step": 4414
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.696389684813754e-05,
+      "loss": 0.056,
+      "step": 4415
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.696320916905444e-05,
+      "loss": 0.0929,
+      "step": 4416
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.696252148997135e-05,
+      "loss": 0.0451,
+      "step": 4417
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.696183381088825e-05,
+      "loss": 0.0244,
+      "step": 4418
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.696114613180516e-05,
+      "loss": 0.0437,
+      "step": 4419
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.6960458452722064e-05,
+      "loss": 0.0405,
+      "step": 4420
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.695977077363897e-05,
+      "loss": 0.0658,
+      "step": 4421
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.6959083094555875e-05,
+      "loss": 0.0609,
+      "step": 4422
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.6958395415472784e-05,
+      "loss": 0.0323,
+      "step": 4423
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6957707736389686e-05,
+      "loss": 0.0767,
+      "step": 4424
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6957020057306595e-05,
+      "loss": 0.0661,
+      "step": 4425
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.69563323782235e-05,
+      "loss": 0.0444,
+      "step": 4426
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6955644699140406e-05,
+      "loss": 0.0378,
+      "step": 4427
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.695495702005731e-05,
+      "loss": 0.0649,
+      "step": 4428
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.695426934097422e-05,
+      "loss": 0.0488,
+      "step": 4429
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.695358166189112e-05,
+      "loss": 0.0418,
+      "step": 4430
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.695289398280803e-05,
+      "loss": 0.0489,
+      "step": 4431
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.695220630372493e-05,
+      "loss": 0.0627,
+      "step": 4432
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.695151862464184e-05,
+      "loss": 0.0429,
+      "step": 4433
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.695083094555874e-05,
+      "loss": 0.0338,
+      "step": 4434
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.695014326647565e-05,
+      "loss": 0.0496,
+      "step": 4435
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.694945558739255e-05,
+      "loss": 0.0391,
+      "step": 4436
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.694876790830946e-05,
+      "loss": 0.0414,
+      "step": 4437
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.694808022922636e-05,
+      "loss": 0.0428,
+      "step": 4438
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.694739255014327e-05,
+      "loss": 0.0314,
+      "step": 4439
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.6946704871060174e-05,
+      "loss": 0.0247,
+      "step": 4440
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.694601719197708e-05,
+      "loss": 0.0434,
+      "step": 4441
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6945329512893985e-05,
+      "loss": 0.0192,
+      "step": 4442
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6944641833810894e-05,
+      "loss": 0.0661,
+      "step": 4443
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6943954154727796e-05,
+      "loss": 0.0392,
+      "step": 4444
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6943266475644705e-05,
+      "loss": 0.0445,
+      "step": 4445
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.694257879656161e-05,
+      "loss": 0.0636,
+      "step": 4446
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6941891117478516e-05,
+      "loss": 0.0666,
+      "step": 4447
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.694120343839542e-05,
+      "loss": 0.0518,
+      "step": 4448
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.694051575931233e-05,
+      "loss": 0.0443,
+      "step": 4449
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.693982808022923e-05,
+      "loss": 0.0372,
+      "step": 4450
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.693914040114613e-05,
+      "loss": 0.0706,
+      "step": 4451
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.693845272206304e-05,
+      "loss": 0.0332,
+      "step": 4452
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.693776504297994e-05,
+      "loss": 0.0425,
+      "step": 4453
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6937077363896844e-05,
+      "loss": 0.05,
+      "step": 4454
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.693638968481375e-05,
+      "loss": 0.0402,
+      "step": 4455
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6935702005730655e-05,
+      "loss": 0.0644,
+      "step": 4456
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6935014326647564e-05,
+      "loss": 0.0542,
+      "step": 4457
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.6934326647564466e-05,
+      "loss": 0.035,
+      "step": 4458
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.6933638968481375e-05,
+      "loss": 0.0577,
+      "step": 4459
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.693295128939828e-05,
+      "loss": 0.0344,
+      "step": 4460
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.6932263610315186e-05,
+      "loss": 0.0453,
+      "step": 4461
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.693157593123209e-05,
+      "loss": 0.0523,
+      "step": 4462
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.6930888252149e-05,
+      "loss": 0.0442,
+      "step": 4463
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.69302005730659e-05,
+      "loss": 0.084,
+      "step": 4464
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.692951289398281e-05,
+      "loss": 0.037,
+      "step": 4465
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.692882521489971e-05,
+      "loss": 0.0385,
+      "step": 4466
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.692813753581662e-05,
+      "loss": 0.0604,
+      "step": 4467
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.692744985673352e-05,
+      "loss": 0.0386,
+      "step": 4468
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.692676217765043e-05,
+      "loss": 0.0631,
+      "step": 4469
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.692607449856733e-05,
+      "loss": 0.0573,
+      "step": 4470
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.692538681948424e-05,
+      "loss": 0.0525,
+      "step": 4471
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.6924699140401143e-05,
+      "loss": 0.0342,
+      "step": 4472
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.692401146131805e-05,
+      "loss": 0.0284,
+      "step": 4473
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.6923323782234954e-05,
+      "loss": 0.0684,
+      "step": 4474
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.692263610315186e-05,
+      "loss": 0.0771,
+      "step": 4475
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.6921948424068765e-05,
+      "loss": 0.0578,
+      "step": 4476
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.6921260744985674e-05,
+      "loss": 0.036,
+      "step": 4477
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.6920573065902576e-05,
+      "loss": 0.0384,
+      "step": 4478
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.6919885386819485e-05,
+      "loss": 0.0477,
+      "step": 4479
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.691919770773639e-05,
+      "loss": 0.0508,
+      "step": 4480
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.6918510028653296e-05,
+      "loss": 0.0476,
+      "step": 4481
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.69178223495702e-05,
+      "loss": 0.0592,
+      "step": 4482
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.691713467048711e-05,
+      "loss": 0.0453,
+      "step": 4483
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.691644699140401e-05,
+      "loss": 0.0431,
+      "step": 4484
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.691575931232092e-05,
+      "loss": 0.0294,
+      "step": 4485
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.691507163323782e-05,
+      "loss": 0.0804,
+      "step": 4486
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.691438395415473e-05,
+      "loss": 0.0156,
+      "step": 4487
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.691369627507163e-05,
+      "loss": 0.0599,
+      "step": 4488
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.691300859598854e-05,
+      "loss": 0.0664,
+      "step": 4489
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.691232091690544e-05,
+      "loss": 0.057,
+      "step": 4490
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.691163323782235e-05,
+      "loss": 0.0369,
+      "step": 4491
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.6910945558739253e-05,
+      "loss": 0.0212,
+      "step": 4492
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.691025787965616e-05,
+      "loss": 0.058,
+      "step": 4493
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.6909570200573064e-05,
+      "loss": 0.0343,
+      "step": 4494
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.690888252148997e-05,
+      "loss": 0.0434,
+      "step": 4495
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.6908194842406875e-05,
+      "loss": 0.0495,
+      "step": 4496
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.6907507163323784e-05,
+      "loss": 0.0385,
+      "step": 4497
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.6906819484240686e-05,
+      "loss": 0.0312,
+      "step": 4498
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.6906131805157595e-05,
+      "loss": 0.0413,
+      "step": 4499
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.69054441260745e-05,
+      "loss": 0.0315,
+      "step": 4500
+    },
+    {
+      "epoch": 2.58,
+      "eval_loss": 0.048270635306835175,
+      "eval_mean_accuracy": 0.8602850814184388,
+      "eval_mean_iou": 0.748761761981972,
+      "eval_overall_accuracy": 0.983471049202813,
+      "eval_per_category_accuracy": [
+        0.9927732521242357,
+        0.8354086874359863,
+        0.8966638995409972,
+        0.7162944865725362
+      ],
+      "eval_per_category_iou": [
+        0.9840000971826649,
+        0.6544011182059266,
+        0.7630360272781914,
+        0.5936098052611051
+      ],
+      "eval_runtime": 290.0955,
+      "eval_samples_per_second": 12.41,
+      "eval_steps_per_second": 1.551,
+      "step": 4500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.6904756446991406e-05,
+      "loss": 0.0465,
+      "step": 4501
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.690406876790831e-05,
+      "loss": 0.0637,
+      "step": 4502
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.690338108882522e-05,
+      "loss": 0.0336,
+      "step": 4503
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.690269340974212e-05,
+      "loss": 0.0739,
+      "step": 4504
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.690200573065903e-05,
+      "loss": 0.0562,
+      "step": 4505
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.690131805157593e-05,
+      "loss": 0.0321,
+      "step": 4506
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.690063037249284e-05,
+      "loss": 0.06,
+      "step": 4507
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.689994269340974e-05,
+      "loss": 0.0573,
+      "step": 4508
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.689925501432665e-05,
+      "loss": 0.058,
+      "step": 4509
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.689856733524355e-05,
+      "loss": 0.0553,
+      "step": 4510
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.689787965616046e-05,
+      "loss": 0.0419,
+      "step": 4511
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.6897191977077363e-05,
+      "loss": 0.0435,
+      "step": 4512
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.689650429799427e-05,
+      "loss": 0.0381,
+      "step": 4513
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.6895816618911174e-05,
+      "loss": 0.0146,
+      "step": 4514
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.689512893982808e-05,
+      "loss": 0.0363,
+      "step": 4515
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.6894441260744985e-05,
+      "loss": 0.0601,
+      "step": 4516
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.6893753581661894e-05,
+      "loss": 0.0442,
+      "step": 4517
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.6893065902578796e-05,
+      "loss": 0.0413,
+      "step": 4518
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.68923782234957e-05,
+      "loss": 0.0602,
+      "step": 4519
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.689169054441261e-05,
+      "loss": 0.0409,
+      "step": 4520
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.689100286532951e-05,
+      "loss": 0.0531,
+      "step": 4521
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.689031518624642e-05,
+      "loss": 0.0389,
+      "step": 4522
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.688962750716332e-05,
+      "loss": 0.0397,
+      "step": 4523
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.688893982808023e-05,
+      "loss": 0.0307,
+      "step": 4524
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.688825214899713e-05,
+      "loss": 0.0698,
+      "step": 4525
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.688756446991404e-05,
+      "loss": 0.0399,
+      "step": 4526
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.688687679083094e-05,
+      "loss": 0.0531,
+      "step": 4527
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.688618911174785e-05,
+      "loss": 0.0831,
+      "step": 4528
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.6885501432664754e-05,
+      "loss": 0.0529,
+      "step": 4529
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.688481375358166e-05,
+      "loss": 0.0748,
+      "step": 4530
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.6884126074498565e-05,
+      "loss": 0.0591,
+      "step": 4531
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.6883438395415473e-05,
+      "loss": 0.0374,
+      "step": 4532
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.6882750716332376e-05,
+      "loss": 0.029,
+      "step": 4533
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.6882063037249284e-05,
+      "loss": 0.0312,
+      "step": 4534
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.6881375358166187e-05,
+      "loss": 0.0358,
+      "step": 4535
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.6880687679083095e-05,
+      "loss": 0.048,
+      "step": 4536
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.688e-05,
+      "loss": 0.0508,
+      "step": 4537
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.6879312320916906e-05,
+      "loss": 0.0322,
+      "step": 4538
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.687862464183381e-05,
+      "loss": 0.0253,
+      "step": 4539
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.687793696275072e-05,
+      "loss": 0.035,
+      "step": 4540
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.687724928366762e-05,
+      "loss": 0.0606,
+      "step": 4541
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.687656160458453e-05,
+      "loss": 0.0707,
+      "step": 4542
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.687587392550143e-05,
+      "loss": 0.0827,
+      "step": 4543
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.687518624641834e-05,
+      "loss": 0.0675,
+      "step": 4544
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.687449856733524e-05,
+      "loss": 0.0505,
+      "step": 4545
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.687381088825215e-05,
+      "loss": 0.0693,
+      "step": 4546
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.687312320916905e-05,
+      "loss": 0.0548,
+      "step": 4547
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.687243553008596e-05,
+      "loss": 0.035,
+      "step": 4548
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6871747851002864e-05,
+      "loss": 0.0535,
+      "step": 4549
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.687106017191977e-05,
+      "loss": 0.0544,
+      "step": 4550
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6870372492836675e-05,
+      "loss": 0.0564,
+      "step": 4551
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.686968481375358e-05,
+      "loss": 0.0402,
+      "step": 4552
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6868997134670485e-05,
+      "loss": 0.0415,
+      "step": 4553
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6868309455587394e-05,
+      "loss": 0.0406,
+      "step": 4554
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6867621776504296e-05,
+      "loss": 0.0569,
+      "step": 4555
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6866934097421205e-05,
+      "loss": 0.1295,
+      "step": 4556
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.686624641833811e-05,
+      "loss": 0.0547,
+      "step": 4557
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6865558739255016e-05,
+      "loss": 0.0657,
+      "step": 4558
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.686487106017192e-05,
+      "loss": 0.0422,
+      "step": 4559
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.686418338108883e-05,
+      "loss": 0.0614,
+      "step": 4560
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.686349570200573e-05,
+      "loss": 0.0609,
+      "step": 4561
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.686280802292264e-05,
+      "loss": 0.0307,
+      "step": 4562
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.686212034383954e-05,
+      "loss": 0.0467,
+      "step": 4563
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.686143266475645e-05,
+      "loss": 0.0769,
+      "step": 4564
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.686074498567335e-05,
+      "loss": 0.0524,
+      "step": 4565
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.686005730659026e-05,
+      "loss": 0.0448,
+      "step": 4566
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.685936962750716e-05,
+      "loss": 0.0519,
+      "step": 4567
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.685868194842407e-05,
+      "loss": 0.0378,
+      "step": 4568
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6857994269340973e-05,
+      "loss": 0.0664,
+      "step": 4569
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.685730659025788e-05,
+      "loss": 0.0417,
+      "step": 4570
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6856618911174784e-05,
+      "loss": 0.037,
+      "step": 4571
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.685593123209169e-05,
+      "loss": 0.0546,
+      "step": 4572
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6855243553008595e-05,
+      "loss": 0.0386,
+      "step": 4573
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6854555873925504e-05,
+      "loss": 0.0847,
+      "step": 4574
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6853868194842406e-05,
+      "loss": 0.0623,
+      "step": 4575
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6853180515759315e-05,
+      "loss": 0.0391,
+      "step": 4576
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.685249283667622e-05,
+      "loss": 0.0494,
+      "step": 4577
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6851805157593126e-05,
+      "loss": 0.0562,
+      "step": 4578
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.685111747851003e-05,
+      "loss": 0.0665,
+      "step": 4579
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.685042979942694e-05,
+      "loss": 0.0511,
+      "step": 4580
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.684974212034384e-05,
+      "loss": 0.0658,
+      "step": 4581
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.684905444126075e-05,
+      "loss": 0.0431,
+      "step": 4582
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.684836676217765e-05,
+      "loss": 0.0164,
+      "step": 4583
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.684767908309456e-05,
+      "loss": 0.0301,
+      "step": 4584
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.684699140401146e-05,
+      "loss": 0.1089,
+      "step": 4585
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.684630372492837e-05,
+      "loss": 0.0354,
+      "step": 4586
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.684561604584527e-05,
+      "loss": 0.0419,
+      "step": 4587
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.684492836676218e-05,
+      "loss": 0.081,
+      "step": 4588
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.6844240687679083e-05,
+      "loss": 0.039,
+      "step": 4589
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.684355300859599e-05,
+      "loss": 0.0485,
+      "step": 4590
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.6842865329512894e-05,
+      "loss": 0.0305,
+      "step": 4591
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.68421776504298e-05,
+      "loss": 0.0358,
+      "step": 4592
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.6841489971346705e-05,
+      "loss": 0.0622,
+      "step": 4593
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.6840802292263614e-05,
+      "loss": 0.0457,
+      "step": 4594
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.6840114613180516e-05,
+      "loss": 0.0246,
+      "step": 4595
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.6839426934097425e-05,
+      "loss": 0.0379,
+      "step": 4596
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.683873925501433e-05,
+      "loss": 0.035,
+      "step": 4597
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.6838051575931236e-05,
+      "loss": 0.0495,
+      "step": 4598
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.683736389684814e-05,
+      "loss": 0.0465,
+      "step": 4599
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.683667621776505e-05,
+      "loss": 0.0604,
+      "step": 4600
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.683598853868195e-05,
+      "loss": 0.0634,
+      "step": 4601
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.683530085959886e-05,
+      "loss": 0.0483,
+      "step": 4602
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.683461318051576e-05,
+      "loss": 0.054,
+      "step": 4603
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.683392550143267e-05,
+      "loss": 0.0335,
+      "step": 4604
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.683323782234957e-05,
+      "loss": 0.0616,
+      "step": 4605
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.683255014326648e-05,
+      "loss": 0.0828,
+      "step": 4606
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.683186246418338e-05,
+      "loss": 0.0382,
+      "step": 4607
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.683117478510029e-05,
+      "loss": 0.0564,
+      "step": 4608
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.6830487106017193e-05,
+      "loss": 0.0657,
+      "step": 4609
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.68297994269341e-05,
+      "loss": 0.0376,
+      "step": 4610
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.6829111747851004e-05,
+      "loss": 0.1022,
+      "step": 4611
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.682842406876791e-05,
+      "loss": 0.041,
+      "step": 4612
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.6827736389684815e-05,
+      "loss": 0.0453,
+      "step": 4613
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.6827048710601724e-05,
+      "loss": 0.0505,
+      "step": 4614
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.6826361031518626e-05,
+      "loss": 0.0453,
+      "step": 4615
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.6825673352435535e-05,
+      "loss": 0.0497,
+      "step": 4616
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.682498567335244e-05,
+      "loss": 0.0412,
+      "step": 4617
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.6824297994269346e-05,
+      "loss": 0.0685,
+      "step": 4618
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.682361031518625e-05,
+      "loss": 0.0418,
+      "step": 4619
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.682292263610316e-05,
+      "loss": 0.0747,
+      "step": 4620
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.682223495702006e-05,
+      "loss": 0.0656,
+      "step": 4621
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.682154727793697e-05,
+      "loss": 0.0537,
+      "step": 4622
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.682085959885387e-05,
+      "loss": 0.0303,
+      "step": 4623
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.682017191977078e-05,
+      "loss": 0.0276,
+      "step": 4624
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.681948424068768e-05,
+      "loss": 0.0681,
+      "step": 4625
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.681879656160459e-05,
+      "loss": 0.0301,
+      "step": 4626
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.681810888252149e-05,
+      "loss": 0.0389,
+      "step": 4627
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.68174212034384e-05,
+      "loss": 0.0416,
+      "step": 4628
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.6816733524355303e-05,
+      "loss": 0.0333,
+      "step": 4629
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.681604584527221e-05,
+      "loss": 0.0467,
+      "step": 4630
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.6815358166189114e-05,
+      "loss": 0.0497,
+      "step": 4631
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.681467048710602e-05,
+      "loss": 0.0557,
+      "step": 4632
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6813982808022925e-05,
+      "loss": 0.054,
+      "step": 4633
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6813295128939834e-05,
+      "loss": 0.0534,
+      "step": 4634
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6812607449856736e-05,
+      "loss": 0.0409,
+      "step": 4635
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.681191977077364e-05,
+      "loss": 0.0387,
+      "step": 4636
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.681123209169055e-05,
+      "loss": 0.0679,
+      "step": 4637
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.681054441260745e-05,
+      "loss": 0.0361,
+      "step": 4638
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.680985673352436e-05,
+      "loss": 0.0683,
+      "step": 4639
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.680916905444126e-05,
+      "loss": 0.0616,
+      "step": 4640
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.680848137535817e-05,
+      "loss": 0.046,
+      "step": 4641
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.680779369627507e-05,
+      "loss": 0.0429,
+      "step": 4642
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.680710601719198e-05,
+      "loss": 0.0544,
+      "step": 4643
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.680641833810888e-05,
+      "loss": 0.0664,
+      "step": 4644
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.680573065902579e-05,
+      "loss": 0.0384,
+      "step": 4645
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6805042979942694e-05,
+      "loss": 0.0484,
+      "step": 4646
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.68043553008596e-05,
+      "loss": 0.0416,
+      "step": 4647
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6803667621776505e-05,
+      "loss": 0.0555,
+      "step": 4648
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6802979942693413e-05,
+      "loss": 0.0399,
+      "step": 4649
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6802292263610316e-05,
+      "loss": 0.0986,
+      "step": 4650
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.6801604584527224e-05,
+      "loss": 0.0835,
+      "step": 4651
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.6800916905444127e-05,
+      "loss": 0.0661,
+      "step": 4652
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.6800229226361035e-05,
+      "loss": 0.0663,
+      "step": 4653
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.679954154727794e-05,
+      "loss": 0.0334,
+      "step": 4654
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.6798853868194846e-05,
+      "loss": 0.0582,
+      "step": 4655
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.679816618911175e-05,
+      "loss": 0.0864,
+      "step": 4656
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.679747851002866e-05,
+      "loss": 0.0323,
+      "step": 4657
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.679679083094556e-05,
+      "loss": 0.0597,
+      "step": 4658
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.679610315186247e-05,
+      "loss": 0.0353,
+      "step": 4659
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.679541547277937e-05,
+      "loss": 0.0663,
+      "step": 4660
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.679472779369628e-05,
+      "loss": 0.0426,
+      "step": 4661
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.679404011461318e-05,
+      "loss": 0.0561,
+      "step": 4662
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.679335243553009e-05,
+      "loss": 0.047,
+      "step": 4663
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.679266475644699e-05,
+      "loss": 0.0573,
+      "step": 4664
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.67919770773639e-05,
+      "loss": 0.024,
+      "step": 4665
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.6791289398280804e-05,
+      "loss": 0.0449,
+      "step": 4666
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.679060171919771e-05,
+      "loss": 0.0297,
+      "step": 4667
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.6789914040114615e-05,
+      "loss": 0.0231,
+      "step": 4668
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.6789226361031523e-05,
+      "loss": 0.0349,
+      "step": 4669
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.6788538681948426e-05,
+      "loss": 0.0476,
+      "step": 4670
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.6787851002865334e-05,
+      "loss": 0.0389,
+      "step": 4671
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.6787163323782237e-05,
+      "loss": 0.0522,
+      "step": 4672
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.6786475644699145e-05,
+      "loss": 0.0455,
+      "step": 4673
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.678578796561605e-05,
+      "loss": 0.0507,
+      "step": 4674
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.6785100286532956e-05,
+      "loss": 0.079,
+      "step": 4675
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.678441260744986e-05,
+      "loss": 0.0491,
+      "step": 4676
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.678372492836677e-05,
+      "loss": 0.0525,
+      "step": 4677
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.678303724928367e-05,
+      "loss": 0.0428,
+      "step": 4678
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.678234957020058e-05,
+      "loss": 0.0431,
+      "step": 4679
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.678166189111748e-05,
+      "loss": 0.0454,
+      "step": 4680
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.678097421203439e-05,
+      "loss": 0.0708,
+      "step": 4681
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.678028653295129e-05,
+      "loss": 0.0458,
+      "step": 4682
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.67795988538682e-05,
+      "loss": 0.0514,
+      "step": 4683
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.67789111747851e-05,
+      "loss": 0.0428,
+      "step": 4684
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.677822349570201e-05,
+      "loss": 0.0834,
+      "step": 4685
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6777535816618914e-05,
+      "loss": 0.0491,
+      "step": 4686
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.677684813753582e-05,
+      "loss": 0.0387,
+      "step": 4687
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6776160458452725e-05,
+      "loss": 0.0497,
+      "step": 4688
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6775472779369633e-05,
+      "loss": 0.0409,
+      "step": 4689
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6774785100286536e-05,
+      "loss": 0.0296,
+      "step": 4690
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6774097421203444e-05,
+      "loss": 0.0469,
+      "step": 4691
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6773409742120347e-05,
+      "loss": 0.0506,
+      "step": 4692
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6772722063037255e-05,
+      "loss": 0.0446,
+      "step": 4693
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.677203438395416e-05,
+      "loss": 0.0441,
+      "step": 4694
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6771346704871066e-05,
+      "loss": 0.0375,
+      "step": 4695
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.677065902578797e-05,
+      "loss": 0.0826,
+      "step": 4696
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.676997134670488e-05,
+      "loss": 0.0347,
+      "step": 4697
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.676928366762178e-05,
+      "loss": 0.0571,
+      "step": 4698
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.676859598853869e-05,
+      "loss": 0.0445,
+      "step": 4699
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.676790830945559e-05,
+      "loss": 0.0299,
+      "step": 4700
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.67672206303725e-05,
+      "loss": 0.0978,
+      "step": 4701
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.67665329512894e-05,
+      "loss": 0.0289,
+      "step": 4702
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.676584527220631e-05,
+      "loss": 0.0497,
+      "step": 4703
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.676515759312321e-05,
+      "loss": 0.0361,
+      "step": 4704
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.6764469914040115e-05,
+      "loss": 0.0455,
+      "step": 4705
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.6763782234957023e-05,
+      "loss": 0.056,
+      "step": 4706
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.6763094555873926e-05,
+      "loss": 0.0576,
+      "step": 4707
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.676240687679083e-05,
+      "loss": 0.0296,
+      "step": 4708
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.6761719197707737e-05,
+      "loss": 0.0545,
+      "step": 4709
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.676103151862464e-05,
+      "loss": 0.0132,
+      "step": 4710
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.676034383954155e-05,
+      "loss": 0.0489,
+      "step": 4711
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.675965616045845e-05,
+      "loss": 0.0666,
+      "step": 4712
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.675896848137536e-05,
+      "loss": 0.0438,
+      "step": 4713
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.675828080229226e-05,
+      "loss": 0.0325,
+      "step": 4714
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.675759312320917e-05,
+      "loss": 0.0416,
+      "step": 4715
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.675690544412607e-05,
+      "loss": 0.0494,
+      "step": 4716
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.675621776504298e-05,
+      "loss": 0.0532,
+      "step": 4717
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.675553008595988e-05,
+      "loss": 0.0571,
+      "step": 4718
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.675484240687679e-05,
+      "loss": 0.0315,
+      "step": 4719
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.6754154727793694e-05,
+      "loss": 0.0241,
+      "step": 4720
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.67534670487106e-05,
+      "loss": 0.026,
+      "step": 4721
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.6752779369627505e-05,
+      "loss": 0.0537,
+      "step": 4722
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.6752091690544414e-05,
+      "loss": 0.0352,
+      "step": 4723
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.6751404011461316e-05,
+      "loss": 0.0626,
+      "step": 4724
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.6750716332378225e-05,
+      "loss": 0.0437,
+      "step": 4725
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.675002865329513e-05,
+      "loss": 0.0589,
+      "step": 4726
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.6749340974212036e-05,
+      "loss": 0.0678,
+      "step": 4727
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.674865329512894e-05,
+      "loss": 0.041,
+      "step": 4728
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.6747965616045847e-05,
+      "loss": 0.0909,
+      "step": 4729
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.674727793696275e-05,
+      "loss": 0.02,
+      "step": 4730
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.674659025787966e-05,
+      "loss": 0.0484,
+      "step": 4731
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.674590257879656e-05,
+      "loss": 0.0466,
+      "step": 4732
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.674521489971347e-05,
+      "loss": 0.0474,
+      "step": 4733
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.674452722063037e-05,
+      "loss": 0.0632,
+      "step": 4734
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.674383954154728e-05,
+      "loss": 0.0239,
+      "step": 4735
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.674315186246418e-05,
+      "loss": 0.0633,
+      "step": 4736
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.674246418338109e-05,
+      "loss": 0.0332,
+      "step": 4737
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.674177650429799e-05,
+      "loss": 0.0486,
+      "step": 4738
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.67410888252149e-05,
+      "loss": 0.0855,
+      "step": 4739
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.6740401146131804e-05,
+      "loss": 0.0491,
+      "step": 4740
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.673971346704871e-05,
+      "loss": 0.0851,
+      "step": 4741
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.6739025787965615e-05,
+      "loss": 0.0456,
+      "step": 4742
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.6738338108882524e-05,
+      "loss": 0.0449,
+      "step": 4743
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.6737650429799426e-05,
+      "loss": 0.0534,
+      "step": 4744
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.6736962750716335e-05,
+      "loss": 0.0922,
+      "step": 4745
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.673627507163324e-05,
+      "loss": 0.0696,
+      "step": 4746
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.6735587392550146e-05,
+      "loss": 0.0282,
+      "step": 4747
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.673489971346705e-05,
+      "loss": 0.0474,
+      "step": 4748
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.6734212034383957e-05,
+      "loss": 0.0396,
+      "step": 4749
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.673352435530086e-05,
+      "loss": 0.0523,
+      "step": 4750
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.673283667621777e-05,
+      "loss": 0.0455,
+      "step": 4751
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.673214899713467e-05,
+      "loss": 0.0512,
+      "step": 4752
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.673146131805158e-05,
+      "loss": 0.0336,
+      "step": 4753
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.673077363896848e-05,
+      "loss": 0.0554,
+      "step": 4754
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.673008595988538e-05,
+      "loss": 0.0866,
+      "step": 4755
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.672939828080229e-05,
+      "loss": 0.0599,
+      "step": 4756
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.6728710601719194e-05,
+      "loss": 0.0451,
+      "step": 4757
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.67280229226361e-05,
+      "loss": 0.0598,
+      "step": 4758
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.6727335243553005e-05,
+      "loss": 0.0372,
+      "step": 4759
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.6726647564469914e-05,
+      "loss": 0.0463,
+      "step": 4760
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.6725959885386816e-05,
+      "loss": 0.0357,
+      "step": 4761
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.6725272206303725e-05,
+      "loss": 0.1349,
+      "step": 4762
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.672458452722063e-05,
+      "loss": 0.0255,
+      "step": 4763
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.6723896848137536e-05,
+      "loss": 0.0313,
+      "step": 4764
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.672320916905444e-05,
+      "loss": 0.0371,
+      "step": 4765
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.672252148997135e-05,
+      "loss": 0.0384,
+      "step": 4766
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.672183381088825e-05,
+      "loss": 0.0474,
+      "step": 4767
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.672114613180516e-05,
+      "loss": 0.0436,
+      "step": 4768
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.672045845272206e-05,
+      "loss": 0.047,
+      "step": 4769
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.671977077363897e-05,
+      "loss": 0.0732,
+      "step": 4770
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.671908309455587e-05,
+      "loss": 0.0537,
+      "step": 4771
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.671839541547278e-05,
+      "loss": 0.0455,
+      "step": 4772
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.671770773638968e-05,
+      "loss": 0.0739,
+      "step": 4773
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.671702005730659e-05,
+      "loss": 0.0585,
+      "step": 4774
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.671633237822349e-05,
+      "loss": 0.0533,
+      "step": 4775
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.67156446991404e-05,
+      "loss": 0.0443,
+      "step": 4776
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.6714957020057304e-05,
+      "loss": 0.0344,
+      "step": 4777
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.671426934097421e-05,
+      "loss": 0.0312,
+      "step": 4778
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.6713581661891115e-05,
+      "loss": 0.035,
+      "step": 4779
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.6712893982808024e-05,
+      "loss": 0.0484,
+      "step": 4780
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.6712206303724926e-05,
+      "loss": 0.0549,
+      "step": 4781
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.6711518624641835e-05,
+      "loss": 0.0448,
+      "step": 4782
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.671083094555874e-05,
+      "loss": 0.0461,
+      "step": 4783
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.6710143266475646e-05,
+      "loss": 0.0549,
+      "step": 4784
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.670945558739255e-05,
+      "loss": 0.0548,
+      "step": 4785
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.670876790830946e-05,
+      "loss": 0.0266,
+      "step": 4786
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.670808022922636e-05,
+      "loss": 0.045,
+      "step": 4787
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.670739255014327e-05,
+      "loss": 0.0594,
+      "step": 4788
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.670670487106017e-05,
+      "loss": 0.0284,
+      "step": 4789
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.670601719197708e-05,
+      "loss": 0.0376,
+      "step": 4790
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.670532951289398e-05,
+      "loss": 0.0275,
+      "step": 4791
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.670464183381089e-05,
+      "loss": 0.0396,
+      "step": 4792
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.670395415472779e-05,
+      "loss": 0.0415,
+      "step": 4793
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.67032664756447e-05,
+      "loss": 0.0434,
+      "step": 4794
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.67025787965616e-05,
+      "loss": 0.0742,
+      "step": 4795
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.670189111747851e-05,
+      "loss": 0.0649,
+      "step": 4796
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.6701203438395414e-05,
+      "loss": 0.0595,
+      "step": 4797
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.670051575931232e-05,
+      "loss": 0.0691,
+      "step": 4798
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.6699828080229225e-05,
+      "loss": 0.0206,
+      "step": 4799
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.6699140401146134e-05,
+      "loss": 0.0321,
+      "step": 4800
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.6698452722063036e-05,
+      "loss": 0.0487,
+      "step": 4801
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.6697765042979945e-05,
+      "loss": 0.0947,
+      "step": 4802
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.669707736389685e-05,
+      "loss": 0.0444,
+      "step": 4803
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.6696389684813756e-05,
+      "loss": 0.0454,
+      "step": 4804
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.669570200573066e-05,
+      "loss": 0.0521,
+      "step": 4805
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.669501432664757e-05,
+      "loss": 0.0746,
+      "step": 4806
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.669432664756447e-05,
+      "loss": 0.0537,
+      "step": 4807
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.669363896848138e-05,
+      "loss": 0.0543,
+      "step": 4808
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.669295128939828e-05,
+      "loss": 0.0323,
+      "step": 4809
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.669226361031519e-05,
+      "loss": 0.048,
+      "step": 4810
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.669157593123209e-05,
+      "loss": 0.046,
+      "step": 4811
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.6690888252149e-05,
+      "loss": 0.0321,
+      "step": 4812
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.66902005730659e-05,
+      "loss": 0.0328,
+      "step": 4813
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.668951289398281e-05,
+      "loss": 0.0449,
+      "step": 4814
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.668882521489971e-05,
+      "loss": 0.0429,
+      "step": 4815
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.668813753581662e-05,
+      "loss": 0.0775,
+      "step": 4816
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.6687449856733524e-05,
+      "loss": 0.0727,
+      "step": 4817
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.668676217765043e-05,
+      "loss": 0.0376,
+      "step": 4818
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.6686074498567335e-05,
+      "loss": 0.0474,
+      "step": 4819
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.6685386819484244e-05,
+      "loss": 0.0543,
+      "step": 4820
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.6684699140401146e-05,
+      "loss": 0.0536,
+      "step": 4821
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.6684011461318055e-05,
+      "loss": 0.0597,
+      "step": 4822
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.668332378223496e-05,
+      "loss": 0.0409,
+      "step": 4823
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.6682636103151866e-05,
+      "loss": 0.0699,
+      "step": 4824
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.668194842406877e-05,
+      "loss": 0.0458,
+      "step": 4825
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.668126074498568e-05,
+      "loss": 0.0424,
+      "step": 4826
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.668057306590258e-05,
+      "loss": 0.0469,
+      "step": 4827
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.667988538681949e-05,
+      "loss": 0.0501,
+      "step": 4828
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.667919770773639e-05,
+      "loss": 0.0433,
+      "step": 4829
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.66785100286533e-05,
+      "loss": 0.0415,
+      "step": 4830
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.66778223495702e-05,
+      "loss": 0.0769,
+      "step": 4831
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.667713467048711e-05,
+      "loss": 0.0496,
+      "step": 4832
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.667644699140401e-05,
+      "loss": 0.0258,
+      "step": 4833
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.667575931232092e-05,
+      "loss": 0.0382,
+      "step": 4834
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.667507163323782e-05,
+      "loss": 0.0331,
+      "step": 4835
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.667438395415473e-05,
+      "loss": 0.0347,
+      "step": 4836
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.6673696275071634e-05,
+      "loss": 0.047,
+      "step": 4837
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.667300859598854e-05,
+      "loss": 0.0518,
+      "step": 4838
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.6672320916905445e-05,
+      "loss": 0.023,
+      "step": 4839
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.6671633237822354e-05,
+      "loss": 0.0385,
+      "step": 4840
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.6670945558739256e-05,
+      "loss": 0.0284,
+      "step": 4841
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.6670257879656165e-05,
+      "loss": 0.0445,
+      "step": 4842
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666957020057307e-05,
+      "loss": 0.0494,
+      "step": 4843
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.6668882521489976e-05,
+      "loss": 0.0343,
+      "step": 4844
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666819484240688e-05,
+      "loss": 0.0732,
+      "step": 4845
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666750716332379e-05,
+      "loss": 0.0357,
+      "step": 4846
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666681948424069e-05,
+      "loss": 0.0893,
+      "step": 4847
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.66661318051576e-05,
+      "loss": 0.0668,
+      "step": 4848
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.66654441260745e-05,
+      "loss": 0.0484,
+      "step": 4849
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666475644699141e-05,
+      "loss": 0.0406,
+      "step": 4850
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666406876790831e-05,
+      "loss": 0.045,
+      "step": 4851
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666338108882522e-05,
+      "loss": 0.0378,
+      "step": 4852
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666269340974212e-05,
+      "loss": 0.0283,
+      "step": 4853
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666200573065903e-05,
+      "loss": 0.0624,
+      "step": 4854
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666131805157593e-05,
+      "loss": 0.033,
+      "step": 4855
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.666063037249284e-05,
+      "loss": 0.0472,
+      "step": 4856
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.6659942693409744e-05,
+      "loss": 0.0602,
+      "step": 4857
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.665925501432665e-05,
+      "loss": 0.0487,
+      "step": 4858
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.6658567335243555e-05,
+      "loss": 0.0811,
+      "step": 4859
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.6657879656160464e-05,
+      "loss": 0.0863,
+      "step": 4860
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.6657191977077366e-05,
+      "loss": 0.1615,
+      "step": 4861
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.6656504297994275e-05,
+      "loss": 0.0723,
+      "step": 4862
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.665581661891118e-05,
+      "loss": 0.0514,
+      "step": 4863
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.6655128939828086e-05,
+      "loss": 0.0649,
+      "step": 4864
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.665444126074499e-05,
+      "loss": 0.0406,
+      "step": 4865
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.6653753581661897e-05,
+      "loss": 0.0627,
+      "step": 4866
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.66530659025788e-05,
+      "loss": 0.0421,
+      "step": 4867
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.665237822349571e-05,
+      "loss": 0.0504,
+      "step": 4868
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.665169054441261e-05,
+      "loss": 0.0627,
+      "step": 4869
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.665100286532952e-05,
+      "loss": 0.0503,
+      "step": 4870
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.665031518624642e-05,
+      "loss": 0.0483,
+      "step": 4871
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.664962750716333e-05,
+      "loss": 0.0571,
+      "step": 4872
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.664893982808023e-05,
+      "loss": 0.0407,
+      "step": 4873
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.6648252148997134e-05,
+      "loss": 0.0653,
+      "step": 4874
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.664756446991404e-05,
+      "loss": 0.0726,
+      "step": 4875
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.6646876790830945e-05,
+      "loss": 0.0516,
+      "step": 4876
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.6646189111747854e-05,
+      "loss": 0.0253,
+      "step": 4877
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.6645501432664756e-05,
+      "loss": 0.0627,
+      "step": 4878
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.6644813753581665e-05,
+      "loss": 0.0397,
+      "step": 4879
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.664412607449857e-05,
+      "loss": 0.0495,
+      "step": 4880
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.6643438395415476e-05,
+      "loss": 0.0778,
+      "step": 4881
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.664275071633238e-05,
+      "loss": 0.0562,
+      "step": 4882
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.664206303724929e-05,
+      "loss": 0.0288,
+      "step": 4883
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.664137535816619e-05,
+      "loss": 0.0387,
+      "step": 4884
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.66406876790831e-05,
+      "loss": 0.0251,
+      "step": 4885
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.664e-05,
+      "loss": 0.0284,
+      "step": 4886
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.663931232091691e-05,
+      "loss": 0.0407,
+      "step": 4887
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.663862464183381e-05,
+      "loss": 0.061,
+      "step": 4888
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.663793696275072e-05,
+      "loss": 0.0817,
+      "step": 4889
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.663724928366762e-05,
+      "loss": 0.0581,
+      "step": 4890
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.663656160458453e-05,
+      "loss": 0.05,
+      "step": 4891
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.663587392550143e-05,
+      "loss": 0.0284,
+      "step": 4892
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.663518624641834e-05,
+      "loss": 0.1165,
+      "step": 4893
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.6634498567335244e-05,
+      "loss": 0.072,
+      "step": 4894
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.663381088825215e-05,
+      "loss": 0.0502,
+      "step": 4895
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.6633123209169055e-05,
+      "loss": 0.0371,
+      "step": 4896
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.6632435530085964e-05,
+      "loss": 0.0543,
+      "step": 4897
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.6631747851002866e-05,
+      "loss": 0.0432,
+      "step": 4898
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.6631060171919775e-05,
+      "loss": 0.0412,
+      "step": 4899
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.663037249283668e-05,
+      "loss": 0.0398,
+      "step": 4900
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.6629684813753586e-05,
+      "loss": 0.0499,
+      "step": 4901
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.662899713467049e-05,
+      "loss": 0.0847,
+      "step": 4902
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.66283094555874e-05,
+      "loss": 0.0434,
+      "step": 4903
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.66276217765043e-05,
+      "loss": 0.0597,
+      "step": 4904
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.662693409742121e-05,
+      "loss": 0.0256,
+      "step": 4905
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.662624641833811e-05,
+      "loss": 0.0298,
+      "step": 4906
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.662555873925502e-05,
+      "loss": 0.0471,
+      "step": 4907
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.662487106017192e-05,
+      "loss": 0.061,
+      "step": 4908
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.662418338108883e-05,
+      "loss": 0.0552,
+      "step": 4909
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.662349570200573e-05,
+      "loss": 0.0607,
+      "step": 4910
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.662280802292264e-05,
+      "loss": 0.0318,
+      "step": 4911
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.662212034383954e-05,
+      "loss": 0.0298,
+      "step": 4912
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.662143266475645e-05,
+      "loss": 0.0351,
+      "step": 4913
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.6620744985673354e-05,
+      "loss": 0.0434,
+      "step": 4914
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.662005730659026e-05,
+      "loss": 0.0473,
+      "step": 4915
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.6619369627507165e-05,
+      "loss": 0.04,
+      "step": 4916
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.6618681948424074e-05,
+      "loss": 0.0586,
+      "step": 4917
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.6617994269340976e-05,
+      "loss": 0.0296,
+      "step": 4918
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.6617306590257885e-05,
+      "loss": 0.0623,
+      "step": 4919
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.661661891117479e-05,
+      "loss": 0.0427,
+      "step": 4920
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.6615931232091696e-05,
+      "loss": 0.0377,
+      "step": 4921
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.66152435530086e-05,
+      "loss": 0.0453,
+      "step": 4922
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.661455587392551e-05,
+      "loss": 0.0302,
+      "step": 4923
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.661386819484241e-05,
+      "loss": 0.055,
+      "step": 4924
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.661318051575932e-05,
+      "loss": 0.0623,
+      "step": 4925
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.661249283667622e-05,
+      "loss": 0.0363,
+      "step": 4926
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.661180515759313e-05,
+      "loss": 0.0491,
+      "step": 4927
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.661111747851003e-05,
+      "loss": 0.0272,
+      "step": 4928
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.661042979942694e-05,
+      "loss": 0.0499,
+      "step": 4929
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.660974212034384e-05,
+      "loss": 0.0526,
+      "step": 4930
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.660905444126075e-05,
+      "loss": 0.0602,
+      "step": 4931
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.660836676217765e-05,
+      "loss": 0.0251,
+      "step": 4932
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.660767908309456e-05,
+      "loss": 0.0872,
+      "step": 4933
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.6606991404011464e-05,
+      "loss": 0.0378,
+      "step": 4934
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.660630372492837e-05,
+      "loss": 0.0394,
+      "step": 4935
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.6605616045845275e-05,
+      "loss": 0.0417,
+      "step": 4936
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.6604928366762184e-05,
+      "loss": 0.0423,
+      "step": 4937
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.6604240687679086e-05,
+      "loss": 0.0568,
+      "step": 4938
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.6603553008595995e-05,
+      "loss": 0.0452,
+      "step": 4939
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.66028653295129e-05,
+      "loss": 0.0481,
+      "step": 4940
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.6602177650429806e-05,
+      "loss": 0.1038,
+      "step": 4941
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.660148997134671e-05,
+      "loss": 0.0874,
+      "step": 4942
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.660080229226362e-05,
+      "loss": 0.1683,
+      "step": 4943
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.660011461318052e-05,
+      "loss": 0.0504,
+      "step": 4944
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.659942693409743e-05,
+      "loss": 0.0377,
+      "step": 4945
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.659873925501433e-05,
+      "loss": 0.0281,
+      "step": 4946
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.659805157593124e-05,
+      "loss": 0.0389,
+      "step": 4947
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.659736389684814e-05,
+      "loss": 0.0585,
+      "step": 4948
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.659667621776505e-05,
+      "loss": 0.1275,
+      "step": 4949
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.659598853868195e-05,
+      "loss": 0.0308,
+      "step": 4950
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.659530085959886e-05,
+      "loss": 0.0517,
+      "step": 4951
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.659461318051576e-05,
+      "loss": 0.05,
+      "step": 4952
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.659392550143267e-05,
+      "loss": 0.0611,
+      "step": 4953
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.6593237822349574e-05,
+      "loss": 0.0518,
+      "step": 4954
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.659255014326648e-05,
+      "loss": 0.0608,
+      "step": 4955
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.6591862464183385e-05,
+      "loss": 0.0601,
+      "step": 4956
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.6591174785100294e-05,
+      "loss": 0.0428,
+      "step": 4957
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.6590487106017196e-05,
+      "loss": 0.0434,
+      "step": 4958
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.65897994269341e-05,
+      "loss": 0.0357,
+      "step": 4959
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.658911174785101e-05,
+      "loss": 0.048,
+      "step": 4960
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.658842406876791e-05,
+      "loss": 0.0722,
+      "step": 4961
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.658773638968481e-05,
+      "loss": 0.0563,
+      "step": 4962
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.658704871060172e-05,
+      "loss": 0.0391,
+      "step": 4963
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.658636103151862e-05,
+      "loss": 0.0562,
+      "step": 4964
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.658567335243553e-05,
+      "loss": 0.0442,
+      "step": 4965
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.658498567335243e-05,
+      "loss": 0.0526,
+      "step": 4966
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.658429799426934e-05,
+      "loss": 0.0384,
+      "step": 4967
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.6583610315186244e-05,
+      "loss": 0.0348,
+      "step": 4968
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.658292263610315e-05,
+      "loss": 0.0612,
+      "step": 4969
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.6582234957020055e-05,
+      "loss": 0.0617,
+      "step": 4970
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.6581547277936964e-05,
+      "loss": 0.0556,
+      "step": 4971
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.6580859598853866e-05,
+      "loss": 0.0423,
+      "step": 4972
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.6580171919770775e-05,
+      "loss": 0.0524,
+      "step": 4973
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.657948424068768e-05,
+      "loss": 0.0467,
+      "step": 4974
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.6578796561604586e-05,
+      "loss": 0.0513,
+      "step": 4975
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.657810888252149e-05,
+      "loss": 0.0555,
+      "step": 4976
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.65774212034384e-05,
+      "loss": 0.0608,
+      "step": 4977
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.65767335243553e-05,
+      "loss": 0.0342,
+      "step": 4978
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.657604584527221e-05,
+      "loss": 0.0327,
+      "step": 4979
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.657535816618911e-05,
+      "loss": 0.0599,
+      "step": 4980
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.657467048710602e-05,
+      "loss": 0.0268,
+      "step": 4981
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.657398280802292e-05,
+      "loss": 0.0442,
+      "step": 4982
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.657329512893983e-05,
+      "loss": 0.0563,
+      "step": 4983
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.657260744985673e-05,
+      "loss": 0.0423,
+      "step": 4984
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.657191977077364e-05,
+      "loss": 0.0508,
+      "step": 4985
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.657123209169054e-05,
+      "loss": 0.0724,
+      "step": 4986
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.657054441260745e-05,
+      "loss": 0.1084,
+      "step": 4987
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.6569856733524354e-05,
+      "loss": 0.0287,
+      "step": 4988
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.656916905444126e-05,
+      "loss": 0.0578,
+      "step": 4989
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.6568481375358165e-05,
+      "loss": 0.0382,
+      "step": 4990
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.6567793696275074e-05,
+      "loss": 0.0281,
+      "step": 4991
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.6567106017191976e-05,
+      "loss": 0.0335,
+      "step": 4992
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.656641833810888e-05,
+      "loss": 0.0461,
+      "step": 4993
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.656573065902579e-05,
+      "loss": 0.0458,
+      "step": 4994
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.656504297994269e-05,
+      "loss": 0.0936,
+      "step": 4995
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.65643553008596e-05,
+      "loss": 0.0616,
+      "step": 4996
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.65636676217765e-05,
+      "loss": 0.0452,
+      "step": 4997
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.656297994269341e-05,
+      "loss": 0.0381,
+      "step": 4998
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.656229226361031e-05,
+      "loss": 0.0229,
+      "step": 4999
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.656160458452722e-05,
+      "loss": 0.0589,
+      "step": 5000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.656091690544412e-05,
+      "loss": 0.0382,
+      "step": 5001
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.656022922636103e-05,
+      "loss": 0.0201,
+      "step": 5002
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.655954154727793e-05,
+      "loss": 0.0553,
+      "step": 5003
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.655885386819484e-05,
+      "loss": 0.0487,
+      "step": 5004
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.6558166189111744e-05,
+      "loss": 0.0658,
+      "step": 5005
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.655747851002865e-05,
+      "loss": 0.0414,
+      "step": 5006
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.6556790830945555e-05,
+      "loss": 0.0296,
+      "step": 5007
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.6556103151862464e-05,
+      "loss": 0.037,
+      "step": 5008
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.6555415472779366e-05,
+      "loss": 0.0683,
+      "step": 5009
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.6554727793696275e-05,
+      "loss": 0.0502,
+      "step": 5010
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.655404011461318e-05,
+      "loss": 0.0454,
+      "step": 5011
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.6553352435530086e-05,
+      "loss": 0.0686,
+      "step": 5012
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.655266475644699e-05,
+      "loss": 0.0224,
+      "step": 5013
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.65519770773639e-05,
+      "loss": 0.0511,
+      "step": 5014
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.65512893982808e-05,
+      "loss": 0.05,
+      "step": 5015
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.655060171919771e-05,
+      "loss": 0.0553,
+      "step": 5016
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.654991404011461e-05,
+      "loss": 0.0494,
+      "step": 5017
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.654922636103152e-05,
+      "loss": 0.0286,
+      "step": 5018
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.654853868194842e-05,
+      "loss": 0.082,
+      "step": 5019
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.654785100286533e-05,
+      "loss": 0.039,
+      "step": 5020
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.654716332378223e-05,
+      "loss": 0.0407,
+      "step": 5021
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.654647564469914e-05,
+      "loss": 0.0427,
+      "step": 5022
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.654578796561604e-05,
+      "loss": 0.064,
+      "step": 5023
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.654510028653295e-05,
+      "loss": 0.0354,
+      "step": 5024
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.6544412607449854e-05,
+      "loss": 0.04,
+      "step": 5025
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.654372492836676e-05,
+      "loss": 0.0361,
+      "step": 5026
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.6543037249283665e-05,
+      "loss": 0.022,
+      "step": 5027
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.6542349570200574e-05,
+      "loss": 0.0537,
+      "step": 5028
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.6541661891117476e-05,
+      "loss": 0.0407,
+      "step": 5029
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.6540974212034385e-05,
+      "loss": 0.046,
+      "step": 5030
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.654028653295129e-05,
+      "loss": 0.0582,
+      "step": 5031
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.6539598853868196e-05,
+      "loss": 0.0524,
+      "step": 5032
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.65389111747851e-05,
+      "loss": 0.0756,
+      "step": 5033
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.653822349570201e-05,
+      "loss": 0.0288,
+      "step": 5034
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.653753581661891e-05,
+      "loss": 0.058,
+      "step": 5035
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.653684813753582e-05,
+      "loss": 0.0779,
+      "step": 5036
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.653616045845272e-05,
+      "loss": 0.0382,
+      "step": 5037
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.653547277936963e-05,
+      "loss": 0.0471,
+      "step": 5038
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.653478510028653e-05,
+      "loss": 0.0497,
+      "step": 5039
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.653409742120344e-05,
+      "loss": 0.0451,
+      "step": 5040
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.653340974212034e-05,
+      "loss": 0.0291,
+      "step": 5041
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.653272206303725e-05,
+      "loss": 0.0462,
+      "step": 5042
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.653203438395415e-05,
+      "loss": 0.0657,
+      "step": 5043
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.653134670487106e-05,
+      "loss": 0.0512,
+      "step": 5044
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.6530659025787964e-05,
+      "loss": 0.0852,
+      "step": 5045
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.652997134670487e-05,
+      "loss": 0.0807,
+      "step": 5046
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.6529283667621775e-05,
+      "loss": 0.0792,
+      "step": 5047
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.6528595988538684e-05,
+      "loss": 0.0456,
+      "step": 5048
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.6527908309455586e-05,
+      "loss": 0.0372,
+      "step": 5049
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.6527220630372495e-05,
+      "loss": 0.0516,
+      "step": 5050
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.65265329512894e-05,
+      "loss": 0.0372,
+      "step": 5051
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6525845272206306e-05,
+      "loss": 0.0405,
+      "step": 5052
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.652515759312321e-05,
+      "loss": 0.0559,
+      "step": 5053
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.652446991404012e-05,
+      "loss": 0.0468,
+      "step": 5054
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.652378223495702e-05,
+      "loss": 0.0438,
+      "step": 5055
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.652309455587393e-05,
+      "loss": 0.0465,
+      "step": 5056
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.652240687679083e-05,
+      "loss": 0.0454,
+      "step": 5057
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.652171919770774e-05,
+      "loss": 0.0407,
+      "step": 5058
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.652103151862464e-05,
+      "loss": 0.0643,
+      "step": 5059
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.652034383954155e-05,
+      "loss": 0.0252,
+      "step": 5060
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.651965616045845e-05,
+      "loss": 0.0308,
+      "step": 5061
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.651896848137536e-05,
+      "loss": 0.0465,
+      "step": 5062
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.651828080229226e-05,
+      "loss": 0.047,
+      "step": 5063
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.651759312320917e-05,
+      "loss": 0.0515,
+      "step": 5064
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6516905444126074e-05,
+      "loss": 0.032,
+      "step": 5065
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.651621776504298e-05,
+      "loss": 0.0405,
+      "step": 5066
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6515530085959885e-05,
+      "loss": 0.0521,
+      "step": 5067
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6514842406876794e-05,
+      "loss": 0.0193,
+      "step": 5068
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6514154727793696e-05,
+      "loss": 0.0334,
+      "step": 5069
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.6513467048710605e-05,
+      "loss": 0.0372,
+      "step": 5070
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.651277936962751e-05,
+      "loss": 0.0318,
+      "step": 5071
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.6512091690544416e-05,
+      "loss": 0.0311,
+      "step": 5072
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.651140401146132e-05,
+      "loss": 0.0423,
+      "step": 5073
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.651071633237823e-05,
+      "loss": 0.0343,
+      "step": 5074
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.651002865329513e-05,
+      "loss": 0.0689,
+      "step": 5075
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.650934097421204e-05,
+      "loss": 0.0951,
+      "step": 5076
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.650865329512894e-05,
+      "loss": 0.0476,
+      "step": 5077
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.650796561604585e-05,
+      "loss": 0.0555,
+      "step": 5078
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.650727793696275e-05,
+      "loss": 0.0268,
+      "step": 5079
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.650659025787966e-05,
+      "loss": 0.0495,
+      "step": 5080
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.650590257879656e-05,
+      "loss": 0.0733,
+      "step": 5081
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.650521489971347e-05,
+      "loss": 0.0415,
+      "step": 5082
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.650452722063037e-05,
+      "loss": 0.0473,
+      "step": 5083
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.650383954154728e-05,
+      "loss": 0.0267,
+      "step": 5084
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.6503151862464184e-05,
+      "loss": 0.0775,
+      "step": 5085
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.650246418338109e-05,
+      "loss": 0.092,
+      "step": 5086
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.6501776504297995e-05,
+      "loss": 0.0684,
+      "step": 5087
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.6501088825214904e-05,
+      "loss": 0.1143,
+      "step": 5088
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.6500401146131806e-05,
+      "loss": 0.0944,
+      "step": 5089
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.6499713467048715e-05,
+      "loss": 0.0335,
+      "step": 5090
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649902578796562e-05,
+      "loss": 0.0399,
+      "step": 5091
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.6498338108882526e-05,
+      "loss": 0.0641,
+      "step": 5092
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649765042979943e-05,
+      "loss": 0.0711,
+      "step": 5093
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649696275071634e-05,
+      "loss": 0.0563,
+      "step": 5094
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649627507163324e-05,
+      "loss": 0.0517,
+      "step": 5095
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649558739255015e-05,
+      "loss": 0.0471,
+      "step": 5096
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649489971346705e-05,
+      "loss": 0.0378,
+      "step": 5097
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649421203438396e-05,
+      "loss": 0.0437,
+      "step": 5098
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649352435530086e-05,
+      "loss": 0.0597,
+      "step": 5099
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649283667621777e-05,
+      "loss": 0.0673,
+      "step": 5100
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649214899713467e-05,
+      "loss": 0.0397,
+      "step": 5101
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649146131805158e-05,
+      "loss": 0.0403,
+      "step": 5102
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649077363896848e-05,
+      "loss": 0.0314,
+      "step": 5103
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 5.649008595988539e-05,
+      "loss": 0.0588,
+      "step": 5104
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.6489398280802294e-05,
+      "loss": 0.0372,
+      "step": 5105
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.64887106017192e-05,
+      "loss": 0.0554,
+      "step": 5106
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.6488022922636105e-05,
+      "loss": 0.0544,
+      "step": 5107
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.6487335243553014e-05,
+      "loss": 0.0576,
+      "step": 5108
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.6486647564469916e-05,
+      "loss": 0.0906,
+      "step": 5109
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.6485959885386825e-05,
+      "loss": 0.0519,
+      "step": 5110
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.648527220630373e-05,
+      "loss": 0.087,
+      "step": 5111
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.648458452722063e-05,
+      "loss": 0.039,
+      "step": 5112
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.648389684813754e-05,
+      "loss": 0.1169,
+      "step": 5113
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.648320916905444e-05,
+      "loss": 0.0412,
+      "step": 5114
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.648252148997135e-05,
+      "loss": 0.0636,
+      "step": 5115
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.648183381088825e-05,
+      "loss": 0.0411,
+      "step": 5116
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.648114613180516e-05,
+      "loss": 0.0463,
+      "step": 5117
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.648045845272206e-05,
+      "loss": 0.046,
+      "step": 5118
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.647977077363897e-05,
+      "loss": 0.0694,
+      "step": 5119
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.647908309455587e-05,
+      "loss": 0.0537,
+      "step": 5120
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 5.647839541547278e-05,
+      "loss": 0.0485,
+      "step": 5121
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.6477707736389684e-05,
+      "loss": 0.0642,
+      "step": 5122
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.647702005730659e-05,
+      "loss": 0.0426,
+      "step": 5123
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.6476332378223495e-05,
+      "loss": 0.036,
+      "step": 5124
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.6475644699140404e-05,
+      "loss": 0.0237,
+      "step": 5125
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.6474957020057306e-05,
+      "loss": 0.0385,
+      "step": 5126
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.6474269340974215e-05,
+      "loss": 0.0589,
+      "step": 5127
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.647358166189112e-05,
+      "loss": 0.0611,
+      "step": 5128
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.6472893982808026e-05,
+      "loss": 0.0418,
+      "step": 5129
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.647220630372493e-05,
+      "loss": 0.0935,
+      "step": 5130
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.647151862464184e-05,
+      "loss": 0.0291,
+      "step": 5131
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.647083094555874e-05,
+      "loss": 0.0412,
+      "step": 5132
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.647014326647565e-05,
+      "loss": 0.0508,
+      "step": 5133
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.646945558739255e-05,
+      "loss": 0.0562,
+      "step": 5134
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.646876790830946e-05,
+      "loss": 0.0407,
+      "step": 5135
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.646808022922636e-05,
+      "loss": 0.0538,
+      "step": 5136
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.646739255014327e-05,
+      "loss": 0.0501,
+      "step": 5137
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.646670487106017e-05,
+      "loss": 0.0373,
+      "step": 5138
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 5.646601719197708e-05,
+      "loss": 0.0917,
+      "step": 5139
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.646532951289398e-05,
+      "loss": 0.0552,
+      "step": 5140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.646464183381089e-05,
+      "loss": 0.0335,
+      "step": 5141
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.6463954154727794e-05,
+      "loss": 0.0526,
+      "step": 5142
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.64632664756447e-05,
+      "loss": 0.0583,
+      "step": 5143
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.6462578796561605e-05,
+      "loss": 0.0363,
+      "step": 5144
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.6461891117478514e-05,
+      "loss": 0.0804,
+      "step": 5145
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.6461203438395416e-05,
+      "loss": 0.0452,
+      "step": 5146
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.6460515759312325e-05,
+      "loss": 0.0418,
+      "step": 5147
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.645982808022923e-05,
+      "loss": 0.0313,
+      "step": 5148
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.6459140401146136e-05,
+      "loss": 0.0544,
+      "step": 5149
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.645845272206304e-05,
+      "loss": 0.0571,
+      "step": 5150
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.645776504297995e-05,
+      "loss": 0.048,
+      "step": 5151
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.645707736389685e-05,
+      "loss": 0.0338,
+      "step": 5152
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.645638968481376e-05,
+      "loss": 0.0569,
+      "step": 5153
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.645570200573066e-05,
+      "loss": 0.0412,
+      "step": 5154
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.645501432664757e-05,
+      "loss": 0.0326,
+      "step": 5155
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 5.645432664756447e-05,
+      "loss": 0.0381,
+      "step": 5156
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.645363896848138e-05,
+      "loss": 0.0505,
+      "step": 5157
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.645295128939828e-05,
+      "loss": 0.0475,
+      "step": 5158
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.645226361031519e-05,
+      "loss": 0.0311,
+      "step": 5159
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.645157593123209e-05,
+      "loss": 0.0563,
+      "step": 5160
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.6450888252149e-05,
+      "loss": 0.0213,
+      "step": 5161
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.6450200573065904e-05,
+      "loss": 0.0344,
+      "step": 5162
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.644951289398281e-05,
+      "loss": 0.0868,
+      "step": 5163
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.6448825214899715e-05,
+      "loss": 0.0446,
+      "step": 5164
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.6448137535816624e-05,
+      "loss": 0.0737,
+      "step": 5165
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.6447449856733526e-05,
+      "loss": 0.0331,
+      "step": 5166
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.6446762177650435e-05,
+      "loss": 0.0657,
+      "step": 5167
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.644607449856734e-05,
+      "loss": 0.0312,
+      "step": 5168
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.6445386819484246e-05,
+      "loss": 0.0352,
+      "step": 5169
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.644469914040115e-05,
+      "loss": 0.0375,
+      "step": 5170
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.644401146131806e-05,
+      "loss": 0.0524,
+      "step": 5171
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.644332378223496e-05,
+      "loss": 0.0375,
+      "step": 5172
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 5.644263610315187e-05,
+      "loss": 0.0647,
+      "step": 5173
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.644194842406877e-05,
+      "loss": 0.0591,
+      "step": 5174
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.644126074498568e-05,
+      "loss": 0.0352,
+      "step": 5175
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.644057306590258e-05,
+      "loss": 0.053,
+      "step": 5176
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.643988538681949e-05,
+      "loss": 0.0565,
+      "step": 5177
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.643919770773639e-05,
+      "loss": 0.042,
+      "step": 5178
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.64385100286533e-05,
+      "loss": 0.0372,
+      "step": 5179
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.64378223495702e-05,
+      "loss": 0.0335,
+      "step": 5180
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.643713467048711e-05,
+      "loss": 0.025,
+      "step": 5181
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6436446991404014e-05,
+      "loss": 0.0745,
+      "step": 5182
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.643575931232092e-05,
+      "loss": 0.0523,
+      "step": 5183
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6435071633237825e-05,
+      "loss": 0.067,
+      "step": 5184
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6434383954154734e-05,
+      "loss": 0.0406,
+      "step": 5185
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6433696275071636e-05,
+      "loss": 0.0415,
+      "step": 5186
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6433008595988545e-05,
+      "loss": 0.047,
+      "step": 5187
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.643232091690545e-05,
+      "loss": 0.0183,
+      "step": 5188
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.6431633237822356e-05,
+      "loss": 0.0436,
+      "step": 5189
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.643094555873926e-05,
+      "loss": 0.0594,
+      "step": 5190
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.643025787965617e-05,
+      "loss": 0.067,
+      "step": 5191
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.642957020057307e-05,
+      "loss": 0.0527,
+      "step": 5192
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.642888252148998e-05,
+      "loss": 0.0405,
+      "step": 5193
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.642819484240688e-05,
+      "loss": 0.0286,
+      "step": 5194
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.642750716332379e-05,
+      "loss": 0.0318,
+      "step": 5195
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.642681948424069e-05,
+      "loss": 0.0431,
+      "step": 5196
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.64261318051576e-05,
+      "loss": 0.0445,
+      "step": 5197
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.64254441260745e-05,
+      "loss": 0.0668,
+      "step": 5198
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.642475644699141e-05,
+      "loss": 0.0456,
+      "step": 5199
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.642406876790831e-05,
+      "loss": 0.0359,
+      "step": 5200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.642338108882522e-05,
+      "loss": 0.0473,
+      "step": 5201
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.6422693409742124e-05,
+      "loss": 0.0581,
+      "step": 5202
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.642200573065903e-05,
+      "loss": 0.0789,
+      "step": 5203
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.6421318051575935e-05,
+      "loss": 0.0445,
+      "step": 5204
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.6420630372492844e-05,
+      "loss": 0.0581,
+      "step": 5205
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.6419942693409746e-05,
+      "loss": 0.0393,
+      "step": 5206
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.6419255014326655e-05,
+      "loss": 0.0361,
+      "step": 5207
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.641856733524356e-05,
+      "loss": 0.0372,
+      "step": 5208
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.6417879656160466e-05,
+      "loss": 0.0571,
+      "step": 5209
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.641719197707737e-05,
+      "loss": 0.1077,
+      "step": 5210
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.641650429799428e-05,
+      "loss": 0.0541,
+      "step": 5211
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.641581661891118e-05,
+      "loss": 0.0552,
+      "step": 5212
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.641512893982808e-05,
+      "loss": 0.0334,
+      "step": 5213
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.641444126074499e-05,
+      "loss": 0.0368,
+      "step": 5214
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.641375358166189e-05,
+      "loss": 0.0717,
+      "step": 5215
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.6413065902578794e-05,
+      "loss": 0.05,
+      "step": 5216
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.64123782234957e-05,
+      "loss": 0.0465,
+      "step": 5217
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.6411690544412605e-05,
+      "loss": 0.0596,
+      "step": 5218
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.6411002865329514e-05,
+      "loss": 0.0613,
+      "step": 5219
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.6410315186246416e-05,
+      "loss": 0.0328,
+      "step": 5220
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.6409627507163325e-05,
+      "loss": 0.0413,
+      "step": 5221
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.640893982808023e-05,
+      "loss": 0.0466,
+      "step": 5222
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.6408252148997136e-05,
+      "loss": 0.0289,
+      "step": 5223
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.640756446991404e-05,
+      "loss": 0.0385,
+      "step": 5224
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.640687679083095e-05,
+      "loss": 0.0392,
+      "step": 5225
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.640618911174785e-05,
+      "loss": 0.0379,
+      "step": 5226
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.640550143266476e-05,
+      "loss": 0.0545,
+      "step": 5227
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.640481375358166e-05,
+      "loss": 0.0764,
+      "step": 5228
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.640412607449857e-05,
+      "loss": 0.0623,
+      "step": 5229
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.640343839541547e-05,
+      "loss": 0.0484,
+      "step": 5230
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.640275071633237e-05,
+      "loss": 0.0182,
+      "step": 5231
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.640206303724928e-05,
+      "loss": 0.2354,
+      "step": 5232
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6401375358166184e-05,
+      "loss": 0.0545,
+      "step": 5233
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.640068767908309e-05,
+      "loss": 0.0473,
+      "step": 5234
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 0.0765,
+      "step": 5235
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6399312320916904e-05,
+      "loss": 0.0447,
+      "step": 5236
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6398624641833806e-05,
+      "loss": 0.0235,
+      "step": 5237
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6397936962750715e-05,
+      "loss": 0.0194,
+      "step": 5238
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.639724928366762e-05,
+      "loss": 0.0422,
+      "step": 5239
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6396561604584526e-05,
+      "loss": 0.0215,
+      "step": 5240
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.639587392550143e-05,
+      "loss": 0.0531,
+      "step": 5241
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.639518624641834e-05,
+      "loss": 0.0571,
+      "step": 5242
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.639449856733524e-05,
+      "loss": 0.0739,
+      "step": 5243
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.639381088825215e-05,
+      "loss": 0.0476,
+      "step": 5244
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.639312320916905e-05,
+      "loss": 0.0145,
+      "step": 5245
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.639243553008596e-05,
+      "loss": 0.0297,
+      "step": 5246
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.639174785100286e-05,
+      "loss": 0.0516,
+      "step": 5247
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.639106017191977e-05,
+      "loss": 0.0408,
+      "step": 5248
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.639037249283667e-05,
+      "loss": 0.0458,
+      "step": 5249
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.638968481375358e-05,
+      "loss": 0.0315,
+      "step": 5250
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.638899713467048e-05,
+      "loss": 0.0483,
+      "step": 5251
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.638830945558739e-05,
+      "loss": 0.0233,
+      "step": 5252
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.6387621776504294e-05,
+      "loss": 0.0421,
+      "step": 5253
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.63869340974212e-05,
+      "loss": 0.0426,
+      "step": 5254
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.6386246418338105e-05,
+      "loss": 0.0285,
+      "step": 5255
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.6385558739255014e-05,
+      "loss": 0.0516,
+      "step": 5256
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.6384871060171916e-05,
+      "loss": 0.064,
+      "step": 5257
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.6384183381088825e-05,
+      "loss": 0.0345,
+      "step": 5258
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.638349570200573e-05,
+      "loss": 0.0195,
+      "step": 5259
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.6382808022922636e-05,
+      "loss": 0.0346,
+      "step": 5260
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 5.638212034383954e-05,
+      "loss": 0.0287,
+      "step": 5261
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.638143266475645e-05,
+      "loss": 0.0334,
+      "step": 5262
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.638074498567335e-05,
+      "loss": 0.0217,
+      "step": 5263
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.638005730659026e-05,
+      "loss": 0.0277,
+      "step": 5264
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.637936962750716e-05,
+      "loss": 0.0258,
+      "step": 5265
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.637868194842407e-05,
+      "loss": 0.0688,
+      "step": 5266
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.637799426934097e-05,
+      "loss": 0.0456,
+      "step": 5267
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.637730659025788e-05,
+      "loss": 0.0483,
+      "step": 5268
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.637661891117478e-05,
+      "loss": 0.0302,
+      "step": 5269
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.637593123209169e-05,
+      "loss": 0.0626,
+      "step": 5270
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.637524355300859e-05,
+      "loss": 0.0387,
+      "step": 5271
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.63745558739255e-05,
+      "loss": 0.0322,
+      "step": 5272
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.6373868194842404e-05,
+      "loss": 0.0311,
+      "step": 5273
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.637318051575931e-05,
+      "loss": 0.0497,
+      "step": 5274
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.6372492836676215e-05,
+      "loss": 0.0456,
+      "step": 5275
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.6371805157593124e-05,
+      "loss": 0.0388,
+      "step": 5276
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.6371117478510026e-05,
+      "loss": 0.0551,
+      "step": 5277
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 5.6370429799426935e-05,
+      "loss": 0.0359,
+      "step": 5278
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636974212034384e-05,
+      "loss": 0.0406,
+      "step": 5279
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.6369054441260746e-05,
+      "loss": 0.0366,
+      "step": 5280
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636836676217765e-05,
+      "loss": 0.0356,
+      "step": 5281
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636767908309456e-05,
+      "loss": 0.0281,
+      "step": 5282
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636699140401146e-05,
+      "loss": 0.0188,
+      "step": 5283
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636630372492837e-05,
+      "loss": 0.0725,
+      "step": 5284
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636561604584527e-05,
+      "loss": 0.0272,
+      "step": 5285
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636492836676218e-05,
+      "loss": 0.0243,
+      "step": 5286
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636424068767908e-05,
+      "loss": 0.0454,
+      "step": 5287
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636355300859599e-05,
+      "loss": 0.0426,
+      "step": 5288
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636286532951289e-05,
+      "loss": 0.0414,
+      "step": 5289
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.63621776504298e-05,
+      "loss": 0.0444,
+      "step": 5290
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.63614899713467e-05,
+      "loss": 0.0336,
+      "step": 5291
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.636080229226361e-05,
+      "loss": 0.0336,
+      "step": 5292
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.6360114613180514e-05,
+      "loss": 0.036,
+      "step": 5293
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.635942693409742e-05,
+      "loss": 0.0211,
+      "step": 5294
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.6358739255014325e-05,
+      "loss": 0.0347,
+      "step": 5295
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 5.6358051575931234e-05,
+      "loss": 0.0401,
+      "step": 5296
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.6357363896848136e-05,
+      "loss": 0.0547,
+      "step": 5297
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.6356676217765045e-05,
+      "loss": 0.0276,
+      "step": 5298
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.635598853868195e-05,
+      "loss": 0.0481,
+      "step": 5299
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.6355300859598856e-05,
+      "loss": 0.0361,
+      "step": 5300
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.635461318051576e-05,
+      "loss": 0.0167,
+      "step": 5301
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.635392550143267e-05,
+      "loss": 0.0525,
+      "step": 5302
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.635323782234957e-05,
+      "loss": 0.0498,
+      "step": 5303
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.635255014326648e-05,
+      "loss": 0.0253,
+      "step": 5304
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.635186246418338e-05,
+      "loss": 0.0211,
+      "step": 5305
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.635117478510029e-05,
+      "loss": 0.0678,
+      "step": 5306
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.635048710601719e-05,
+      "loss": 0.0451,
+      "step": 5307
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.63497994269341e-05,
+      "loss": 0.0245,
+      "step": 5308
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.6349111747851e-05,
+      "loss": 0.0286,
+      "step": 5309
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.634842406876791e-05,
+      "loss": 0.0419,
+      "step": 5310
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.634773638968481e-05,
+      "loss": 0.0336,
+      "step": 5311
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.634704871060172e-05,
+      "loss": 0.0544,
+      "step": 5312
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 5.6346361031518624e-05,
+      "loss": 0.0319,
+      "step": 5313
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.634567335243553e-05,
+      "loss": 0.0373,
+      "step": 5314
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.6344985673352435e-05,
+      "loss": 0.0165,
+      "step": 5315
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.6344297994269344e-05,
+      "loss": 0.0265,
+      "step": 5316
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.6343610315186246e-05,
+      "loss": 0.032,
+      "step": 5317
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.6342922636103155e-05,
+      "loss": 0.0559,
+      "step": 5318
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.634223495702006e-05,
+      "loss": 0.0326,
+      "step": 5319
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.6341547277936966e-05,
+      "loss": 0.0274,
+      "step": 5320
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.634085959885387e-05,
+      "loss": 0.0385,
+      "step": 5321
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.634017191977078e-05,
+      "loss": 0.0418,
+      "step": 5322
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.633948424068768e-05,
+      "loss": 0.0471,
+      "step": 5323
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.633879656160459e-05,
+      "loss": 0.039,
+      "step": 5324
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.633810888252149e-05,
+      "loss": 0.0548,
+      "step": 5325
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.63374212034384e-05,
+      "loss": 0.0347,
+      "step": 5326
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.63367335243553e-05,
+      "loss": 0.0281,
+      "step": 5327
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.633604584527221e-05,
+      "loss": 0.0524,
+      "step": 5328
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.633535816618911e-05,
+      "loss": 0.0411,
+      "step": 5329
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 5.633467048710602e-05,
+      "loss": 0.0406,
+      "step": 5330
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.633398280802292e-05,
+      "loss": 0.0425,
+      "step": 5331
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.633329512893983e-05,
+      "loss": 0.0461,
+      "step": 5332
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.6332607449856734e-05,
+      "loss": 0.0208,
+      "step": 5333
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.633191977077364e-05,
+      "loss": 0.0442,
+      "step": 5334
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.6331232091690545e-05,
+      "loss": 0.0433,
+      "step": 5335
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.6330544412607454e-05,
+      "loss": 0.0477,
+      "step": 5336
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.6329856733524356e-05,
+      "loss": 0.0415,
+      "step": 5337
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.6329169054441265e-05,
+      "loss": 0.0474,
+      "step": 5338
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.632848137535817e-05,
+      "loss": 0.0843,
+      "step": 5339
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.6327793696275076e-05,
+      "loss": 0.0395,
+      "step": 5340
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.632710601719198e-05,
+      "loss": 0.0405,
+      "step": 5341
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.632641833810889e-05,
+      "loss": 0.0281,
+      "step": 5342
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.632573065902579e-05,
+      "loss": 0.0398,
+      "step": 5343
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.63250429799427e-05,
+      "loss": 0.0435,
+      "step": 5344
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.63243553008596e-05,
+      "loss": 0.0331,
+      "step": 5345
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.632366762177651e-05,
+      "loss": 0.0305,
+      "step": 5346
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.632297994269341e-05,
+      "loss": 0.0392,
+      "step": 5347
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 5.632229226361032e-05,
+      "loss": 0.0436,
+      "step": 5348
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.632160458452722e-05,
+      "loss": 0.0269,
+      "step": 5349
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.6320916905444124e-05,
+      "loss": 0.0471,
+      "step": 5350
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.632022922636103e-05,
+      "loss": 0.0435,
+      "step": 5351
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.6319541547277935e-05,
+      "loss": 0.0802,
+      "step": 5352
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.6318853868194844e-05,
+      "loss": 0.0303,
+      "step": 5353
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.6318166189111746e-05,
+      "loss": 0.0423,
+      "step": 5354
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.6317478510028655e-05,
+      "loss": 0.0482,
+      "step": 5355
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.631679083094556e-05,
+      "loss": 0.0431,
+      "step": 5356
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.6316103151862466e-05,
+      "loss": 0.0561,
+      "step": 5357
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.631541547277937e-05,
+      "loss": 0.0485,
+      "step": 5358
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.631472779369628e-05,
+      "loss": 0.0223,
+      "step": 5359
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.631404011461318e-05,
+      "loss": 0.0341,
+      "step": 5360
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.631335243553009e-05,
+      "loss": 0.0188,
+      "step": 5361
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.631266475644699e-05,
+      "loss": 0.0234,
+      "step": 5362
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.63119770773639e-05,
+      "loss": 0.0363,
+      "step": 5363
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.63112893982808e-05,
+      "loss": 0.0626,
+      "step": 5364
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 5.631060171919771e-05,
+      "loss": 0.0197,
+      "step": 5365
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.630991404011461e-05,
+      "loss": 0.0295,
+      "step": 5366
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.630922636103152e-05,
+      "loss": 0.0337,
+      "step": 5367
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.630853868194842e-05,
+      "loss": 0.0367,
+      "step": 5368
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.630785100286533e-05,
+      "loss": 0.0424,
+      "step": 5369
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.6307163323782234e-05,
+      "loss": 0.0615,
+      "step": 5370
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.630647564469914e-05,
+      "loss": 0.0215,
+      "step": 5371
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.6305787965616045e-05,
+      "loss": 0.0342,
+      "step": 5372
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.6305100286532954e-05,
+      "loss": 0.0443,
+      "step": 5373
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.6304412607449856e-05,
+      "loss": 0.0582,
+      "step": 5374
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.6303724928366765e-05,
+      "loss": 0.045,
+      "step": 5375
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.630303724928367e-05,
+      "loss": 0.0386,
+      "step": 5376
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.6302349570200576e-05,
+      "loss": 0.0274,
+      "step": 5377
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.630166189111748e-05,
+      "loss": 0.0237,
+      "step": 5378
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.630097421203439e-05,
+      "loss": 0.0263,
+      "step": 5379
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.630028653295129e-05,
+      "loss": 0.0267,
+      "step": 5380
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.62995988538682e-05,
+      "loss": 0.039,
+      "step": 5381
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.62989111747851e-05,
+      "loss": 0.0286,
+      "step": 5382
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 5.629822349570201e-05,
+      "loss": 0.0385,
+      "step": 5383
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.629753581661891e-05,
+      "loss": 0.0223,
+      "step": 5384
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.629684813753582e-05,
+      "loss": 0.0344,
+      "step": 5385
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.629616045845272e-05,
+      "loss": 0.0455,
+      "step": 5386
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.629547277936963e-05,
+      "loss": 0.0353,
+      "step": 5387
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.629478510028653e-05,
+      "loss": 0.0255,
+      "step": 5388
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.629409742120344e-05,
+      "loss": 0.0368,
+      "step": 5389
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.6293409742120344e-05,
+      "loss": 0.0447,
+      "step": 5390
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.629272206303725e-05,
+      "loss": 0.0468,
+      "step": 5391
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.6292034383954155e-05,
+      "loss": 0.0321,
+      "step": 5392
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.6291346704871064e-05,
+      "loss": 0.0298,
+      "step": 5393
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.6290659025787966e-05,
+      "loss": 0.0183,
+      "step": 5394
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.6289971346704875e-05,
+      "loss": 0.0375,
+      "step": 5395
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.628928366762178e-05,
+      "loss": 0.0305,
+      "step": 5396
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.6288595988538686e-05,
+      "loss": 0.0181,
+      "step": 5397
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.628790830945559e-05,
+      "loss": 0.0433,
+      "step": 5398
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.62872206303725e-05,
+      "loss": 0.0283,
+      "step": 5399
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 5.62865329512894e-05,
+      "loss": 0.0589,
+      "step": 5400
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.628584527220631e-05,
+      "loss": 0.0258,
+      "step": 5401
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.628515759312321e-05,
+      "loss": 0.0409,
+      "step": 5402
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.628446991404012e-05,
+      "loss": 0.0552,
+      "step": 5403
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.628378223495702e-05,
+      "loss": 0.039,
+      "step": 5404
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.628309455587393e-05,
+      "loss": 0.0501,
+      "step": 5405
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.628240687679083e-05,
+      "loss": 0.032,
+      "step": 5406
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.628171919770774e-05,
+      "loss": 0.024,
+      "step": 5407
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.628103151862464e-05,
+      "loss": 0.0585,
+      "step": 5408
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.628034383954155e-05,
+      "loss": 0.0398,
+      "step": 5409
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.6279656160458454e-05,
+      "loss": 0.0432,
+      "step": 5410
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.627896848137536e-05,
+      "loss": 0.0492,
+      "step": 5411
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.6278280802292265e-05,
+      "loss": 0.0265,
+      "step": 5412
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.6277593123209174e-05,
+      "loss": 0.0399,
+      "step": 5413
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.6276905444126076e-05,
+      "loss": 0.0412,
+      "step": 5414
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.6276217765042985e-05,
+      "loss": 0.049,
+      "step": 5415
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.627553008595989e-05,
+      "loss": 0.0146,
+      "step": 5416
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.6274842406876796e-05,
+      "loss": 0.1009,
+      "step": 5417
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 5.62741547277937e-05,
+      "loss": 0.0212,
+      "step": 5418
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.627346704871061e-05,
+      "loss": 0.0617,
+      "step": 5419
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.627277936962751e-05,
+      "loss": 0.0426,
+      "step": 5420
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.627209169054442e-05,
+      "loss": 0.0349,
+      "step": 5421
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.627140401146132e-05,
+      "loss": 0.0431,
+      "step": 5422
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.627071633237823e-05,
+      "loss": 0.0227,
+      "step": 5423
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.627002865329513e-05,
+      "loss": 0.0341,
+      "step": 5424
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.626934097421204e-05,
+      "loss": 0.0434,
+      "step": 5425
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.626865329512894e-05,
+      "loss": 0.0388,
+      "step": 5426
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.626796561604585e-05,
+      "loss": 0.0289,
+      "step": 5427
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.626727793696275e-05,
+      "loss": 0.0356,
+      "step": 5428
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.626659025787966e-05,
+      "loss": 0.0395,
+      "step": 5429
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.6265902578796564e-05,
+      "loss": 0.0528,
+      "step": 5430
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.626521489971347e-05,
+      "loss": 0.0252,
+      "step": 5431
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.6264527220630375e-05,
+      "loss": 0.0324,
+      "step": 5432
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.6263839541547284e-05,
+      "loss": 0.0352,
+      "step": 5433
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.6263151862464186e-05,
+      "loss": 0.043,
+      "step": 5434
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 5.6262464183381095e-05,
+      "loss": 0.035,
+      "step": 5435
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.6261776504298e-05,
+      "loss": 0.0289,
+      "step": 5436
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.6261088825214906e-05,
+      "loss": 0.0531,
+      "step": 5437
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.626040114613181e-05,
+      "loss": 0.0422,
+      "step": 5438
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625971346704872e-05,
+      "loss": 0.0535,
+      "step": 5439
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625902578796562e-05,
+      "loss": 0.0474,
+      "step": 5440
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625833810888253e-05,
+      "loss": 0.0473,
+      "step": 5441
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625765042979943e-05,
+      "loss": 0.0625,
+      "step": 5442
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625696275071634e-05,
+      "loss": 0.0371,
+      "step": 5443
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625627507163324e-05,
+      "loss": 0.0221,
+      "step": 5444
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625558739255015e-05,
+      "loss": 0.0522,
+      "step": 5445
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625489971346705e-05,
+      "loss": 0.0398,
+      "step": 5446
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625421203438396e-05,
+      "loss": 0.0456,
+      "step": 5447
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625352435530086e-05,
+      "loss": 0.0284,
+      "step": 5448
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625283667621777e-05,
+      "loss": 0.0375,
+      "step": 5449
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.6252148997134674e-05,
+      "loss": 0.0358,
+      "step": 5450
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.625146131805158e-05,
+      "loss": 0.0294,
+      "step": 5451
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.6250773638968485e-05,
+      "loss": 0.033,
+      "step": 5452
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 5.6250085959885394e-05,
+      "loss": 0.0316,
+      "step": 5453
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.6249398280802296e-05,
+      "loss": 0.0417,
+      "step": 5454
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.6248710601719205e-05,
+      "loss": 0.0289,
+      "step": 5455
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.624802292263611e-05,
+      "loss": 0.0354,
+      "step": 5456
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.6247335243553016e-05,
+      "loss": 0.029,
+      "step": 5457
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.624664756446992e-05,
+      "loss": 0.0293,
+      "step": 5458
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.624595988538683e-05,
+      "loss": 0.0395,
+      "step": 5459
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.624527220630373e-05,
+      "loss": 0.0244,
+      "step": 5460
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.624458452722064e-05,
+      "loss": 0.0404,
+      "step": 5461
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.624389684813754e-05,
+      "loss": 0.0477,
+      "step": 5462
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.624320916905445e-05,
+      "loss": 0.0365,
+      "step": 5463
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.624252148997135e-05,
+      "loss": 0.0388,
+      "step": 5464
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.624183381088826e-05,
+      "loss": 0.0265,
+      "step": 5465
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.624114613180516e-05,
+      "loss": 0.06,
+      "step": 5466
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.6240458452722064e-05,
+      "loss": 0.0533,
+      "step": 5467
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.623977077363897e-05,
+      "loss": 0.0356,
+      "step": 5468
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.6239083094555875e-05,
+      "loss": 0.0417,
+      "step": 5469
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 5.623839541547278e-05,
+      "loss": 0.0256,
+      "step": 5470
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.6237707736389686e-05,
+      "loss": 0.0294,
+      "step": 5471
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.623702005730659e-05,
+      "loss": 0.0326,
+      "step": 5472
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.623633237822349e-05,
+      "loss": 0.0199,
+      "step": 5473
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.62356446991404e-05,
+      "loss": 0.0331,
+      "step": 5474
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.62349570200573e-05,
+      "loss": 0.0556,
+      "step": 5475
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.623426934097421e-05,
+      "loss": 0.0384,
+      "step": 5476
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.623358166189111e-05,
+      "loss": 0.0259,
+      "step": 5477
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.623289398280802e-05,
+      "loss": 0.0421,
+      "step": 5478
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.6232206303724924e-05,
+      "loss": 0.0356,
+      "step": 5479
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.623151862464183e-05,
+      "loss": 0.0166,
+      "step": 5480
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.6230830945558735e-05,
+      "loss": 0.0356,
+      "step": 5481
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.6230143266475643e-05,
+      "loss": 0.0284,
+      "step": 5482
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.6229455587392546e-05,
+      "loss": 0.0208,
+      "step": 5483
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.6228767908309454e-05,
+      "loss": 0.0308,
+      "step": 5484
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.6228080229226357e-05,
+      "loss": 0.0359,
+      "step": 5485
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.6227392550143265e-05,
+      "loss": 0.0394,
+      "step": 5486
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.622670487106017e-05,
+      "loss": 0.0341,
+      "step": 5487
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 5.6226017191977076e-05,
+      "loss": 0.0539,
+      "step": 5488
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.622532951289398e-05,
+      "loss": 0.0363,
+      "step": 5489
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.622464183381089e-05,
+      "loss": 0.0348,
+      "step": 5490
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.622395415472779e-05,
+      "loss": 0.0323,
+      "step": 5491
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.62232664756447e-05,
+      "loss": 0.0486,
+      "step": 5492
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.62225787965616e-05,
+      "loss": 0.0332,
+      "step": 5493
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.622189111747851e-05,
+      "loss": 0.0678,
+      "step": 5494
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.622120343839541e-05,
+      "loss": 0.0396,
+      "step": 5495
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.622051575931232e-05,
+      "loss": 0.0253,
+      "step": 5496
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.621982808022922e-05,
+      "loss": 0.0464,
+      "step": 5497
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.621914040114613e-05,
+      "loss": 0.0329,
+      "step": 5498
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.6218452722063034e-05,
+      "loss": 0.0336,
+      "step": 5499
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.621776504297994e-05,
+      "loss": 0.0457,
+      "step": 5500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.6217077363896845e-05,
+      "loss": 0.0383,
+      "step": 5501
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.6216389684813753e-05,
+      "loss": 0.0357,
+      "step": 5502
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.6215702005730656e-05,
+      "loss": 0.0367,
+      "step": 5503
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.6215014326647564e-05,
+      "loss": 0.0273,
+      "step": 5504
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 5.6214326647564467e-05,
+      "loss": 0.0674,
+      "step": 5505
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.6213638968481375e-05,
+      "loss": 0.032,
+      "step": 5506
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.621295128939828e-05,
+      "loss": 0.0332,
+      "step": 5507
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.6212263610315186e-05,
+      "loss": 0.0286,
+      "step": 5508
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.621157593123209e-05,
+      "loss": 0.0266,
+      "step": 5509
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.6210888252149e-05,
+      "loss": 0.0469,
+      "step": 5510
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.62102005730659e-05,
+      "loss": 0.0333,
+      "step": 5511
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.620951289398281e-05,
+      "loss": 0.0262,
+      "step": 5512
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.620882521489971e-05,
+      "loss": 0.0756,
+      "step": 5513
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.620813753581662e-05,
+      "loss": 0.0532,
+      "step": 5514
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.620744985673352e-05,
+      "loss": 0.0372,
+      "step": 5515
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.620676217765043e-05,
+      "loss": 0.0386,
+      "step": 5516
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.620607449856733e-05,
+      "loss": 0.0421,
+      "step": 5517
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.620538681948424e-05,
+      "loss": 0.0434,
+      "step": 5518
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.6204699140401144e-05,
+      "loss": 0.0235,
+      "step": 5519
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.620401146131805e-05,
+      "loss": 0.0483,
+      "step": 5520
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.6203323782234955e-05,
+      "loss": 0.0402,
+      "step": 5521
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 5.6202636103151863e-05,
+      "loss": 0.0215,
+      "step": 5522
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.6201948424068766e-05,
+      "loss": 0.0205,
+      "step": 5523
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.6201260744985674e-05,
+      "loss": 0.0191,
+      "step": 5524
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.6200573065902577e-05,
+      "loss": 0.048,
+      "step": 5525
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.6199885386819485e-05,
+      "loss": 0.0291,
+      "step": 5526
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619919770773639e-05,
+      "loss": 0.0285,
+      "step": 5527
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.6198510028653296e-05,
+      "loss": 0.0593,
+      "step": 5528
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.61978223495702e-05,
+      "loss": 0.0337,
+      "step": 5529
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619713467048711e-05,
+      "loss": 0.0482,
+      "step": 5530
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619644699140401e-05,
+      "loss": 0.0478,
+      "step": 5531
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619575931232092e-05,
+      "loss": 0.0465,
+      "step": 5532
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619507163323782e-05,
+      "loss": 0.0433,
+      "step": 5533
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619438395415473e-05,
+      "loss": 0.0319,
+      "step": 5534
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619369627507163e-05,
+      "loss": 0.0419,
+      "step": 5535
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619300859598854e-05,
+      "loss": 0.0391,
+      "step": 5536
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619232091690544e-05,
+      "loss": 0.0203,
+      "step": 5537
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619163323782235e-05,
+      "loss": 0.0323,
+      "step": 5538
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.6190945558739254e-05,
+      "loss": 0.0392,
+      "step": 5539
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 5.619025787965616e-05,
+      "loss": 0.0586,
+      "step": 5540
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.6189570200573065e-05,
+      "loss": 0.0209,
+      "step": 5541
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.6188882521489973e-05,
+      "loss": 0.0433,
+      "step": 5542
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.6188194842406876e-05,
+      "loss": 0.0449,
+      "step": 5543
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.6187507163323784e-05,
+      "loss": 0.0259,
+      "step": 5544
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.6186819484240687e-05,
+      "loss": 0.0649,
+      "step": 5545
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.6186131805157595e-05,
+      "loss": 0.0364,
+      "step": 5546
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.61854441260745e-05,
+      "loss": 0.0301,
+      "step": 5547
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.6184756446991406e-05,
+      "loss": 0.0583,
+      "step": 5548
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.618406876790831e-05,
+      "loss": 0.0458,
+      "step": 5549
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.618338108882522e-05,
+      "loss": 0.0461,
+      "step": 5550
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.618269340974212e-05,
+      "loss": 0.0399,
+      "step": 5551
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.618200573065903e-05,
+      "loss": 0.0399,
+      "step": 5552
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.618131805157593e-05,
+      "loss": 0.0388,
+      "step": 5553
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.618063037249284e-05,
+      "loss": 0.0229,
+      "step": 5554
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.617994269340974e-05,
+      "loss": 0.0416,
+      "step": 5555
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.617925501432665e-05,
+      "loss": 0.0347,
+      "step": 5556
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 5.617856733524355e-05,
+      "loss": 0.0525,
+      "step": 5557
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.617787965616046e-05,
+      "loss": 0.031,
+      "step": 5558
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.6177191977077364e-05,
+      "loss": 0.0311,
+      "step": 5559
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.617650429799427e-05,
+      "loss": 0.0287,
+      "step": 5560
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.6175816618911175e-05,
+      "loss": 0.0348,
+      "step": 5561
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.6175128939828083e-05,
+      "loss": 0.0291,
+      "step": 5562
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.6174441260744986e-05,
+      "loss": 0.0419,
+      "step": 5563
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.6173753581661894e-05,
+      "loss": 0.0345,
+      "step": 5564
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.6173065902578797e-05,
+      "loss": 0.017,
+      "step": 5565
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.6172378223495705e-05,
+      "loss": 0.0208,
+      "step": 5566
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.617169054441261e-05,
+      "loss": 0.0277,
+      "step": 5567
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.6171002865329516e-05,
+      "loss": 0.0465,
+      "step": 5568
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.617031518624642e-05,
+      "loss": 0.0524,
+      "step": 5569
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.616962750716333e-05,
+      "loss": 0.0367,
+      "step": 5570
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.616893982808023e-05,
+      "loss": 0.0257,
+      "step": 5571
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.616825214899714e-05,
+      "loss": 0.0667,
+      "step": 5572
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.616756446991404e-05,
+      "loss": 0.0318,
+      "step": 5573
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.616687679083095e-05,
+      "loss": 0.0263,
+      "step": 5574
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 5.616618911174785e-05,
+      "loss": 0.0229,
+      "step": 5575
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.616550143266476e-05,
+      "loss": 0.0377,
+      "step": 5576
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.616481375358166e-05,
+      "loss": 0.0329,
+      "step": 5577
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.616412607449857e-05,
+      "loss": 0.0342,
+      "step": 5578
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.6163438395415474e-05,
+      "loss": 0.0325,
+      "step": 5579
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.616275071633238e-05,
+      "loss": 0.0218,
+      "step": 5580
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.6162063037249285e-05,
+      "loss": 0.0462,
+      "step": 5581
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.6161375358166193e-05,
+      "loss": 0.0246,
+      "step": 5582
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.6160687679083096e-05,
+      "loss": 0.0412,
+      "step": 5583
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.6160000000000004e-05,
+      "loss": 0.0446,
+      "step": 5584
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.6159312320916907e-05,
+      "loss": 0.0347,
+      "step": 5585
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.6158624641833815e-05,
+      "loss": 0.0388,
+      "step": 5586
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.615793696275072e-05,
+      "loss": 0.0264,
+      "step": 5587
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.615724928366762e-05,
+      "loss": 0.0241,
+      "step": 5588
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.615656160458453e-05,
+      "loss": 0.0452,
+      "step": 5589
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.615587392550143e-05,
+      "loss": 0.025,
+      "step": 5590
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.615518624641834e-05,
+      "loss": 0.0445,
+      "step": 5591
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 5.615449856733524e-05,
+      "loss": 0.0194,
+      "step": 5592
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.615381088825215e-05,
+      "loss": 0.0305,
+      "step": 5593
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.615312320916905e-05,
+      "loss": 0.0286,
+      "step": 5594
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.615243553008596e-05,
+      "loss": 0.0272,
+      "step": 5595
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.6151747851002864e-05,
+      "loss": 0.0308,
+      "step": 5596
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.615106017191977e-05,
+      "loss": 0.046,
+      "step": 5597
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.6150372492836675e-05,
+      "loss": 0.0552,
+      "step": 5598
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.6149684813753584e-05,
+      "loss": 0.0452,
+      "step": 5599
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.6148997134670486e-05,
+      "loss": 0.0336,
+      "step": 5600
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.6148309455587395e-05,
+      "loss": 0.0427,
+      "step": 5601
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.61476217765043e-05,
+      "loss": 0.0214,
+      "step": 5602
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.6146934097421206e-05,
+      "loss": 0.0163,
+      "step": 5603
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.614624641833811e-05,
+      "loss": 0.0388,
+      "step": 5604
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.6145558739255017e-05,
+      "loss": 0.0337,
+      "step": 5605
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.614487106017192e-05,
+      "loss": 0.0531,
+      "step": 5606
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.614418338108883e-05,
+      "loss": 0.0379,
+      "step": 5607
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.614349570200573e-05,
+      "loss": 0.0302,
+      "step": 5608
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.614280802292264e-05,
+      "loss": 0.0424,
+      "step": 5609
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.614212034383954e-05,
+      "loss": 0.0373,
+      "step": 5610
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.614143266475645e-05,
+      "loss": 0.0399,
+      "step": 5611
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.614074498567335e-05,
+      "loss": 0.0268,
+      "step": 5612
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.614005730659026e-05,
+      "loss": 0.0267,
+      "step": 5613
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.613936962750716e-05,
+      "loss": 0.0697,
+      "step": 5614
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.613868194842407e-05,
+      "loss": 0.0451,
+      "step": 5615
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.6137994269340974e-05,
+      "loss": 0.0437,
+      "step": 5616
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.613730659025788e-05,
+      "loss": 0.065,
+      "step": 5617
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.6136618911174785e-05,
+      "loss": 0.0278,
+      "step": 5618
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.6135931232091693e-05,
+      "loss": 0.0538,
+      "step": 5619
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.6135243553008596e-05,
+      "loss": 0.0258,
+      "step": 5620
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.6134555873925504e-05,
+      "loss": 0.0418,
+      "step": 5621
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.6133868194842407e-05,
+      "loss": 0.0444,
+      "step": 5622
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.6133180515759315e-05,
+      "loss": 0.0287,
+      "step": 5623
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.613249283667622e-05,
+      "loss": 0.045,
+      "step": 5624
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.6131805157593126e-05,
+      "loss": 0.0544,
+      "step": 5625
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.613111747851003e-05,
+      "loss": 0.0391,
+      "step": 5626
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.613042979942694e-05,
+      "loss": 0.0442,
+      "step": 5627
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.612974212034384e-05,
+      "loss": 0.0345,
+      "step": 5628
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.612905444126075e-05,
+      "loss": 0.047,
+      "step": 5629
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.612836676217765e-05,
+      "loss": 0.0312,
+      "step": 5630
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.612767908309456e-05,
+      "loss": 0.0381,
+      "step": 5631
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.612699140401146e-05,
+      "loss": 0.0284,
+      "step": 5632
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.612630372492837e-05,
+      "loss": 0.0301,
+      "step": 5633
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.612561604584527e-05,
+      "loss": 0.0181,
+      "step": 5634
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.612492836676218e-05,
+      "loss": 0.0488,
+      "step": 5635
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.6124240687679084e-05,
+      "loss": 0.0357,
+      "step": 5636
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.612355300859599e-05,
+      "loss": 0.0269,
+      "step": 5637
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.6122865329512895e-05,
+      "loss": 0.0553,
+      "step": 5638
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.6122177650429803e-05,
+      "loss": 0.0329,
+      "step": 5639
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.6121489971346706e-05,
+      "loss": 0.0283,
+      "step": 5640
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.6120802292263614e-05,
+      "loss": 0.0312,
+      "step": 5641
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.6120114613180517e-05,
+      "loss": 0.0165,
+      "step": 5642
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.6119426934097425e-05,
+      "loss": 0.0442,
+      "step": 5643
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.611873925501433e-05,
+      "loss": 0.0356,
+      "step": 5644
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.6118051575931236e-05,
+      "loss": 0.0198,
+      "step": 5645
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.611736389684814e-05,
+      "loss": 0.0254,
+      "step": 5646
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.611667621776505e-05,
+      "loss": 0.0349,
+      "step": 5647
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.611598853868195e-05,
+      "loss": 0.0601,
+      "step": 5648
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.611530085959886e-05,
+      "loss": 0.0366,
+      "step": 5649
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.611461318051576e-05,
+      "loss": 0.0384,
+      "step": 5650
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.611392550143267e-05,
+      "loss": 0.0341,
+      "step": 5651
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.611323782234957e-05,
+      "loss": 0.0452,
+      "step": 5652
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.611255014326648e-05,
+      "loss": 0.0597,
+      "step": 5653
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.611186246418338e-05,
+      "loss": 0.0475,
+      "step": 5654
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.611117478510029e-05,
+      "loss": 0.0422,
+      "step": 5655
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.6110487106017194e-05,
+      "loss": 0.0524,
+      "step": 5656
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.61097994269341e-05,
+      "loss": 0.0368,
+      "step": 5657
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.6109111747851005e-05,
+      "loss": 0.0346,
+      "step": 5658
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.6108424068767913e-05,
+      "loss": 0.0681,
+      "step": 5659
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.6107736389684816e-05,
+      "loss": 0.0349,
+      "step": 5660
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.6107048710601724e-05,
+      "loss": 0.0327,
+      "step": 5661
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.6106361031518627e-05,
+      "loss": 0.031,
+      "step": 5662
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.6105673352435535e-05,
+      "loss": 0.0247,
+      "step": 5663
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.610498567335244e-05,
+      "loss": 0.0324,
+      "step": 5664
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.6104297994269346e-05,
+      "loss": 0.0337,
+      "step": 5665
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.610361031518625e-05,
+      "loss": 0.034,
+      "step": 5666
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.610292263610316e-05,
+      "loss": 0.0518,
+      "step": 5667
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.610223495702006e-05,
+      "loss": 0.0416,
+      "step": 5668
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.610154727793697e-05,
+      "loss": 0.0442,
+      "step": 5669
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.610085959885387e-05,
+      "loss": 0.0342,
+      "step": 5670
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.610017191977078e-05,
+      "loss": 0.0333,
+      "step": 5671
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.609948424068768e-05,
+      "loss": 0.0329,
+      "step": 5672
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.609879656160459e-05,
+      "loss": 0.0466,
+      "step": 5673
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.609810888252149e-05,
+      "loss": 0.0518,
+      "step": 5674
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.60974212034384e-05,
+      "loss": 0.0404,
+      "step": 5675
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.6096733524355304e-05,
+      "loss": 0.0368,
+      "step": 5676
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.609604584527221e-05,
+      "loss": 0.0612,
+      "step": 5677
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.6095358166189115e-05,
+      "loss": 0.0398,
+      "step": 5678
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.6094670487106023e-05,
+      "loss": 0.0465,
+      "step": 5679
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6093982808022926e-05,
+      "loss": 0.0599,
+      "step": 5680
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6093295128939834e-05,
+      "loss": 0.0357,
+      "step": 5681
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6092607449856737e-05,
+      "loss": 0.0422,
+      "step": 5682
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6091919770773645e-05,
+      "loss": 0.0441,
+      "step": 5683
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.609123209169055e-05,
+      "loss": 0.0439,
+      "step": 5684
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6090544412607456e-05,
+      "loss": 0.0522,
+      "step": 5685
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.608985673352436e-05,
+      "loss": 0.0296,
+      "step": 5686
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.608916905444127e-05,
+      "loss": 0.036,
+      "step": 5687
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.608848137535817e-05,
+      "loss": 0.0401,
+      "step": 5688
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.608779369627508e-05,
+      "loss": 0.0436,
+      "step": 5689
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.608710601719198e-05,
+      "loss": 0.0227,
+      "step": 5690
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.608641833810889e-05,
+      "loss": 0.0778,
+      "step": 5691
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.608573065902579e-05,
+      "loss": 0.0472,
+      "step": 5692
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.60850429799427e-05,
+      "loss": 0.0339,
+      "step": 5693
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.60843553008596e-05,
+      "loss": 0.0308,
+      "step": 5694
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.608366762177651e-05,
+      "loss": 0.0513,
+      "step": 5695
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.6082979942693414e-05,
+      "loss": 0.0604,
+      "step": 5696
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.608229226361032e-05,
+      "loss": 0.025,
+      "step": 5697
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6081604584527225e-05,
+      "loss": 0.0268,
+      "step": 5698
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6080916905444133e-05,
+      "loss": 0.0432,
+      "step": 5699
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6080229226361036e-05,
+      "loss": 0.0527,
+      "step": 5700
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6079541547277944e-05,
+      "loss": 0.0392,
+      "step": 5701
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6078853868194847e-05,
+      "loss": 0.0406,
+      "step": 5702
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6078166189111755e-05,
+      "loss": 0.0432,
+      "step": 5703
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.607747851002866e-05,
+      "loss": 0.0287,
+      "step": 5704
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6076790830945566e-05,
+      "loss": 0.0419,
+      "step": 5705
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.607610315186247e-05,
+      "loss": 0.0278,
+      "step": 5706
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.607541547277937e-05,
+      "loss": 0.0345,
+      "step": 5707
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.607472779369628e-05,
+      "loss": 0.0324,
+      "step": 5708
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.607404011461318e-05,
+      "loss": 0.0597,
+      "step": 5709
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.607335243553009e-05,
+      "loss": 0.038,
+      "step": 5710
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.607266475644699e-05,
+      "loss": 0.0367,
+      "step": 5711
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.60719770773639e-05,
+      "loss": 0.0451,
+      "step": 5712
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6071289398280804e-05,
+      "loss": 0.0248,
+      "step": 5713
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.607060171919771e-05,
+      "loss": 0.0382,
+      "step": 5714
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6069914040114615e-05,
+      "loss": 0.0417,
+      "step": 5715
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6069226361031524e-05,
+      "loss": 0.0376,
+      "step": 5716
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6068538681948426e-05,
+      "loss": 0.051,
+      "step": 5717
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6067851002865335e-05,
+      "loss": 0.0372,
+      "step": 5718
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.606716332378224e-05,
+      "loss": 0.0788,
+      "step": 5719
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6066475644699146e-05,
+      "loss": 0.0409,
+      "step": 5720
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.606578796561605e-05,
+      "loss": 0.0501,
+      "step": 5721
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6065100286532957e-05,
+      "loss": 0.0291,
+      "step": 5722
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.606441260744986e-05,
+      "loss": 0.0198,
+      "step": 5723
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.606372492836676e-05,
+      "loss": 0.0262,
+      "step": 5724
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.606303724928367e-05,
+      "loss": 0.0355,
+      "step": 5725
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.606234957020057e-05,
+      "loss": 0.0338,
+      "step": 5726
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6061661891117474e-05,
+      "loss": 0.0402,
+      "step": 5727
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.606097421203438e-05,
+      "loss": 0.037,
+      "step": 5728
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6060286532951285e-05,
+      "loss": 0.0167,
+      "step": 5729
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6059598853868194e-05,
+      "loss": 0.0931,
+      "step": 5730
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6058911174785096e-05,
+      "loss": 0.0244,
+      "step": 5731
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.6058223495702005e-05,
+      "loss": 0.0434,
+      "step": 5732
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.605753581661891e-05,
+      "loss": 0.0357,
+      "step": 5733
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.6056848137535816e-05,
+      "loss": 0.0423,
+      "step": 5734
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.605616045845272e-05,
+      "loss": 0.0373,
+      "step": 5735
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.605547277936963e-05,
+      "loss": 0.0228,
+      "step": 5736
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.605478510028653e-05,
+      "loss": 0.0457,
+      "step": 5737
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.605409742120344e-05,
+      "loss": 0.0224,
+      "step": 5738
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.605340974212034e-05,
+      "loss": 0.0193,
+      "step": 5739
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.605272206303725e-05,
+      "loss": 0.0399,
+      "step": 5740
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.605203438395415e-05,
+      "loss": 0.036,
+      "step": 5741
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.605134670487106e-05,
+      "loss": 0.023,
+      "step": 5742
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.605065902578796e-05,
+      "loss": 0.0331,
+      "step": 5743
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.604997134670487e-05,
+      "loss": 0.0386,
+      "step": 5744
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.604928366762177e-05,
+      "loss": 0.0337,
+      "step": 5745
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.604859598853868e-05,
+      "loss": 0.087,
+      "step": 5746
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.6047908309455584e-05,
+      "loss": 0.0435,
+      "step": 5747
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.604722063037249e-05,
+      "loss": 0.0206,
+      "step": 5748
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.6046532951289395e-05,
+      "loss": 0.0432,
+      "step": 5749
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.6045845272206304e-05,
+      "loss": 0.0349,
+      "step": 5750
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.6045157593123206e-05,
+      "loss": 0.0339,
+      "step": 5751
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.6044469914040115e-05,
+      "loss": 0.0393,
+      "step": 5752
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.604378223495702e-05,
+      "loss": 0.0441,
+      "step": 5753
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.6043094555873926e-05,
+      "loss": 0.0249,
+      "step": 5754
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.604240687679083e-05,
+      "loss": 0.0419,
+      "step": 5755
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.604171919770774e-05,
+      "loss": 0.0515,
+      "step": 5756
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.604103151862464e-05,
+      "loss": 0.0394,
+      "step": 5757
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.604034383954155e-05,
+      "loss": 0.0384,
+      "step": 5758
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.603965616045845e-05,
+      "loss": 0.0233,
+      "step": 5759
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.603896848137536e-05,
+      "loss": 0.0539,
+      "step": 5760
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.603828080229226e-05,
+      "loss": 0.0452,
+      "step": 5761
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.603759312320917e-05,
+      "loss": 0.1045,
+      "step": 5762
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.603690544412607e-05,
+      "loss": 0.0635,
+      "step": 5763
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.603621776504298e-05,
+      "loss": 0.0494,
+      "step": 5764
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.603553008595988e-05,
+      "loss": 0.0311,
+      "step": 5765
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.603484240687679e-05,
+      "loss": 0.042,
+      "step": 5766
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.6034154727793694e-05,
+      "loss": 0.0309,
+      "step": 5767
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.60334670487106e-05,
+      "loss": 0.0424,
+      "step": 5768
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.6032779369627505e-05,
+      "loss": 0.0315,
+      "step": 5769
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.6032091690544414e-05,
+      "loss": 0.0345,
+      "step": 5770
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.6031404011461316e-05,
+      "loss": 0.0304,
+      "step": 5771
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.6030716332378225e-05,
+      "loss": 0.03,
+      "step": 5772
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.603002865329513e-05,
+      "loss": 0.045,
+      "step": 5773
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.6029340974212036e-05,
+      "loss": 0.028,
+      "step": 5774
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.602865329512894e-05,
+      "loss": 0.0442,
+      "step": 5775
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.602796561604585e-05,
+      "loss": 0.0419,
+      "step": 5776
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.602727793696275e-05,
+      "loss": 0.032,
+      "step": 5777
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.602659025787966e-05,
+      "loss": 0.028,
+      "step": 5778
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.602590257879656e-05,
+      "loss": 0.0236,
+      "step": 5779
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.602521489971347e-05,
+      "loss": 0.043,
+      "step": 5780
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.602452722063037e-05,
+      "loss": 0.05,
+      "step": 5781
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.602383954154728e-05,
+      "loss": 0.051,
+      "step": 5782
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.602315186246418e-05,
+      "loss": 0.0449,
+      "step": 5783
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.602246418338109e-05,
+      "loss": 0.0284,
+      "step": 5784
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.602177650429799e-05,
+      "loss": 0.0312,
+      "step": 5785
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.60210888252149e-05,
+      "loss": 0.0303,
+      "step": 5786
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.6020401146131804e-05,
+      "loss": 0.0386,
+      "step": 5787
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.601971346704871e-05,
+      "loss": 0.038,
+      "step": 5788
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.6019025787965615e-05,
+      "loss": 0.0585,
+      "step": 5789
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.6018338108882524e-05,
+      "loss": 0.03,
+      "step": 5790
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.6017650429799426e-05,
+      "loss": 0.0412,
+      "step": 5791
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.6016962750716335e-05,
+      "loss": 0.0263,
+      "step": 5792
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.601627507163324e-05,
+      "loss": 0.0239,
+      "step": 5793
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.6015587392550146e-05,
+      "loss": 0.0237,
+      "step": 5794
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.601489971346705e-05,
+      "loss": 0.0239,
+      "step": 5795
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.601421203438396e-05,
+      "loss": 0.0243,
+      "step": 5796
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.601352435530086e-05,
+      "loss": 0.0395,
+      "step": 5797
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.601283667621777e-05,
+      "loss": 0.0355,
+      "step": 5798
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.601214899713467e-05,
+      "loss": 0.0303,
+      "step": 5799
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.601146131805158e-05,
+      "loss": 0.0451,
+      "step": 5800
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.601077363896848e-05,
+      "loss": 0.0379,
+      "step": 5801
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.601008595988539e-05,
+      "loss": 0.0309,
+      "step": 5802
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.600939828080229e-05,
+      "loss": 0.0356,
+      "step": 5803
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.60087106017192e-05,
+      "loss": 0.0599,
+      "step": 5804
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.60080229226361e-05,
+      "loss": 0.0344,
+      "step": 5805
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.600733524355301e-05,
+      "loss": 0.0327,
+      "step": 5806
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.6006647564469914e-05,
+      "loss": 0.0328,
+      "step": 5807
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.600595988538682e-05,
+      "loss": 0.027,
+      "step": 5808
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.6005272206303725e-05,
+      "loss": 0.0404,
+      "step": 5809
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.6004584527220634e-05,
+      "loss": 0.0288,
+      "step": 5810
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.6003896848137536e-05,
+      "loss": 0.0401,
+      "step": 5811
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.6003209169054445e-05,
+      "loss": 0.0632,
+      "step": 5812
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.600252148997135e-05,
+      "loss": 0.0419,
+      "step": 5813
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.6001833810888256e-05,
+      "loss": 0.0184,
+      "step": 5814
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.600114613180516e-05,
+      "loss": 0.0256,
+      "step": 5815
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.600045845272207e-05,
+      "loss": 0.0489,
+      "step": 5816
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.599977077363897e-05,
+      "loss": 0.0328,
+      "step": 5817
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.599908309455588e-05,
+      "loss": 0.0459,
+      "step": 5818
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.599839541547278e-05,
+      "loss": 0.059,
+      "step": 5819
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.599770773638969e-05,
+      "loss": 0.0332,
+      "step": 5820
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.599702005730659e-05,
+      "loss": 0.0304,
+      "step": 5821
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.59963323782235e-05,
+      "loss": 0.026,
+      "step": 5822
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.59956446991404e-05,
+      "loss": 0.0225,
+      "step": 5823
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.599495702005731e-05,
+      "loss": 0.0294,
+      "step": 5824
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.599426934097421e-05,
+      "loss": 0.0502,
+      "step": 5825
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.5993581661891115e-05,
+      "loss": 0.0398,
+      "step": 5826
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.5992893982808024e-05,
+      "loss": 0.031,
+      "step": 5827
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.5992206303724926e-05,
+      "loss": 0.0448,
+      "step": 5828
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.5991518624641835e-05,
+      "loss": 0.0325,
+      "step": 5829
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.599083094555874e-05,
+      "loss": 0.0314,
+      "step": 5830
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.5990143266475646e-05,
+      "loss": 0.0306,
+      "step": 5831
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.598945558739255e-05,
+      "loss": 0.0678,
+      "step": 5832
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.598876790830946e-05,
+      "loss": 0.0777,
+      "step": 5833
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.598808022922636e-05,
+      "loss": 0.0326,
+      "step": 5834
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.598739255014327e-05,
+      "loss": 0.0309,
+      "step": 5835
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.598670487106017e-05,
+      "loss": 0.0366,
+      "step": 5836
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.598601719197708e-05,
+      "loss": 0.0382,
+      "step": 5837
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.598532951289398e-05,
+      "loss": 0.0356,
+      "step": 5838
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.598464183381089e-05,
+      "loss": 0.0382,
+      "step": 5839
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.598395415472779e-05,
+      "loss": 0.0501,
+      "step": 5840
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.59832664756447e-05,
+      "loss": 0.0428,
+      "step": 5841
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.59825787965616e-05,
+      "loss": 0.0274,
+      "step": 5842
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.598189111747851e-05,
+      "loss": 0.0252,
+      "step": 5843
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.5981203438395414e-05,
+      "loss": 0.0541,
+      "step": 5844
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.598051575931232e-05,
+      "loss": 0.0224,
+      "step": 5845
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.5979828080229225e-05,
+      "loss": 0.022,
+      "step": 5846
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.5979140401146134e-05,
+      "loss": 0.0346,
+      "step": 5847
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.5978452722063036e-05,
+      "loss": 0.021,
+      "step": 5848
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.5977765042979945e-05,
+      "loss": 0.0412,
+      "step": 5849
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.597707736389685e-05,
+      "loss": 0.0215,
+      "step": 5850
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.5976389684813756e-05,
+      "loss": 0.0564,
+      "step": 5851
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.597570200573066e-05,
+      "loss": 0.0396,
+      "step": 5852
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.597501432664757e-05,
+      "loss": 0.0452,
+      "step": 5853
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.597432664756447e-05,
+      "loss": 0.0542,
+      "step": 5854
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.597363896848138e-05,
+      "loss": 0.0255,
+      "step": 5855
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.597295128939828e-05,
+      "loss": 0.029,
+      "step": 5856
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.597226361031519e-05,
+      "loss": 0.0234,
+      "step": 5857
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.597157593123209e-05,
+      "loss": 0.0286,
+      "step": 5858
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.5970888252149e-05,
+      "loss": 0.0261,
+      "step": 5859
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.59702005730659e-05,
+      "loss": 0.039,
+      "step": 5860
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.596951289398281e-05,
+      "loss": 0.0333,
+      "step": 5861
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.596882521489971e-05,
+      "loss": 0.0599,
+      "step": 5862
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.596813753581662e-05,
+      "loss": 0.0458,
+      "step": 5863
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.5967449856733524e-05,
+      "loss": 0.0349,
+      "step": 5864
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.596676217765043e-05,
+      "loss": 0.047,
+      "step": 5865
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.5966074498567335e-05,
+      "loss": 0.0669,
+      "step": 5866
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.5965386819484244e-05,
+      "loss": 0.0341,
+      "step": 5867
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.5964699140401146e-05,
+      "loss": 0.0459,
+      "step": 5868
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.5964011461318055e-05,
+      "loss": 0.0246,
+      "step": 5869
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.596332378223496e-05,
+      "loss": 0.0406,
+      "step": 5870
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.5962636103151866e-05,
+      "loss": 0.0253,
+      "step": 5871
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.596194842406877e-05,
+      "loss": 0.0473,
+      "step": 5872
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.596126074498568e-05,
+      "loss": 0.0235,
+      "step": 5873
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.596057306590258e-05,
+      "loss": 0.0159,
+      "step": 5874
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.595988538681949e-05,
+      "loss": 0.0368,
+      "step": 5875
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.595919770773639e-05,
+      "loss": 0.0437,
+      "step": 5876
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.59585100286533e-05,
+      "loss": 0.0376,
+      "step": 5877
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.59578223495702e-05,
+      "loss": 0.0332,
+      "step": 5878
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.595713467048711e-05,
+      "loss": 0.0347,
+      "step": 5879
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.595644699140401e-05,
+      "loss": 0.0188,
+      "step": 5880
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.595575931232092e-05,
+      "loss": 0.022,
+      "step": 5881
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.595507163323782e-05,
+      "loss": 0.0524,
+      "step": 5882
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.595438395415473e-05,
+      "loss": 0.0464,
+      "step": 5883
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.5953696275071634e-05,
+      "loss": 0.0323,
+      "step": 5884
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.595300859598854e-05,
+      "loss": 0.0473,
+      "step": 5885
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.5952320916905445e-05,
+      "loss": 0.1459,
+      "step": 5886
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.5951633237822354e-05,
+      "loss": 0.0377,
+      "step": 5887
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.5950945558739256e-05,
+      "loss": 0.0196,
+      "step": 5888
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.5950257879656165e-05,
+      "loss": 0.0407,
+      "step": 5889
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594957020057307e-05,
+      "loss": 0.0598,
+      "step": 5890
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.5948882521489976e-05,
+      "loss": 0.0264,
+      "step": 5891
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594819484240688e-05,
+      "loss": 0.0387,
+      "step": 5892
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594750716332379e-05,
+      "loss": 0.0304,
+      "step": 5893
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594681948424069e-05,
+      "loss": 0.0454,
+      "step": 5894
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.59461318051576e-05,
+      "loss": 0.0443,
+      "step": 5895
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.59454441260745e-05,
+      "loss": 0.0313,
+      "step": 5896
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594475644699141e-05,
+      "loss": 0.0313,
+      "step": 5897
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594406876790831e-05,
+      "loss": 0.0367,
+      "step": 5898
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594338108882522e-05,
+      "loss": 0.0476,
+      "step": 5899
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594269340974212e-05,
+      "loss": 0.054,
+      "step": 5900
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594200573065903e-05,
+      "loss": 0.0189,
+      "step": 5901
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594131805157593e-05,
+      "loss": 0.042,
+      "step": 5902
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.594063037249284e-05,
+      "loss": 0.0355,
+      "step": 5903
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.5939942693409744e-05,
+      "loss": 0.054,
+      "step": 5904
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.593925501432665e-05,
+      "loss": 0.0319,
+      "step": 5905
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.5938567335243555e-05,
+      "loss": 0.0337,
+      "step": 5906
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.5937879656160464e-05,
+      "loss": 0.0558,
+      "step": 5907
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.5937191977077366e-05,
+      "loss": 0.052,
+      "step": 5908
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.5936504297994275e-05,
+      "loss": 0.0295,
+      "step": 5909
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.593581661891118e-05,
+      "loss": 0.0268,
+      "step": 5910
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.5935128939828086e-05,
+      "loss": 0.0268,
+      "step": 5911
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.593444126074499e-05,
+      "loss": 0.0453,
+      "step": 5912
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.59337535816619e-05,
+      "loss": 0.0461,
+      "step": 5913
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.59330659025788e-05,
+      "loss": 0.0438,
+      "step": 5914
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.593237822349571e-05,
+      "loss": 0.0362,
+      "step": 5915
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.593169054441261e-05,
+      "loss": 0.018,
+      "step": 5916
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.593100286532952e-05,
+      "loss": 0.0184,
+      "step": 5917
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.593031518624642e-05,
+      "loss": 0.0324,
+      "step": 5918
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.592962750716333e-05,
+      "loss": 0.055,
+      "step": 5919
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.592893982808023e-05,
+      "loss": 0.0341,
+      "step": 5920
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.592825214899714e-05,
+      "loss": 0.0343,
+      "step": 5921
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.592756446991404e-05,
+      "loss": 0.0605,
+      "step": 5922
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.592687679083095e-05,
+      "loss": 0.0499,
+      "step": 5923
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 5.5926189111747854e-05,
+      "loss": 0.0329,
+      "step": 5924
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.592550143266476e-05,
+      "loss": 0.0587,
+      "step": 5925
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.5924813753581665e-05,
+      "loss": 0.0258,
+      "step": 5926
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.5924126074498574e-05,
+      "loss": 0.0335,
+      "step": 5927
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.5923438395415476e-05,
+      "loss": 0.0669,
+      "step": 5928
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.5922750716332385e-05,
+      "loss": 0.0278,
+      "step": 5929
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.592206303724929e-05,
+      "loss": 0.0256,
+      "step": 5930
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.5921375358166196e-05,
+      "loss": 0.0188,
+      "step": 5931
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.59206876790831e-05,
+      "loss": 0.0253,
+      "step": 5932
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.592000000000001e-05,
+      "loss": 0.0565,
+      "step": 5933
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.591931232091691e-05,
+      "loss": 0.0332,
+      "step": 5934
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.591862464183382e-05,
+      "loss": 0.0559,
+      "step": 5935
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.591793696275072e-05,
+      "loss": 0.0189,
+      "step": 5936
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.591724928366763e-05,
+      "loss": 0.0324,
+      "step": 5937
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.591656160458453e-05,
+      "loss": 0.0335,
+      "step": 5938
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.591587392550144e-05,
+      "loss": 0.0354,
+      "step": 5939
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.591518624641834e-05,
+      "loss": 0.0208,
+      "step": 5940
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 5.591449856733525e-05,
+      "loss": 0.0546,
+      "step": 5941
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.591381088825215e-05,
+      "loss": 0.0202,
+      "step": 5942
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.591312320916906e-05,
+      "loss": 0.0401,
+      "step": 5943
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.5912435530085964e-05,
+      "loss": 0.0446,
+      "step": 5944
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.5911747851002866e-05,
+      "loss": 0.0416,
+      "step": 5945
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.5911060171919775e-05,
+      "loss": 0.0525,
+      "step": 5946
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.591037249283668e-05,
+      "loss": 0.0398,
+      "step": 5947
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.5909684813753586e-05,
+      "loss": 0.0525,
+      "step": 5948
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.590899713467049e-05,
+      "loss": 0.0454,
+      "step": 5949
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.59083094555874e-05,
+      "loss": 0.0639,
+      "step": 5950
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.59076217765043e-05,
+      "loss": 0.0371,
+      "step": 5951
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.590693409742121e-05,
+      "loss": 0.0357,
+      "step": 5952
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.590624641833811e-05,
+      "loss": 0.0533,
+      "step": 5953
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.590555873925502e-05,
+      "loss": 0.0638,
+      "step": 5954
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.590487106017192e-05,
+      "loss": 0.0351,
+      "step": 5955
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.590418338108883e-05,
+      "loss": 0.0518,
+      "step": 5956
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.590349570200573e-05,
+      "loss": 0.0456,
+      "step": 5957
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.590280802292264e-05,
+      "loss": 0.044,
+      "step": 5958
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 5.590212034383954e-05,
+      "loss": 0.0625,
+      "step": 5959
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.590143266475645e-05,
+      "loss": 0.043,
+      "step": 5960
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.5900744985673354e-05,
+      "loss": 0.0445,
+      "step": 5961
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.590005730659026e-05,
+      "loss": 0.0402,
+      "step": 5962
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.5899369627507165e-05,
+      "loss": 0.0238,
+      "step": 5963
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.5898681948424074e-05,
+      "loss": 0.0409,
+      "step": 5964
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.5897994269340976e-05,
+      "loss": 0.059,
+      "step": 5965
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.5897306590257885e-05,
+      "loss": 0.0179,
+      "step": 5966
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.589661891117479e-05,
+      "loss": 0.0415,
+      "step": 5967
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.5895931232091696e-05,
+      "loss": 0.0492,
+      "step": 5968
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.58952435530086e-05,
+      "loss": 0.037,
+      "step": 5969
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.589455587392551e-05,
+      "loss": 0.0221,
+      "step": 5970
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.589386819484241e-05,
+      "loss": 0.0237,
+      "step": 5971
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.589318051575932e-05,
+      "loss": 0.0511,
+      "step": 5972
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.589249283667622e-05,
+      "loss": 0.0295,
+      "step": 5973
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.589180515759313e-05,
+      "loss": 0.0444,
+      "step": 5974
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.589111747851003e-05,
+      "loss": 0.0327,
+      "step": 5975
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 5.589042979942694e-05,
+      "loss": 0.0408,
+      "step": 5976
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.588974212034384e-05,
+      "loss": 0.0338,
+      "step": 5977
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.5889054441260744e-05,
+      "loss": 0.0504,
+      "step": 5978
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.588836676217765e-05,
+      "loss": 0.0512,
+      "step": 5979
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.5887679083094555e-05,
+      "loss": 0.0433,
+      "step": 5980
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.588699140401146e-05,
+      "loss": 0.0476,
+      "step": 5981
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.5886303724928366e-05,
+      "loss": 0.0293,
+      "step": 5982
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.588561604584527e-05,
+      "loss": 0.0265,
+      "step": 5983
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.588492836676218e-05,
+      "loss": 0.0326,
+      "step": 5984
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.588424068767908e-05,
+      "loss": 0.0275,
+      "step": 5985
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.588355300859599e-05,
+      "loss": 0.0299,
+      "step": 5986
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.588286532951289e-05,
+      "loss": 0.0343,
+      "step": 5987
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.58821776504298e-05,
+      "loss": 0.0408,
+      "step": 5988
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.58814899713467e-05,
+      "loss": 0.0239,
+      "step": 5989
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.588080229226361e-05,
+      "loss": 0.031,
+      "step": 5990
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.588011461318051e-05,
+      "loss": 0.0517,
+      "step": 5991
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.587942693409742e-05,
+      "loss": 0.0538,
+      "step": 5992
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.587873925501432e-05,
+      "loss": 0.0304,
+      "step": 5993
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 5.587805157593123e-05,
+      "loss": 0.0374,
+      "step": 5994
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.5877363896848134e-05,
+      "loss": 0.0687,
+      "step": 5995
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.587667621776504e-05,
+      "loss": 0.0203,
+      "step": 5996
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.5875988538681945e-05,
+      "loss": 0.045,
+      "step": 5997
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.5875300859598854e-05,
+      "loss": 0.0314,
+      "step": 5998
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.5874613180515756e-05,
+      "loss": 0.0301,
+      "step": 5999
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 5.5873925501432665e-05,
+      "loss": 0.0322,
+      "step": 6000
+    },
+    {
+      "epoch": 3.44,
+      "eval_loss": 0.04754304513335228,
+      "eval_mean_accuracy": 0.8718316352110502,
+      "eval_mean_iou": 0.7735643375797914,
+      "eval_overall_accuracy": 0.985195844438341,
+      "eval_per_category_accuracy": [
+        0.9947443950950807,
+        0.8724620187459422,
+        0.9168990202071478,
+        0.7032211067960306
+      ],
+      "eval_per_category_iou": [
+        0.9856305598923848,
+        0.7027964682093609,
+        0.7920541219748636,
+        0.6137762002425566
+      ],
+      "eval_runtime": 244.4892,
+      "eval_samples_per_second": 14.725,
+      "eval_steps_per_second": 1.841,
+      "step": 6000
+    }
+  ],
+  "max_steps": 87250,
+  "num_train_epochs": 50,
+  "total_flos": 1.4493725389121126e+19,
+  "trial_name": null,
+  "trial_params": null
+}