diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,48085 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9604706306089984,
+  "eval_steps": 1000,
+  "global_step": 8000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.99999982216222e-05,
+      "loss": 5.46,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999992886489043e-05,
+      "loss": 7.1523,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999983994601295e-05,
+      "loss": 5.3145,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999997154596021e-05,
+      "loss": 5.5693,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999995554056757e-05,
+      "loss": 5.5674,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999993597842566e-05,
+      "loss": 5.5332,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999912859537236e-05,
+      "loss": 4.8945,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999988618390561e-05,
+      "loss": 4.8462,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999855951534566e-05,
+      "loss": 5.1289,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999822162428416e-05,
+      "loss": 4.9102,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999784816591964e-05,
+      "loss": 4.48,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999974391403051e-05,
+      "loss": 4.0273,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999969945474989e-05,
+      "loss": 4.5864,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999965143875642e-05,
+      "loss": 3.6118,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999599866056927e-05,
+      "loss": 4.1108,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999954473665876e-05,
+      "loss": 3.8984,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999948605056976e-05,
+      "loss": 3.4907,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999942380779828e-05,
+      "loss": 3.6602,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999358008353154e-05,
+      "loss": 3.6274,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999288652243757e-05,
+      "loss": 2.9629,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999215739479965e-05,
+      "loss": 2.8955,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999139270072146e-05,
+      "loss": 2.8516,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9999059244031176e-05,
+      "loss": 2.6045,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999897566136843e-05,
+      "loss": 2.3472,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999888852209582e-05,
+      "loss": 2.1572,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998797826225744e-05,
+      "loss": 2.0295,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999870357377108e-05,
+      "loss": 1.8174,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998605764745255e-05,
+      "loss": 1.9019,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998504399162186e-05,
+      "loss": 1.6545,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999839947703629e-05,
+      "loss": 1.5762,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999829099838249e-05,
+      "loss": 1.3013,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998178963216227e-05,
+      "loss": 1.2012,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9998063371553425e-05,
+      "loss": 1.1617,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997944223410556e-05,
+      "loss": 1.1313,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999782151880456e-05,
+      "loss": 1.0817,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997695257752875e-05,
+      "loss": 1.2112,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999756544027349e-05,
+      "loss": 0.9791,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999743206638486e-05,
+      "loss": 0.8965,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999729513610597e-05,
+      "loss": 0.7751,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.999715464945629e-05,
+      "loss": 0.9619,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9997010606455814e-05,
+      "loss": 0.8671,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996863007125025e-05,
+      "loss": 0.8119,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996711851484944e-05,
+      "loss": 0.8314,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999655713955705e-05,
+      "loss": 0.8772,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999639887136338e-05,
+      "loss": 0.8112,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996237046926434e-05,
+      "loss": 0.9047,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9996071666269226e-05,
+      "loss": 1.0377,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995902729415306e-05,
+      "loss": 0.7456,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995730236388696e-05,
+      "loss": 0.8188,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995554187213946e-05,
+      "loss": 1.0289,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999537458191609e-05,
+      "loss": 0.8851,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999519142052069e-05,
+      "loss": 0.9148,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9995004703053795e-05,
+      "loss": 0.8207,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999481442954198e-05,
+      "loss": 0.7682,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994620600012314e-05,
+      "loss": 0.894,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999442321449237e-05,
+      "loss": 0.7436,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999422227301023e-05,
+      "loss": 0.8976,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9994017775594476e-05,
+      "loss": 1.0,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999380972227421e-05,
+      "loss": 0.8563,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999359811307903e-05,
+      "loss": 0.6584,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999338294803905e-05,
+      "loss": 0.9342,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9993164227184864e-05,
+      "loss": 0.8384,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99929419505476e-05,
+      "loss": 0.8362,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999271611815888e-05,
+      "loss": 1.0142,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992486730050834e-05,
+      "loss": 0.8809,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99922537862561e-05,
+      "loss": 0.7258,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9992017286807814e-05,
+      "loss": 0.8152,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999177723173962e-05,
+      "loss": 0.5397,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9991533621085676e-05,
+      "loss": 0.6564,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999128645488064e-05,
+      "loss": 0.6403,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999103573315967e-05,
+      "loss": 0.6254,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.999078145595845e-05,
+      "loss": 0.6275,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990523623313146e-05,
+      "loss": 0.8439,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9990262235260444e-05,
+      "loss": 0.8895,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9989997291837524e-05,
+      "loss": 0.7084,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998972879308209e-05,
+      "loss": 0.8926,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998945673903233e-05,
+      "loss": 0.545,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998918112972696e-05,
+      "loss": 0.7977,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988901965205185e-05,
+      "loss": 0.8218,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998861924550673e-05,
+      "loss": 0.8107,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988332970671803e-05,
+      "loss": 0.6561,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9988043140741145e-05,
+      "loss": 0.8785,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998774975575599e-05,
+      "loss": 0.8685,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998745281575806e-05,
+      "loss": 0.8899,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998715232078963e-05,
+      "loss": 0.45,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998684827089343e-05,
+      "loss": 0.65,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986540666112716e-05,
+      "loss": 0.6326,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9986229506491264e-05,
+      "loss": 0.8008,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998591479207334e-05,
+      "loss": 0.826,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998559652290371e-05,
+      "loss": 0.6747,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998527469902766e-05,
+      "loss": 0.5486,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998494932049098e-05,
+      "loss": 0.8922,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998462038733995e-05,
+      "loss": 0.8973,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998428789962139e-05,
+      "loss": 0.7823,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998395185738257e-05,
+      "loss": 0.6039,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998361226067132e-05,
+      "loss": 0.8076,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998326910953596e-05,
+      "loss": 0.6896,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998292240402529e-05,
+      "loss": 0.5089,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998257214418866e-05,
+      "loss": 0.678,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998221833007588e-05,
+      "loss": 0.5447,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9981860961737296e-05,
+      "loss": 0.9568,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998150003922375e-05,
+      "loss": 0.6177,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998113556258659e-05,
+      "loss": 0.6303,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998076753187767e-05,
+      "loss": 0.7928,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998039594714936e-05,
+      "loss": 0.5879,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.998002080845451e-05,
+      "loss": 0.8993,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99796421158465e-05,
+      "loss": 0.6246,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.99792598693792e-05,
+      "loss": 0.8381,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9978874069107005e-05,
+      "loss": 0.5032,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997848471508479e-05,
+      "loss": 0.6895,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997809180736796e-05,
+      "loss": 0.5519,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997769534601239e-05,
+      "loss": 0.7189,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9977295331074516e-05,
+      "loss": 0.8541,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976891762611235e-05,
+      "loss": 0.6348,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976484640679954e-05,
+      "loss": 0.7569,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9976073965338605e-05,
+      "loss": 0.6807,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9975659736645606e-05,
+      "loss": 0.6747,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997524195465991e-05,
+      "loss": 0.9092,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997482061944092e-05,
+      "loss": 0.7045,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997439573104862e-05,
+      "loss": 0.5674,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997396728954343e-05,
+      "loss": 0.8207,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997353529498631e-05,
+      "loss": 0.875,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997309974743873e-05,
+      "loss": 0.6039,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.997266064696264e-05,
+      "loss": 0.6326,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.997221799362053e-05,
+      "loss": 0.6976,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9971771787475366e-05,
+      "loss": 0.5513,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9971322028590615e-05,
+      "loss": 0.8132,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9970868717030295e-05,
+      "loss": 0.7363,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9970411852858876e-05,
+      "loss": 0.7117,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996995143614136e-05,
+      "loss": 0.8324,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996948746694326e-05,
+      "loss": 0.6816,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996901994533057e-05,
+      "loss": 0.5261,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996854887136982e-05,
+      "loss": 0.7924,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9968074245128024e-05,
+      "loss": 0.5662,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99675960666727e-05,
+      "loss": 1.023,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996711433607189e-05,
+      "loss": 0.6558,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996662905339412e-05,
+      "loss": 0.6,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996614021870844e-05,
+      "loss": 0.8712,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996564783208439e-05,
+      "loss": 0.8904,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9965151893592024e-05,
+      "loss": 0.6578,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9964652403301905e-05,
+      "loss": 0.5598,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996414936128508e-05,
+      "loss": 0.5261,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996364276761313e-05,
+      "loss": 0.7261,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996313262235813e-05,
+      "loss": 0.6993,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9962618925592655e-05,
+      "loss": 1.0376,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996210167738978e-05,
+      "loss": 0.6467,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996158087782311e-05,
+      "loss": 0.8088,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996105652696672e-05,
+      "loss": 0.8817,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.996052862489522e-05,
+      "loss": 0.6288,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995999717168373e-05,
+      "loss": 0.9591,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995946216740783e-05,
+      "loss": 0.863,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995892361214366e-05,
+      "loss": 0.7505,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995838150596783e-05,
+      "loss": 0.7709,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9957835848957454e-05,
+      "loss": 0.6696,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995728664119019e-05,
+      "loss": 0.8364,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9956733882744154e-05,
+      "loss": 0.9604,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9956177573697995e-05,
+      "loss": 0.7547,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995561771413085e-05,
+      "loss": 0.8249,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9955054304122375e-05,
+      "loss": 0.8757,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995448734375273e-05,
+      "loss": 0.5397,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9953916833102576e-05,
+      "loss": 0.5649,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9953342772253074e-05,
+      "loss": 0.7036,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9952765161285906e-05,
+      "loss": 0.7315,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995218400028324e-05,
+      "loss": 0.9,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995159928932777e-05,
+      "loss": 0.6579,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995101102850266e-05,
+      "loss": 0.6979,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.995041921789162e-05,
+      "loss": 0.7496,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9949823857578836e-05,
+      "loss": 0.606,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994922494764903e-05,
+      "loss": 0.5994,
+      "step": 169
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994862248818738e-05,
+      "loss": 0.9365,
+      "step": 170
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994801647927963e-05,
+      "loss": 0.3735,
+      "step": 171
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994740692101198e-05,
+      "loss": 0.4648,
+      "step": 172
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994679381347114e-05,
+      "loss": 0.5766,
+      "step": 173
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994617715674436e-05,
+      "loss": 0.7181,
+      "step": 174
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9945556950919354e-05,
+      "loss": 0.707,
+      "step": 175
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994493319608437e-05,
+      "loss": 0.5309,
+      "step": 176
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9944305892328145e-05,
+      "loss": 0.5834,
+      "step": 177
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994367503973993e-05,
+      "loss": 0.6697,
+      "step": 178
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994304063840947e-05,
+      "loss": 0.707,
+      "step": 179
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994240268842703e-05,
+      "loss": 0.5599,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9941761189883364e-05,
+      "loss": 0.6802,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994111614286974e-05,
+      "loss": 0.7852,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.994046754747792e-05,
+      "loss": 1.0704,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99398154038002e-05,
+      "loss": 0.5792,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993915971192935e-05,
+      "loss": 0.7611,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993850047195865e-05,
+      "loss": 0.4584,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99378376839819e-05,
+      "loss": 0.7277,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993717134809338e-05,
+      "loss": 0.6825,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993650146438791e-05,
+      "loss": 0.4676,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993582803296078e-05,
+      "loss": 0.6905,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9935151053907815e-05,
+      "loss": 0.7367,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993447052732531e-05,
+      "loss": 0.5679,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9933786453310086e-05,
+      "loss": 0.7251,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9933098831959473e-05,
+      "loss": 0.5801,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99324076633713e-05,
+      "loss": 0.8254,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.99317129476439e-05,
+      "loss": 0.8307,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9931014684876104e-05,
+      "loss": 0.5263,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.993031287516726e-05,
+      "loss": 0.6677,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9929607518617216e-05,
+      "loss": 0.6378,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9928898615326316e-05,
+      "loss": 0.7197,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992818616539542e-05,
+      "loss": 0.4792,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9927470168925885e-05,
+      "loss": 0.9016,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992675062601958e-05,
+      "loss": 0.8376,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9926027536778877e-05,
+      "loss": 0.639,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9925300901306646e-05,
+      "loss": 0.7834,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992457071970627e-05,
+      "loss": 0.6763,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.992383699208162e-05,
+      "loss": 0.8359,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9923099718537094e-05,
+      "loss": 0.7219,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9922358899177576e-05,
+      "loss": 0.5292,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992161453410848e-05,
+      "loss": 0.4614,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992086662343569e-05,
+      "loss": 0.7055,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.992011516726562e-05,
+      "loss": 0.6606,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9919360165705176e-05,
+      "loss": 0.588,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991860161886177e-05,
+      "loss": 0.7551,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9917839526843324e-05,
+      "loss": 0.8033,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991707388975826e-05,
+      "loss": 0.727,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991630470771551e-05,
+      "loss": 0.7256,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99155319808245e-05,
+      "loss": 0.5542,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991475570919516e-05,
+      "loss": 0.7157,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9913975892937954e-05,
+      "loss": 0.8693,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.99131925321638e-05,
+      "loss": 0.4419,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991240562698416e-05,
+      "loss": 0.5813,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9911615177510986e-05,
+      "loss": 0.8186,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991082118385673e-05,
+      "loss": 0.5242,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.991002364613436e-05,
+      "loss": 0.6923,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9909222564457335e-05,
+      "loss": 0.6424,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990841793893964e-05,
+      "loss": 0.8932,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990760976969573e-05,
+      "loss": 0.6252,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9906798056840596e-05,
+      "loss": 0.6616,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990598280048972e-05,
+      "loss": 0.6171,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990516400075908e-05,
+      "loss": 0.6716,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990434165776517e-05,
+      "loss": 0.5376,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9903515771625e-05,
+      "loss": 0.8225,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.990268634245605e-05,
+      "loss": 0.6288,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9901853370376326e-05,
+      "loss": 0.7289,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9901016855504347e-05,
+      "loss": 0.729,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9900176797959106e-05,
+      "loss": 0.6198,
+      "step": 237
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989933319786013e-05,
+      "loss": 0.6304,
+      "step": 238
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9898486055327445e-05,
+      "loss": 0.5477,
+      "step": 239
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9897635370481556e-05,
+      "loss": 0.9713,
+      "step": 240
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98967811434435e-05,
+      "loss": 0.7369,
+      "step": 241
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989592337433481e-05,
+      "loss": 0.775,
+      "step": 242
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9895062063277514e-05,
+      "loss": 0.6835,
+      "step": 243
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989419721039416e-05,
+      "loss": 0.6682,
+      "step": 244
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9893328815807785e-05,
+      "loss": 0.5733,
+      "step": 245
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989245687964194e-05,
+      "loss": 0.6537,
+      "step": 246
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.989158140202067e-05,
+      "loss": 0.6251,
+      "step": 247
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9890702383068535e-05,
+      "loss": 0.4703,
+      "step": 248
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9889819822910584e-05,
+      "loss": 0.5748,
+      "step": 249
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988893372167239e-05,
+      "loss": 0.9214,
+      "step": 250
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988804407948001e-05,
+      "loss": 0.5891,
+      "step": 251
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988715089646002e-05,
+      "loss": 1.0165,
+      "step": 252
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9886254172739486e-05,
+      "loss": 0.5927,
+      "step": 253
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9885353908446e-05,
+      "loss": 0.554,
+      "step": 254
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988445010370763e-05,
+      "loss": 0.679,
+      "step": 255
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988354275865296e-05,
+      "loss": 0.5224,
+      "step": 256
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988263187341108e-05,
+      "loss": 0.6447,
+      "step": 257
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988171744811159e-05,
+      "loss": 0.7822,
+      "step": 258
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.988079948288457e-05,
+      "loss": 0.6774,
+      "step": 259
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9879877977860635e-05,
+      "loss": 0.5884,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9878952933170875e-05,
+      "loss": 0.6078,
+      "step": 261
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98780243489469e-05,
+      "loss": 0.652,
+      "step": 262
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987709222532083e-05,
+      "loss": 0.8943,
+      "step": 263
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987615656242527e-05,
+      "loss": 0.7563,
+      "step": 264
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9875217360393335e-05,
+      "loss": 0.64,
+      "step": 265
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987427461935865e-05,
+      "loss": 0.6506,
+      "step": 266
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987332833945533e-05,
+      "loss": 0.6356,
+      "step": 267
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9872378520818015e-05,
+      "loss": 0.6131,
+      "step": 268
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987142516358183e-05,
+      "loss": 1.0162,
+      "step": 269
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.987046826788241e-05,
+      "loss": 0.6153,
+      "step": 270
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9869507833855886e-05,
+      "loss": 0.8401,
+      "step": 271
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986854386163891e-05,
+      "loss": 1.0875,
+      "step": 272
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9867576351368624e-05,
+      "loss": 1.0283,
+      "step": 273
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9866605303182665e-05,
+      "loss": 0.7184,
+      "step": 274
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98656307172192e-05,
+      "loss": 0.7462,
+      "step": 275
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986465259361688e-05,
+      "loss": 0.6588,
+      "step": 276
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9863670932514855e-05,
+      "loss": 0.6968,
+      "step": 277
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986268573405279e-05,
+      "loss": 0.7891,
+      "step": 278
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986169699837085e-05,
+      "loss": 0.5815,
+      "step": 279
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.986070472560971e-05,
+      "loss": 0.9948,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9859708915910524e-05,
+      "loss": 0.8211,
+      "step": 281
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985870956941498e-05,
+      "loss": 0.8236,
+      "step": 282
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9857706686265245e-05,
+      "loss": 0.8545,
+      "step": 283
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985670026660402e-05,
+      "loss": 0.8745,
+      "step": 284
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985569031057445e-05,
+      "loss": 0.7574,
+      "step": 285
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985467681832027e-05,
+      "loss": 0.6704,
+      "step": 286
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985365978998563e-05,
+      "loss": 0.79,
+      "step": 287
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.985263922571525e-05,
+      "loss": 0.7021,
+      "step": 288
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9851615125654305e-05,
+      "loss": 0.761,
+      "step": 289
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.98505874899485e-05,
+      "loss": 0.5789,
+      "step": 290
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984955631874405e-05,
+      "loss": 0.5641,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984852161218764e-05,
+      "loss": 0.637,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9847483370426495e-05,
+      "loss": 0.5307,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9846441593608316e-05,
+      "loss": 0.7209,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984539628188132e-05,
+      "loss": 0.5544,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984434743539422e-05,
+      "loss": 0.5932,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9843295054296246e-05,
+      "loss": 0.652,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9842239138737114e-05,
+      "loss": 0.683,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984117968886704e-05,
+      "loss": 0.6388,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.984011670483677e-05,
+      "loss": 0.7194,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983905018679752e-05,
+      "loss": 0.6823,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983798013490103e-05,
+      "loss": 0.6699,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983690654929954e-05,
+      "loss": 0.6157,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983582943014578e-05,
+      "loss": 0.8528,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9834748777593e-05,
+      "loss": 0.5225,
+      "step": 305
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983366459179494e-05,
+      "loss": 0.6972,
+      "step": 306
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9832576872905856e-05,
+      "loss": 0.689,
+      "step": 307
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.983148562108049e-05,
+      "loss": 0.6334,
+      "step": 308
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9830390836474095e-05,
+      "loss": 0.7823,
+      "step": 309
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9829292519242424e-05,
+      "loss": 0.5346,
+      "step": 310
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982819066954174e-05,
+      "loss": 0.6865,
+      "step": 311
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98270852875288e-05,
+      "loss": 0.6978,
+      "step": 312
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982597637336087e-05,
+      "loss": 0.4336,
+      "step": 313
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9824863927195715e-05,
+      "loss": 0.5337,
+      "step": 314
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98237479491916e-05,
+      "loss": 0.5604,
+      "step": 315
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.98226284395073e-05,
+      "loss": 0.5585,
+      "step": 316
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982150539830208e-05,
+      "loss": 0.9023,
+      "step": 317
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.982037882573573e-05,
+      "loss": 0.7083,
+      "step": 318
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981924872196851e-05,
+      "loss": 0.7563,
+      "step": 319
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981811508716121e-05,
+      "loss": 0.7632,
+      "step": 320
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9816977921475113e-05,
+      "loss": 0.6455,
+      "step": 321
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9815837225072e-05,
+      "loss": 0.7181,
+      "step": 322
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981469299811416e-05,
+      "loss": 0.6406,
+      "step": 323
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981354524076439e-05,
+      "loss": 0.5086,
+      "step": 324
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.981239395318597e-05,
+      "loss": 0.657,
+      "step": 325
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9811239135542695e-05,
+      "loss": 0.7577,
+      "step": 326
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9810080787998855e-05,
+      "loss": 0.7915,
+      "step": 327
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9808918910719274e-05,
+      "loss": 0.5222,
+      "step": 328
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980775350386922e-05,
+      "loss": 0.5697,
+      "step": 329
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980658456761453e-05,
+      "loss": 0.8407,
+      "step": 330
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9805412102121486e-05,
+      "loss": 0.7925,
+      "step": 331
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980423610755689e-05,
+      "loss": 0.4104,
+      "step": 332
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980305658408807e-05,
+      "loss": 0.9498,
+      "step": 333
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.980187353188282e-05,
+      "loss": 0.7555,
+      "step": 334
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9800686951109465e-05,
+      "loss": 0.8007,
+      "step": 335
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9799496841936816e-05,
+      "loss": 0.6558,
+      "step": 336
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979830320453419e-05,
+      "loss": 0.5542,
+      "step": 337
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979710603907141e-05,
+      "loss": 0.799,
+      "step": 338
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979590534571878e-05,
+      "loss": 0.6392,
+      "step": 339
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979470112464715e-05,
+      "loss": 0.8561,
+      "step": 340
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979349337602782e-05,
+      "loss": 0.6457,
+      "step": 341
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979228210003263e-05,
+      "loss": 0.7187,
+      "step": 342
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97910672968339e-05,
+      "loss": 0.5437,
+      "step": 343
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978984896660448e-05,
+      "loss": 0.5943,
+      "step": 344
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978862710951768e-05,
+      "loss": 0.766,
+      "step": 345
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978740172574735e-05,
+      "loss": 0.6271,
+      "step": 346
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9786172815467815e-05,
+      "loss": 0.603,
+      "step": 347
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9784940378853905e-05,
+      "loss": 0.7604,
+      "step": 348
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978370441608098e-05,
+      "loss": 0.5464,
+      "step": 349
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978246492732487e-05,
+      "loss": 0.6904,
+      "step": 350
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.978122191276191e-05,
+      "loss": 0.6101,
+      "step": 351
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977997537256895e-05,
+      "loss": 0.4756,
+      "step": 352
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9778725306923344e-05,
+      "loss": 0.7855,
+      "step": 353
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977747171600293e-05,
+      "loss": 0.6639,
+      "step": 354
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977621459998606e-05,
+      "loss": 0.7948,
+      "step": 355
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977495395905158e-05,
+      "loss": 0.5737,
+      "step": 356
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977368979337884e-05,
+      "loss": 0.7281,
+      "step": 357
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9772422103147706e-05,
+      "loss": 0.7661,
+      "step": 358
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.977115088853852e-05,
+      "loss": 0.707,
+      "step": 359
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976987614973214e-05,
+      "loss": 0.4374,
+      "step": 360
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976859788690993e-05,
+      "loss": 0.5703,
+      "step": 361
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976731610025374e-05,
+      "loss": 0.7291,
+      "step": 362
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976603078994594e-05,
+      "loss": 0.6506,
+      "step": 363
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976474195616938e-05,
+      "loss": 0.8446,
+      "step": 364
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.976344959910743e-05,
+      "loss": 0.4817,
+      "step": 365
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9762153718943955e-05,
+      "loss": 0.8449,
+      "step": 366
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9760854315863317e-05,
+      "loss": 0.699,
+      "step": 367
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9759551390050374e-05,
+      "loss": 0.5606,
+      "step": 368
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975824494169051e-05,
+      "loss": 0.6755,
+      "step": 369
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9756934970969586e-05,
+      "loss": 0.5291,
+      "step": 370
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9755621478073973e-05,
+      "loss": 0.5761,
+      "step": 371
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975430446319054e-05,
+      "loss": 0.5203,
+      "step": 372
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.975298392650666e-05,
+      "loss": 0.6236,
+      "step": 373
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.97516598682102e-05,
+      "loss": 0.362,
+      "step": 374
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.975033228848954e-05,
+      "loss": 0.4697,
+      "step": 375
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974900118753356e-05,
+      "loss": 0.6818,
+      "step": 376
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9747666565531635e-05,
+      "loss": 0.6012,
+      "step": 377
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974632842267363e-05,
+      "loss": 0.9537,
+      "step": 378
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974498675914994e-05,
+      "loss": 0.6309,
+      "step": 379
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9743641575151426e-05,
+      "loss": 0.7528,
+      "step": 380
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9742292870869475e-05,
+      "loss": 0.6019,
+      "step": 381
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.974094064649597e-05,
+      "loss": 0.6727,
+      "step": 382
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9739584902223293e-05,
+      "loss": 0.6952,
+      "step": 383
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9738225638244326e-05,
+      "loss": 0.6967,
+      "step": 384
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973686285475245e-05,
+      "loss": 0.7175,
+      "step": 385
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973549655194154e-05,
+      "loss": 0.6754,
+      "step": 386
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9734126730006e-05,
+      "loss": 0.4955,
+      "step": 387
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9732753389140696e-05,
+      "loss": 0.6056,
+      "step": 388
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973137652954103e-05,
+      "loss": 0.7049,
+      "step": 389
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9729996151402866e-05,
+      "loss": 0.616,
+      "step": 390
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972861225492261e-05,
+      "loss": 0.5414,
+      "step": 391
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972722484029715e-05,
+      "loss": 0.6295,
+      "step": 392
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9725833907723855e-05,
+      "loss": 0.4691,
+      "step": 393
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972443945740063e-05,
+      "loss": 0.85,
+      "step": 394
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9723041489525857e-05,
+      "loss": 0.7158,
+      "step": 395
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972164000429843e-05,
+      "loss": 0.7802,
+      "step": 396
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.972023500191774e-05,
+      "loss": 0.7663,
+      "step": 397
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9718826482583666e-05,
+      "loss": 0.4505,
+      "step": 398
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9717414446496604e-05,
+      "loss": 0.6,
+      "step": 399
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971599889385745e-05,
+      "loss": 0.6686,
+      "step": 400
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971457982486759e-05,
+      "loss": 0.8933,
+      "step": 401
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971315723972891e-05,
+      "loss": 0.5766,
+      "step": 402
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9711731138643815e-05,
+      "loss": 0.656,
+      "step": 403
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.971030152181518e-05,
+      "loss": 0.7134,
+      "step": 404
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970886838944642e-05,
+      "loss": 0.5447,
+      "step": 405
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97074317417414e-05,
+      "loss": 0.4781,
+      "step": 406
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.970599157890453e-05,
+      "loss": 0.6373,
+      "step": 407
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9704547901140694e-05,
+      "loss": 0.5851,
+      "step": 408
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.97031007086553e-05,
+      "loss": 0.7894,
+      "step": 409
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9701650001654224e-05,
+      "loss": 0.6064,
+      "step": 410
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9700195780343857e-05,
+      "loss": 0.7509,
+      "step": 411
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9698738044931096e-05,
+      "loss": 0.6545,
+      "step": 412
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9697276795623337e-05,
+      "loss": 0.7742,
+      "step": 413
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9695812032628474e-05,
+      "loss": 0.5869,
+      "step": 414
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969434375615489e-05,
+      "loss": 0.5945,
+      "step": 415
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969287196641149e-05,
+      "loss": 0.7372,
+      "step": 416
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.969139666360765e-05,
+      "loss": 0.6243,
+      "step": 417
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968991784795327e-05,
+      "loss": 0.4756,
+      "step": 418
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9688435519658744e-05,
+      "loss": 0.591,
+      "step": 419
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968694967893496e-05,
+      "loss": 0.5408,
+      "step": 420
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96854603259933e-05,
+      "loss": 0.8085,
+      "step": 421
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9683967461045675e-05,
+      "loss": 0.6293,
+      "step": 422
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968247108430446e-05,
+      "loss": 0.651,
+      "step": 423
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.968097119598255e-05,
+      "loss": 0.7825,
+      "step": 424
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967946779629333e-05,
+      "loss": 0.6425,
+      "step": 425
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9677960885450694e-05,
+      "loss": 0.7379,
+      "step": 426
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9676450463669035e-05,
+      "loss": 0.8749,
+      "step": 427
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967493653116323e-05,
+      "loss": 0.6492,
+      "step": 428
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967341908814866e-05,
+      "loss": 0.916,
+      "step": 429
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967189813484124e-05,
+      "loss": 0.4033,
+      "step": 430
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.967037367145733e-05,
+      "loss": 0.5085,
+      "step": 431
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966884569821383e-05,
+      "loss": 0.7063,
+      "step": 432
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966731421532812e-05,
+      "loss": 0.6682,
+      "step": 433
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966577922301808e-05,
+      "loss": 0.7107,
+      "step": 434
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.96642407215021e-05,
+      "loss": 0.7999,
+      "step": 435
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9662698710999066e-05,
+      "loss": 0.7546,
+      "step": 436
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.966115319172836e-05,
+      "loss": 0.6126,
+      "step": 437
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9659604163909846e-05,
+      "loss": 0.4661,
+      "step": 438
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9658051627763924e-05,
+      "loss": 0.6428,
+      "step": 439
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9656495583511464e-05,
+      "loss": 0.7189,
+      "step": 440
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965493603137385e-05,
+      "loss": 0.7906,
+      "step": 441
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9653372971572965e-05,
+      "loss": 0.6865,
+      "step": 442
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965180640433117e-05,
+      "loss": 0.6224,
+      "step": 443
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.965023632987135e-05,
+      "loss": 0.6097,
+      "step": 444
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9648662748416886e-05,
+      "loss": 0.6149,
+      "step": 445
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.964708566019164e-05,
+      "loss": 0.6367,
+      "step": 446
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9645505065419985e-05,
+      "loss": 0.4513,
+      "step": 447
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9643920964326806e-05,
+      "loss": 0.9803,
+      "step": 448
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9642333357137464e-05,
+      "loss": 0.5621,
+      "step": 449
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9640742244077825e-05,
+      "loss": 0.3833,
+      "step": 450
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963914762537426e-05,
+      "loss": 0.8129,
+      "step": 451
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963754950125364e-05,
+      "loss": 0.5831,
+      "step": 452
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963594787194333e-05,
+      "loss": 0.6182,
+      "step": 453
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963434273767119e-05,
+      "loss": 0.7505,
+      "step": 454
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9632734098665586e-05,
+      "loss": 0.6577,
+      "step": 455
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.963112195515537e-05,
+      "loss": 0.6859,
+      "step": 456
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.962950630736992e-05,
+      "loss": 0.6545,
+      "step": 457
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.962788715553908e-05,
+      "loss": 0.5172,
+      "step": 458
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.96262644998932e-05,
+      "loss": 0.6263,
+      "step": 459
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.962463834066316e-05,
+      "loss": 0.8149,
+      "step": 460
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9623008678080304e-05,
+      "loss": 0.6213,
+      "step": 461
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9621375512376475e-05,
+      "loss": 0.8157,
+      "step": 462
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961973884378403e-05,
+      "loss": 0.6943,
+      "step": 463
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961809867253583e-05,
+      "loss": 0.7336,
+      "step": 464
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.96164549988652e-05,
+      "loss": 0.7457,
+      "step": 465
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9614807823005994e-05,
+      "loss": 0.807,
+      "step": 466
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961315714519257e-05,
+      "loss": 0.713,
+      "step": 467
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.961150296565975e-05,
+      "loss": 0.6162,
+      "step": 468
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9609845284642885e-05,
+      "loss": 0.6956,
+      "step": 469
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9608184102377816e-05,
+      "loss": 0.5938,
+      "step": 470
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960651941910087e-05,
+      "loss": 0.5889,
+      "step": 471
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.96048512350489e-05,
+      "loss": 0.7258,
+      "step": 472
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9603179550459223e-05,
+      "loss": 0.6503,
+      "step": 473
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.960150436556967e-05,
+      "loss": 0.7913,
+      "step": 474
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9599825680618574e-05,
+      "loss": 0.4902,
+      "step": 475
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959814349584476e-05,
+      "loss": 0.4564,
+      "step": 476
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9596457811487556e-05,
+      "loss": 0.4055,
+      "step": 477
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9594768627786784e-05,
+      "loss": 1.0184,
+      "step": 478
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.959307594498277e-05,
+      "loss": 0.7488,
+      "step": 479
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9591379763316314e-05,
+      "loss": 0.6749,
+      "step": 480
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9589680083028754e-05,
+      "loss": 0.6179,
+      "step": 481
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9587976904361893e-05,
+      "loss": 0.7342,
+      "step": 482
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958627022755804e-05,
+      "loss": 0.5132,
+      "step": 483
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9584560052860016e-05,
+      "loss": 0.6995,
+      "step": 484
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958284638051111e-05,
+      "loss": 0.7373,
+      "step": 485
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.958112921075514e-05,
+      "loss": 0.6716,
+      "step": 486
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95794085438364e-05,
+      "loss": 0.5728,
+      "step": 487
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957768437999971e-05,
+      "loss": 0.7855,
+      "step": 488
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957595671949034e-05,
+      "loss": 0.6957,
+      "step": 489
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95742255625541e-05,
+      "loss": 0.819,
+      "step": 490
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957249090943727e-05,
+      "loss": 0.8107,
+      "step": 491
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.957075276038666e-05,
+      "loss": 0.8528,
+      "step": 492
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.956901111564954e-05,
+      "loss": 0.6467,
+      "step": 493
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9567265975473696e-05,
+      "loss": 0.5813,
+      "step": 494
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9565517340107416e-05,
+      "loss": 0.7831,
+      "step": 495
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9563765209799475e-05,
+      "loss": 0.64,
+      "step": 496
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.956200958479914e-05,
+      "loss": 0.5341,
+      "step": 497
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9560250465356205e-05,
+      "loss": 0.6768,
+      "step": 498
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955848785172092e-05,
+      "loss": 0.5309,
+      "step": 499
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955672174414407e-05,
+      "loss": 0.4952,
+      "step": 500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9554952142876905e-05,
+      "loss": 0.3887,
+      "step": 501
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955317904817119e-05,
+      "loss": 0.6719,
+      "step": 502
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.955140246027919e-05,
+      "loss": 0.6742,
+      "step": 503
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9549622379453655e-05,
+      "loss": 0.9116,
+      "step": 504
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9547838805947835e-05,
+      "loss": 0.6121,
+      "step": 505
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.954605174001549e-05,
+      "loss": 0.7318,
+      "step": 506
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9544261181910855e-05,
+      "loss": 0.7917,
+      "step": 507
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.954246713188868e-05,
+      "loss": 0.6719,
+      "step": 508
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95406695902042e-05,
+      "loss": 0.6075,
+      "step": 509
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.953886855711315e-05,
+      "loss": 0.5865,
+      "step": 510
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9537064032871775e-05,
+      "loss": 0.4234,
+      "step": 511
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9535256017736795e-05,
+      "loss": 0.6541,
+      "step": 512
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9533444511965446e-05,
+      "loss": 0.7175,
+      "step": 513
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9531629515815436e-05,
+      "loss": 0.6281,
+      "step": 514
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9529811029544995e-05,
+      "loss": 0.6893,
+      "step": 515
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9527989053412835e-05,
+      "loss": 0.782,
+      "step": 516
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.952616358767817e-05,
+      "loss": 0.7469,
+      "step": 517
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9524334632600725e-05,
+      "loss": 0.875,
+      "step": 518
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.952250218844068e-05,
+      "loss": 0.8128,
+      "step": 519
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9520666255458756e-05,
+      "loss": 0.5368,
+      "step": 520
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9518826833916136e-05,
+      "loss": 0.6597,
+      "step": 521
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.951698392407453e-05,
+      "loss": 0.549,
+      "step": 522
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.951513752619613e-05,
+      "loss": 0.7985,
+      "step": 523
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9513287640543605e-05,
+      "loss": 0.8244,
+      "step": 524
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9511434267380154e-05,
+      "loss": 0.5784,
+      "step": 525
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9509577406969456e-05,
+      "loss": 0.6083,
+      "step": 526
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9507717059575676e-05,
+      "loss": 0.4724,
+      "step": 527
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.95058532254635e-05,
+      "loss": 0.3543,
+      "step": 528
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9503985904898096e-05,
+      "loss": 0.3708,
+      "step": 529
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950211509814511e-05,
+      "loss": 0.7684,
+      "step": 530
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.950024080547072e-05,
+      "loss": 0.7386,
+      "step": 531
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949836302714158e-05,
+      "loss": 0.4572,
+      "step": 532
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949648176342483e-05,
+      "loss": 0.7646,
+      "step": 533
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949459701458813e-05,
+      "loss": 0.7853,
+      "step": 534
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949270878089962e-05,
+      "loss": 0.589,
+      "step": 535
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.949081706262794e-05,
+      "loss": 0.6311,
+      "step": 536
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9488921860042225e-05,
+      "loss": 0.6125,
+      "step": 537
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9487023173412105e-05,
+      "loss": 0.6943,
+      "step": 538
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.948512100300771e-05,
+      "loss": 0.5626,
+      "step": 539
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.948321534909965e-05,
+      "loss": 0.733,
+      "step": 540
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.948130621195906e-05,
+      "loss": 0.5031,
+      "step": 541
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9479393591857536e-05,
+      "loss": 0.8029,
+      "step": 542
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94774774890672e-05,
+      "loss": 0.7456,
+      "step": 543
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947555790386066e-05,
+      "loss": 0.5938,
+      "step": 544
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9473634836511e-05,
+      "loss": 0.6412,
+      "step": 545
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.947170828729183e-05,
+      "loss": 0.6935,
+      "step": 546
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946977825647723e-05,
+      "loss": 0.6416,
+      "step": 547
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9467844744341784e-05,
+      "loss": 0.6892,
+      "step": 548
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946590775116059e-05,
+      "loss": 0.4821,
+      "step": 549
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9463967277209214e-05,
+      "loss": 0.8346,
+      "step": 550
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.946202332276372e-05,
+      "loss": 0.6229,
+      "step": 551
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9460075888100685e-05,
+      "loss": 0.7448,
+      "step": 552
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945812497349717e-05,
+      "loss": 0.6213,
+      "step": 553
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945617057923072e-05,
+      "loss": 0.6925,
+      "step": 554
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9454212705579405e-05,
+      "loss": 0.6123,
+      "step": 555
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945225135282177e-05,
+      "loss": 0.8844,
+      "step": 556
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.945028652123685e-05,
+      "loss": 0.7752,
+      "step": 557
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944831821110417e-05,
+      "loss": 0.6953,
+      "step": 558
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944634642270379e-05,
+      "loss": 0.5179,
+      "step": 559
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9444371156316215e-05,
+      "loss": 0.6899,
+      "step": 560
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944239241222248e-05,
+      "loss": 0.4427,
+      "step": 561
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.944041019070409e-05,
+      "loss": 0.3546,
+      "step": 562
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9438424492043055e-05,
+      "loss": 0.4957,
+      "step": 563
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94364353165219e-05,
+      "loss": 0.7468,
+      "step": 564
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9434442664423605e-05,
+      "loss": 0.8872,
+      "step": 565
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.943244653603167e-05,
+      "loss": 0.7649,
+      "step": 566
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9430446931630105e-05,
+      "loss": 0.7209,
+      "step": 567
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942844385150336e-05,
+      "loss": 0.7921,
+      "step": 568
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942643729593643e-05,
+      "loss": 0.7439,
+      "step": 569
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942442726521479e-05,
+      "loss": 0.627,
+      "step": 570
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942241375962441e-05,
+      "loss": 0.5808,
+      "step": 571
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.942039677945175e-05,
+      "loss": 0.5465,
+      "step": 572
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9418376324983755e-05,
+      "loss": 0.7043,
+      "step": 573
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941635239650789e-05,
+      "loss": 0.5361,
+      "step": 574
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9414324994312086e-05,
+      "loss": 0.5844,
+      "step": 575
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.94122941186848e-05,
+      "loss": 0.6288,
+      "step": 576
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.941025976991496e-05,
+      "loss": 0.4109,
+      "step": 577
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940822194829198e-05,
+      "loss": 0.6526,
+      "step": 578
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9406180654105784e-05,
+      "loss": 0.597,
+      "step": 579
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9404135887646794e-05,
+      "loss": 0.4664,
+      "step": 580
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940208764920593e-05,
+      "loss": 0.6071,
+      "step": 581
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.940003593907457e-05,
+      "loss": 0.4779,
+      "step": 582
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.939798075754464e-05,
+      "loss": 0.5935,
+      "step": 583
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.939592210490851e-05,
+      "loss": 0.5991,
+      "step": 584
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9393859981459064e-05,
+      "loss": 0.649,
+      "step": 585
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9391794387489696e-05,
+      "loss": 0.7437,
+      "step": 586
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938972532329427e-05,
+      "loss": 0.5355,
+      "step": 587
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938765278916715e-05,
+      "loss": 0.6618,
+      "step": 588
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.93855767854032e-05,
+      "loss": 0.5402,
+      "step": 589
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.938349731229778e-05,
+      "loss": 0.7387,
+      "step": 590
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9381414370146716e-05,
+      "loss": 0.5386,
+      "step": 591
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937932795924637e-05,
+      "loss": 0.4341,
+      "step": 592
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937723807989356e-05,
+      "loss": 0.7961,
+      "step": 593
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.937514473238562e-05,
+      "loss": 0.5328,
+      "step": 594
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9373047917020385e-05,
+      "loss": 0.6803,
+      "step": 595
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9370947634096156e-05,
+      "loss": 0.568,
+      "step": 596
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9368843883911744e-05,
+      "loss": 0.728,
+      "step": 597
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9366736666766435e-05,
+      "loss": 0.7673,
+      "step": 598
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9364625982960044e-05,
+      "loss": 0.6602,
+      "step": 599
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936251183279284e-05,
+      "loss": 0.5541,
+      "step": 600
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936039421656563e-05,
+      "loss": 0.5701,
+      "step": 601
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935827313457967e-05,
+      "loss": 0.8304,
+      "step": 602
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935614858713672e-05,
+      "loss": 0.7207,
+      "step": 603
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935402057453905e-05,
+      "loss": 0.759,
+      "step": 604
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.935188909708942e-05,
+      "loss": 0.9485,
+      "step": 605
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9349754155091056e-05,
+      "loss": 0.5145,
+      "step": 606
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9347615748847716e-05,
+      "loss": 0.5754,
+      "step": 607
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934547387866362e-05,
+      "loss": 0.7388,
+      "step": 608
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.934332854484349e-05,
+      "loss": 0.6056,
+      "step": 609
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9341179747692554e-05,
+      "loss": 0.7318,
+      "step": 610
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933902748751652e-05,
+      "loss": 0.9205,
+      "step": 611
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933687176462159e-05,
+      "loss": 0.5657,
+      "step": 612
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.933471257931445e-05,
+      "loss": 0.7589,
+      "step": 613
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.93325499319023e-05,
+      "loss": 0.665,
+      "step": 614
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.93303838226928e-05,
+      "loss": 0.7736,
+      "step": 615
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9328214251994157e-05,
+      "loss": 0.7369,
+      "step": 616
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9326041220115003e-05,
+      "loss": 0.5004,
+      "step": 617
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.932386472736451e-05,
+      "loss": 0.8621,
+      "step": 618
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.932168477405233e-05,
+      "loss": 0.626,
+      "step": 619
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.931950136048861e-05,
+      "loss": 0.7479,
+      "step": 620
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.931731448698397e-05,
+      "loss": 0.5895,
+      "step": 621
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9315124153849546e-05,
+      "loss": 0.7133,
+      "step": 622
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.931293036139695e-05,
+      "loss": 0.5624,
+      "step": 623
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.931073310993831e-05,
+      "loss": 0.7548,
+      "step": 624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9308532399786214e-05,
+      "loss": 0.9662,
+      "step": 625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930632823125376e-05,
+      "loss": 0.7777,
+      "step": 626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930412060465454e-05,
+      "loss": 0.4913,
+      "step": 627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.930190952030263e-05,
+      "loss": 0.7241,
+      "step": 628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929969497851261e-05,
+      "loss": 0.7365,
+      "step": 629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929747697959952e-05,
+      "loss": 0.8709,
+      "step": 630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929525552387895e-05,
+      "loss": 0.6407,
+      "step": 631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.929303061166691e-05,
+      "loss": 0.775,
+      "step": 632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9290802243279976e-05,
+      "loss": 0.8483,
+      "step": 633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9288570419035147e-05,
+      "loss": 1.0604,
+      "step": 634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928633513924996e-05,
+      "loss": 0.8256,
+      "step": 635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.928409640424243e-05,
+      "loss": 0.7682,
+      "step": 636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9281854214331056e-05,
+      "loss": 0.8142,
+      "step": 637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927960856983483e-05,
+      "loss": 0.6088,
+      "step": 638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927735947107326e-05,
+      "loss": 0.5822,
+      "step": 639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927510691836631e-05,
+      "loss": 0.7809,
+      "step": 640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.927285091203445e-05,
+      "loss": 0.9087,
+      "step": 641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9270591452398655e-05,
+      "loss": 0.8923,
+      "step": 642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.926832853978037e-05,
+      "loss": 0.6715,
+      "step": 643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9266062174501535e-05,
+      "loss": 0.4525,
+      "step": 644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92637923568846e-05,
+      "loss": 0.5788,
+      "step": 645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9261519087252486e-05,
+      "loss": 0.7011,
+      "step": 646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925924236592861e-05,
+      "loss": 0.598,
+      "step": 647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9256962193236874e-05,
+      "loss": 0.7066,
+      "step": 648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925467856950169e-05,
+      "loss": 0.6576,
+      "step": 649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925239149504795e-05,
+      "loss": 0.5895,
+      "step": 650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.925010097020103e-05,
+      "loss": 0.4862,
+      "step": 651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924780699528681e-05,
+      "loss": 0.7725,
+      "step": 652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9245509570631656e-05,
+      "loss": 0.4415,
+      "step": 653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924320869656241e-05,
+      "loss": 0.8077,
+      "step": 654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.924090437340643e-05,
+      "loss": 0.684,
+      "step": 655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923859660149154e-05,
+      "loss": 0.9005,
+      "step": 656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9236285381146084e-05,
+      "loss": 0.7167,
+      "step": 657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9233970712698876e-05,
+      "loss": 0.7717,
+      "step": 658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.923165259647922e-05,
+      "loss": 0.7469,
+      "step": 659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.922933103281691e-05,
+      "loss": 0.4445,
+      "step": 660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9227006022042235e-05,
+      "loss": 0.8721,
+      "step": 661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9224677564485985e-05,
+      "loss": 0.6564,
+      "step": 662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9222345660479424e-05,
+      "loss": 0.4603,
+      "step": 663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9220010310354316e-05,
+      "loss": 0.835,
+      "step": 664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.921767151444291e-05,
+      "loss": 0.6079,
+      "step": 665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.921532927307795e-05,
+      "loss": 0.7051,
+      "step": 666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.921298358659266e-05,
+      "loss": 0.6295,
+      "step": 667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9210634455320764e-05,
+      "loss": 1.209,
+      "step": 668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9208281879596474e-05,
+      "loss": 0.9042,
+      "step": 669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.92059258597545e-05,
+      "loss": 0.8073,
+      "step": 670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.920356639613002e-05,
+      "loss": 0.9003,
+      "step": 671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9201203489058725e-05,
+      "loss": 0.467,
+      "step": 672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.919883713887677e-05,
+      "loss": 0.5991,
+      "step": 673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.919646734592084e-05,
+      "loss": 0.562,
+      "step": 674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.919409411052807e-05,
+      "loss": 0.7762,
+      "step": 675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.919171743303611e-05,
+      "loss": 0.4568,
+      "step": 676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.918933731378308e-05,
+      "loss": 0.3574,
+      "step": 677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.918695375310761e-05,
+      "loss": 0.6516,
+      "step": 678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.91845667513488e-05,
+      "loss": 0.623,
+      "step": 679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9182176308846264e-05,
+      "loss": 0.645,
+      "step": 680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9179782425940076e-05,
+      "loss": 0.6949,
+      "step": 681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.917738510297082e-05,
+      "loss": 0.6143,
+      "step": 682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.917498434027956e-05,
+      "loss": 0.7482,
+      "step": 683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.917258013820786e-05,
+      "loss": 0.5978,
+      "step": 684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.917017249709777e-05,
+      "loss": 0.6876,
+      "step": 685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.916776141729181e-05,
+      "loss": 0.6459,
+      "step": 686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9165346899133016e-05,
+      "loss": 0.5558,
+      "step": 687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.91629289429649e-05,
+      "loss": 0.8462,
+      "step": 688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9160507549131464e-05,
+      "loss": 0.6956,
+      "step": 689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9158082717977195e-05,
+      "loss": 0.7213,
+      "step": 690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.915565444984709e-05,
+      "loss": 0.7166,
+      "step": 691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9153222745086605e-05,
+      "loss": 0.5835,
+      "step": 692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.915078760404171e-05,
+      "loss": 0.5818,
+      "step": 693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.914834902705883e-05,
+      "loss": 0.7939,
+      "step": 694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.914590701448493e-05,
+      "loss": 0.675,
+      "step": 695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.914346156666743e-05,
+      "loss": 0.8684,
+      "step": 696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9141012683954225e-05,
+      "loss": 0.7618,
+      "step": 697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9138560366693744e-05,
+      "loss": 0.3676,
+      "step": 698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9136104615234856e-05,
+      "loss": 0.6624,
+      "step": 699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.913364542992696e-05,
+      "loss": 0.5545,
+      "step": 700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9131182811119915e-05,
+      "loss": 0.6949,
+      "step": 701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9128716759164073e-05,
+      "loss": 0.8967,
+      "step": 702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9126247274410295e-05,
+      "loss": 0.4507,
+      "step": 703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.91237743572099e-05,
+      "loss": 0.5518,
+      "step": 704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.9121298007914725e-05,
+      "loss": 0.594,
+      "step": 705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.911881822687707e-05,
+      "loss": 0.7625,
+      "step": 706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.911633501444973e-05,
+      "loss": 0.7824,
+      "step": 707
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911384837098601e-05,
+      "loss": 0.6897,
+      "step": 708
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.911135829683967e-05,
+      "loss": 0.7174,
+      "step": 709
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9108864792364975e-05,
+      "loss": 0.7281,
+      "step": 710
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910636785791669e-05,
+      "loss": 0.5364,
+      "step": 711
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910386749385003e-05,
+      "loss": 1.129,
+      "step": 712
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.910136370052074e-05,
+      "loss": 0.6174,
+      "step": 713
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909885647828504e-05,
+      "loss": 0.5695,
+      "step": 714
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9096345827499625e-05,
+      "loss": 0.6647,
+      "step": 715
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909383174852167e-05,
+      "loss": 0.4971,
+      "step": 716
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.909131424170889e-05,
+      "loss": 0.9954,
+      "step": 717
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908879330741941e-05,
+      "loss": 0.699,
+      "step": 718
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9086268946011915e-05,
+      "loss": 0.6012,
+      "step": 719
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.908374115784553e-05,
+      "loss": 0.5986,
+      "step": 720
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9081209943279897e-05,
+      "loss": 0.6256,
+      "step": 721
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.907867530267512e-05,
+      "loss": 0.6591,
+      "step": 722
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.907613723639181e-05,
+      "loss": 0.6381,
+      "step": 723
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.907359574479106e-05,
+      "loss": 0.519,
+      "step": 724
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9071050828234435e-05,
+      "loss": 0.5101,
+      "step": 725
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.906850248708401e-05,
+      "loss": 0.648,
+      "step": 726
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9065950721702345e-05,
+      "loss": 0.6758,
+      "step": 727
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9063395532452475e-05,
+      "loss": 0.9672,
+      "step": 728
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9060836919697927e-05,
+      "loss": 0.5112,
+      "step": 729
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9058274883802715e-05,
+      "loss": 0.6354,
+      "step": 730
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9055709425131335e-05,
+      "loss": 0.6672,
+      "step": 731
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.905314054404878e-05,
+      "loss": 0.9341,
+      "step": 732
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.905056824092054e-05,
+      "loss": 0.8474,
+      "step": 733
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904799251611254e-05,
+      "loss": 0.5189,
+      "step": 734
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904541336999127e-05,
+      "loss": 0.3746,
+      "step": 735
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904283080292364e-05,
+      "loss": 0.6005,
+      "step": 736
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.904024481527708e-05,
+      "loss": 0.6505,
+      "step": 737
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9037655407419504e-05,
+      "loss": 0.7672,
+      "step": 738
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9035062579719293e-05,
+      "loss": 0.4974,
+      "step": 739
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.903246633254535e-05,
+      "loss": 0.6677,
+      "step": 740
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.902986666626702e-05,
+      "loss": 0.8069,
+      "step": 741
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.902726358125418e-05,
+      "loss": 0.6062,
+      "step": 742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.902465707787716e-05,
+      "loss": 0.7814,
+      "step": 743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.902204715650679e-05,
+      "loss": 0.4542,
+      "step": 744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.901943381751438e-05,
+      "loss": 0.4033,
+      "step": 745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.901681706127174e-05,
+      "loss": 0.6306,
+      "step": 746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.901419688815116e-05,
+      "loss": 0.4829,
+      "step": 747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.901157329852538e-05,
+      "loss": 0.7854,
+      "step": 748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.90089462927677e-05,
+      "loss": 0.7018,
+      "step": 749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.900631587125184e-05,
+      "loss": 0.9645,
+      "step": 750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.900368203435204e-05,
+      "loss": 0.7626,
+      "step": 751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9001044782443006e-05,
+      "loss": 0.6776,
+      "step": 752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.899840411589996e-05,
+      "loss": 0.8715,
+      "step": 753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.899576003509857e-05,
+      "loss": 0.6869,
+      "step": 754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8993112540415026e-05,
+      "loss": 0.6192,
+      "step": 755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.899046163222597e-05,
+      "loss": 0.6499,
+      "step": 756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.898780731090857e-05,
+      "loss": 0.5635,
+      "step": 757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.898514957684044e-05,
+      "loss": 0.4792,
+      "step": 758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8982488430399695e-05,
+      "loss": 0.8527,
+      "step": 759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.897982387196495e-05,
+      "loss": 0.6845,
+      "step": 760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.897715590191528e-05,
+      "loss": 0.7014,
+      "step": 761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.897448452063027e-05,
+      "loss": 0.5905,
+      "step": 762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.897180972848997e-05,
+      "loss": 0.5338,
+      "step": 763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.896913152587492e-05,
+      "loss": 0.5837,
+      "step": 764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.896644991316615e-05,
+      "loss": 0.8565,
+      "step": 765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8963764890745175e-05,
+      "loss": 0.7392,
+      "step": 766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8961076458994004e-05,
+      "loss": 0.5167,
+      "step": 767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.895838461829511e-05,
+      "loss": 0.6016,
+      "step": 768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8955689369031465e-05,
+      "loss": 0.6708,
+      "step": 769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8952990711586514e-05,
+      "loss": 0.665,
+      "step": 770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.89502886463442e-05,
+      "loss": 0.5544,
+      "step": 771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.894758317368896e-05,
+      "loss": 0.8858,
+      "step": 772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8944874294005693e-05,
+      "loss": 0.7191,
+      "step": 773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.894216200767978e-05,
+      "loss": 0.6097,
+      "step": 774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.893944631509712e-05,
+      "loss": 0.827,
+      "step": 775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8936727216644054e-05,
+      "loss": 0.8695,
+      "step": 776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.893400471270744e-05,
+      "loss": 0.8041,
+      "step": 777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8931278803674615e-05,
+      "loss": 0.5615,
+      "step": 778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.892854948993338e-05,
+      "loss": 0.6602,
+      "step": 779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.892581677187204e-05,
+      "loss": 0.9706,
+      "step": 780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.89230806498794e-05,
+      "loss": 0.75,
+      "step": 781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.89203411243447e-05,
+      "loss": 0.5475,
+      "step": 782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.891759819565771e-05,
+      "loss": 0.5734,
+      "step": 783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.891485186420865e-05,
+      "loss": 0.4893,
+      "step": 784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.891210213038826e-05,
+      "loss": 0.7552,
+      "step": 785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8909348994587734e-05,
+      "loss": 0.3919,
+      "step": 786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.890659245719876e-05,
+      "loss": 0.6639,
+      "step": 787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8903832518613525e-05,
+      "loss": 0.5349,
+      "step": 788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.890106917922468e-05,
+      "loss": 0.4397,
+      "step": 789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8898302439425356e-05,
+      "loss": 0.7298,
+      "step": 790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.889553229960918e-05,
+      "loss": 0.6219,
+      "step": 791
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.889275876017027e-05,
+      "loss": 0.5439,
+      "step": 792
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8889981821503214e-05,
+      "loss": 0.8646,
+      "step": 793
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8887201484003094e-05,
+      "loss": 0.4868,
+      "step": 794
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.888441774806545e-05,
+      "loss": 0.6772,
+      "step": 795
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.888163061408634e-05,
+      "loss": 0.8711,
+      "step": 796
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8878840082462296e-05,
+      "loss": 0.7512,
+      "step": 797
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8876046153590306e-05,
+      "loss": 0.6742,
+      "step": 798
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887324882786788e-05,
+      "loss": 0.7078,
+      "step": 799
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.887044810569299e-05,
+      "loss": 0.8666,
+      "step": 800
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8867643987464095e-05,
+      "loss": 0.559,
+      "step": 801
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8864836473580144e-05,
+      "loss": 0.5545,
+      "step": 802
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.886202556444055e-05,
+      "loss": 0.4231,
+      "step": 803
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885921126044523e-05,
+      "loss": 0.405,
+      "step": 804
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.885639356199456e-05,
+      "loss": 0.8341,
+      "step": 805
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8853572469489444e-05,
+      "loss": 0.5811,
+      "step": 806
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8850747983331226e-05,
+      "loss": 0.4906,
+      "step": 807
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.884792010392173e-05,
+      "loss": 0.7888,
+      "step": 808
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.88450888316633e-05,
+      "loss": 0.6743,
+      "step": 809
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.884225416695873e-05,
+      "loss": 0.7007,
+      "step": 810
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883941611021132e-05,
+      "loss": 0.8535,
+      "step": 811
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883657466182483e-05,
+      "loss": 0.6927,
+      "step": 812
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8833729822203524e-05,
+      "loss": 0.5036,
+      "step": 813
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.883088159175212e-05,
+      "loss": 0.752,
+      "step": 814
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882802997087585e-05,
+      "loss": 0.4038,
+      "step": 815
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.882517495998042e-05,
+      "loss": 0.4766,
+      "step": 816
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8822316559472e-05,
+      "loss": 0.6957,
+      "step": 817
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.881945476975727e-05,
+      "loss": 0.4474,
+      "step": 818
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8816589591243355e-05,
+      "loss": 0.7323,
+      "step": 819
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.88137210243379e-05,
+      "loss": 0.5557,
+      "step": 820
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8810849069449015e-05,
+      "loss": 0.5647,
+      "step": 821
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.88079737269853e-05,
+      "loss": 0.5712,
+      "step": 822
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880509499735582e-05,
+      "loss": 0.5068,
+      "step": 823
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.880221288097014e-05,
+      "loss": 0.5197,
+      "step": 824
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.879932737823829e-05,
+      "loss": 0.7712,
+      "step": 825
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8796438489570796e-05,
+      "loss": 0.7556,
+      "step": 826
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8793546215378664e-05,
+      "loss": 0.7889,
+      "step": 827
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.879065055607338e-05,
+      "loss": 0.5131,
+      "step": 828
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8787751512066904e-05,
+      "loss": 0.5765,
+      "step": 829
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.878484908377168e-05,
+      "loss": 0.5921,
+      "step": 830
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.878194327160065e-05,
+      "loss": 0.7196,
+      "step": 831
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.877903407596721e-05,
+      "loss": 0.5747,
+      "step": 832
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.877612149728527e-05,
+      "loss": 0.7026,
+      "step": 833
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8773205535969194e-05,
+      "loss": 0.4887,
+      "step": 834
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.877028619243382e-05,
+      "loss": 0.4189,
+      "step": 835
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8767363467094505e-05,
+      "loss": 0.6846,
+      "step": 836
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.876443736036707e-05,
+      "loss": 0.5571,
+      "step": 837
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8761507872667786e-05,
+      "loss": 0.7281,
+      "step": 838
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8758575004413464e-05,
+      "loss": 0.7783,
+      "step": 839
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.875563875602134e-05,
+      "loss": 0.5683,
+      "step": 840
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.875269912790916e-05,
+      "loss": 0.6116,
+      "step": 841
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.874975612049515e-05,
+      "loss": 0.7064,
+      "step": 842
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.874680973419802e-05,
+      "loss": 0.7161,
+      "step": 843
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8743859969436934e-05,
+      "loss": 0.5457,
+      "step": 844
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.874090682663157e-05,
+      "loss": 0.7711,
+      "step": 845
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.873795030620206e-05,
+      "loss": 0.9191,
+      "step": 846
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8734990408569045e-05,
+      "loss": 0.3196,
+      "step": 847
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.873202713415362e-05,
+      "loss": 0.886,
+      "step": 848
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8729060483377374e-05,
+      "loss": 0.6678,
+      "step": 849
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8726090456662364e-05,
+      "loss": 0.6702,
+      "step": 850
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.872311705443116e-05,
+      "loss": 0.7061,
+      "step": 851
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.872014027710676e-05,
+      "loss": 0.465,
+      "step": 852
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.871716012511268e-05,
+      "loss": 0.5834,
+      "step": 853
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.871417659887292e-05,
+      "loss": 0.5504,
+      "step": 854
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.871118969881193e-05,
+      "loss": 0.7189,
+      "step": 855
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.870819942535467e-05,
+      "loss": 0.707,
+      "step": 856
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.870520577892656e-05,
+      "loss": 0.6282,
+      "step": 857
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8702208759953506e-05,
+      "loss": 0.6487,
+      "step": 858
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.86992083688619e-05,
+      "loss": 0.5934,
+      "step": 859
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.86962046060786e-05,
+      "loss": 0.5895,
+      "step": 860
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.869319747203096e-05,
+      "loss": 0.6092,
+      "step": 861
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.86901869671468e-05,
+      "loss": 0.7437,
+      "step": 862
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.868717309185443e-05,
+      "loss": 0.7044,
+      "step": 863
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.868415584658263e-05,
+      "loss": 0.6445,
+      "step": 864
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.868113523176067e-05,
+      "loss": 0.5497,
+      "step": 865
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.867811124781828e-05,
+      "loss": 0.4526,
+      "step": 866
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8675083895185704e-05,
+      "loss": 0.6394,
+      "step": 867
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.867205317429363e-05,
+      "loss": 0.8167,
+      "step": 868
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.866901908557324e-05,
+      "loss": 0.5768,
+      "step": 869
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.86659816294562e-05,
+      "loss": 0.6527,
+      "step": 870
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.866294080637465e-05,
+      "loss": 0.9283,
+      "step": 871
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.865989661676119e-05,
+      "loss": 0.7034,
+      "step": 872
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.865684906104895e-05,
+      "loss": 0.6473,
+      "step": 873
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8653798139671484e-05,
+      "loss": 0.8488,
+      "step": 874
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.865074385306284e-05,
+      "loss": 0.5204,
+      "step": 875
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864768620165758e-05,
+      "loss": 0.5566,
+      "step": 876
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864462518589069e-05,
+      "loss": 0.5604,
+      "step": 877
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.864156080619768e-05,
+      "loss": 0.6311,
+      "step": 878
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.863849306301451e-05,
+      "loss": 0.9286,
+      "step": 879
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.863542195677763e-05,
+      "loss": 0.8607,
+      "step": 880
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8632347487923965e-05,
+      "loss": 0.5667,
+      "step": 881
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862926965689093e-05,
+      "loss": 0.779,
+      "step": 882
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8626188464116395e-05,
+      "loss": 0.7826,
+      "step": 883
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8623103910038734e-05,
+      "loss": 0.8848,
+      "step": 884
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.862001599509678e-05,
+      "loss": 0.583,
+      "step": 885
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.861692471972985e-05,
+      "loss": 0.7987,
+      "step": 886
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.861383008437776e-05,
+      "loss": 0.7043,
+      "step": 887
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.861073208948075e-05,
+      "loss": 0.5354,
+      "step": 888
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.860763073547959e-05,
+      "loss": 0.8232,
+      "step": 889
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.860452602281552e-05,
+      "loss": 0.5787,
+      "step": 890
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.860141795193024e-05,
+      "loss": 0.388,
+      "step": 891
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8598306523265935e-05,
+      "loss": 0.5726,
+      "step": 892
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8595191737265264e-05,
+      "loss": 0.6519,
+      "step": 893
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.859207359437138e-05,
+      "loss": 0.7241,
+      "step": 894
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858895209502788e-05,
+      "loss": 0.4349,
+      "step": 895
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8585827239678895e-05,
+      "loss": 0.7367,
+      "step": 896
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.858269902876897e-05,
+      "loss": 0.8832,
+      "step": 897
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.857956746274317e-05,
+      "loss": 0.8083,
+      "step": 898
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.857643254204701e-05,
+      "loss": 0.8872,
+      "step": 899
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.857329426712651e-05,
+      "loss": 0.6477,
+      "step": 900
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.857015263842815e-05,
+      "loss": 0.8799,
+      "step": 901
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856700765639889e-05,
+      "loss": 0.609,
+      "step": 902
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856385932148616e-05,
+      "loss": 0.8192,
+      "step": 903
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.856070763413789e-05,
+      "loss": 0.7486,
+      "step": 904
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.855755259480245e-05,
+      "loss": 0.5706,
+      "step": 905
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.855439420392872e-05,
+      "loss": 0.6296,
+      "step": 906
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8551232461966055e-05,
+      "loss": 0.8281,
+      "step": 907
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.854806736936426e-05,
+      "loss": 0.8153,
+      "step": 908
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8544898926573635e-05,
+      "loss": 0.7478,
+      "step": 909
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8541727134044967e-05,
+      "loss": 0.5553,
+      "step": 910
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8538551992229494e-05,
+      "loss": 0.5511,
+      "step": 911
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.853537350157896e-05,
+      "loss": 0.6156,
+      "step": 912
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.853219166254555e-05,
+      "loss": 0.7314,
+      "step": 913
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.852900647558196e-05,
+      "loss": 1.0112,
+      "step": 914
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.852581794114134e-05,
+      "loss": 0.8484,
+      "step": 915
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8522626059677327e-05,
+      "loss": 0.5981,
+      "step": 916
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.851943083164403e-05,
+      "loss": 0.7449,
+      "step": 917
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.851623225749603e-05,
+      "loss": 0.5046,
+      "step": 918
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8513030337688404e-05,
+      "loss": 0.7037,
+      "step": 919
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.850982507267667e-05,
+      "loss": 0.6122,
+      "step": 920
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.850661646291685e-05,
+      "loss": 0.8582,
+      "step": 921
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8503404508865436e-05,
+      "loss": 0.489,
+      "step": 922
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.85001892109794e-05,
+      "loss": 0.7435,
+      "step": 923
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.849697056971616e-05,
+      "loss": 0.7841,
+      "step": 924
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.849374858553366e-05,
+      "loss": 0.7609,
+      "step": 925
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8490523258890276e-05,
+      "loss": 0.8044,
+      "step": 926
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8487294590244876e-05,
+      "loss": 0.6549,
+      "step": 927
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.848406258005681e-05,
+      "loss": 0.9293,
+      "step": 928
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8480827228785895e-05,
+      "loss": 0.988,
+      "step": 929
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.847758853689243e-05,
+      "loss": 0.7781,
+      "step": 930
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.847434650483717e-05,
+      "loss": 0.6949,
+      "step": 931
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.847110113308138e-05,
+      "loss": 0.7006,
+      "step": 932
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.846785242208675e-05,
+      "loss": 0.5422,
+      "step": 933
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8464600372315504e-05,
+      "loss": 0.7977,
+      "step": 934
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8461344984230306e-05,
+      "loss": 0.6237,
+      "step": 935
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8458086258294286e-05,
+      "loss": 0.7167,
+      "step": 936
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.845482419497108e-05,
+      "loss": 0.6533,
+      "step": 937
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8451558794724774e-05,
+      "loss": 0.6242,
+      "step": 938
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.844829005801994e-05,
+      "loss": 0.6619,
+      "step": 939
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8445017985321614e-05,
+      "loss": 0.5986,
+      "step": 940
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.844174257709533e-05,
+      "loss": 0.6526,
+      "step": 941
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8438463833807076e-05,
+      "loss": 0.4991,
+      "step": 942
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8435181755923307e-05,
+      "loss": 0.5048,
+      "step": 943
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.843189634391098e-05,
+      "loss": 0.597,
+      "step": 944
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.84286075982375e-05,
+      "loss": 0.5275,
+      "step": 945
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.842531551937077e-05,
+      "loss": 0.5992,
+      "step": 946
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.842202010777914e-05,
+      "loss": 0.6174,
+      "step": 947
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.841872136393146e-05,
+      "loss": 0.614,
+      "step": 948
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8415419288297045e-05,
+      "loss": 0.7894,
+      "step": 949
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8412113881345675e-05,
+      "loss": 0.9255,
+      "step": 950
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8408805143547605e-05,
+      "loss": 0.7516,
+      "step": 951
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.840549307537359e-05,
+      "loss": 1.0145,
+      "step": 952
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8402177677294815e-05,
+      "loss": 0.5981,
+      "step": 953
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.839885894978298e-05,
+      "loss": 0.8187,
+      "step": 954
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.839553689331023e-05,
+      "loss": 0.5689,
+      "step": 955
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8392211508349206e-05,
+      "loss": 0.5773,
+      "step": 956
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8388882795373e-05,
+      "loss": 0.5862,
+      "step": 957
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.838555075485519e-05,
+      "loss": 0.579,
+      "step": 958
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.838221538726983e-05,
+      "loss": 0.6018,
+      "step": 959
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837887669309146e-05,
+      "loss": 0.6908,
+      "step": 960
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837553467279505e-05,
+      "loss": 0.5647,
+      "step": 961
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8372189326856076e-05,
+      "loss": 0.7656,
+      "step": 962
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.836884065575049e-05,
+      "loss": 0.7924,
+      "step": 963
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83654886599547e-05,
+      "loss": 0.8434,
+      "step": 964
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.83621333399456e-05,
+      "loss": 0.6745,
+      "step": 965
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.835877469620055e-05,
+      "loss": 0.4787,
+      "step": 966
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8355412729197384e-05,
+      "loss": 0.671,
+      "step": 967
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.835204743941442e-05,
+      "loss": 0.6542,
+      "step": 968
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.834867882733042e-05,
+      "loss": 0.705,
+      "step": 969
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8345306893424646e-05,
+      "loss": 0.7163,
+      "step": 970
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.834193163817683e-05,
+      "loss": 0.5934,
+      "step": 971
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.833855306206717e-05,
+      "loss": 0.734,
+      "step": 972
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.833517116557632e-05,
+      "loss": 0.8181,
+      "step": 973
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.833178594918545e-05,
+      "loss": 0.6569,
+      "step": 974
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.832839741337615e-05,
+      "loss": 0.5505,
+      "step": 975
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8325005558630524e-05,
+      "loss": 0.5422,
+      "step": 976
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8321610385431124e-05,
+      "loss": 0.8015,
+      "step": 977
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.831821189426098e-05,
+      "loss": 0.5079,
+      "step": 978
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8314810085603604e-05,
+      "loss": 0.9849,
+      "step": 979
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.831140495994297e-05,
+      "loss": 0.623,
+      "step": 980
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.830799651776353e-05,
+      "loss": 0.9143,
+      "step": 981
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8304584759550186e-05,
+      "loss": 0.8716,
+      "step": 982
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.830116968578835e-05,
+      "loss": 0.8327,
+      "step": 983
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.829775129696388e-05,
+      "loss": 0.6307,
+      "step": 984
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8294329593563104e-05,
+      "loss": 0.8268,
+      "step": 985
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.829090457607284e-05,
+      "loss": 0.6508,
+      "step": 986
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.828747624498034e-05,
+      "loss": 0.6678,
+      "step": 987
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.828404460077339e-05,
+      "loss": 0.8898,
+      "step": 988
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.828060964394019e-05,
+      "loss": 0.6019,
+      "step": 989
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.827717137496943e-05,
+      "loss": 0.8221,
+      "step": 990
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.827372979435029e-05,
+      "loss": 0.7231,
+      "step": 991
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.827028490257238e-05,
+      "loss": 0.6391,
+      "step": 992
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8266836700125825e-05,
+      "loss": 0.6262,
+      "step": 993
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8263385187501194e-05,
+      "loss": 0.4205,
+      "step": 994
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.825993036518953e-05,
+      "loss": 0.6487,
+      "step": 995
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.825647223368237e-05,
+      "loss": 0.6628,
+      "step": 996
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8253010793471674e-05,
+      "loss": 0.5193,
+      "step": 997
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.824954604504993e-05,
+      "loss": 0.8226,
+      "step": 998
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.824607798891005e-05,
+      "loss": 0.7922,
+      "step": 999
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.824260662554544e-05,
+      "loss": 0.79,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 0.73876953125,
+      "eval_runtime": 3.6677,
+      "eval_samples_per_second": 27.265,
+      "eval_steps_per_second": 4.635,
+      "step": 1000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.823913195544998e-05,
+      "loss": 0.8851,
+      "step": 1001
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8235653979118e-05,
+      "loss": 0.9105,
+      "step": 1002
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.823217269704432e-05,
+      "loss": 0.6243,
+      "step": 1003
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8228688109724216e-05,
+      "loss": 0.8767,
+      "step": 1004
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.822520021765346e-05,
+      "loss": 0.7766,
+      "step": 1005
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8221709021328246e-05,
+      "loss": 0.5068,
+      "step": 1006
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.821821452124528e-05,
+      "loss": 0.8279,
+      "step": 1007
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.821471671790173e-05,
+      "loss": 0.5951,
+      "step": 1008
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.821121561179523e-05,
+      "loss": 0.7791,
+      "step": 1009
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8207711203423886e-05,
+      "loss": 0.8087,
+      "step": 1010
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8204203493286256e-05,
+      "loss": 0.6902,
+      "step": 1011
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.820069248188139e-05,
+      "loss": 0.8698,
+      "step": 1012
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8197178169708804e-05,
+      "loss": 0.8274,
+      "step": 1013
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.819366055726848e-05,
+      "loss": 0.4873,
+      "step": 1014
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.819013964506087e-05,
+      "loss": 0.4871,
+      "step": 1015
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.818661543358689e-05,
+      "loss": 0.5518,
+      "step": 1016
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8183087923347926e-05,
+      "loss": 0.9852,
+      "step": 1017
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8179557114845854e-05,
+      "loss": 0.6652,
+      "step": 1018
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.817602300858298e-05,
+      "loss": 0.7183,
+      "step": 1019
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.817248560506213e-05,
+      "loss": 0.5499,
+      "step": 1020
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.816894490478655e-05,
+      "loss": 0.5274,
+      "step": 1021
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.816540090825998e-05,
+      "loss": 0.8313,
+      "step": 1022
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.816185361598663e-05,
+      "loss": 0.4417,
+      "step": 1023
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8158303028471174e-05,
+      "loss": 0.4507,
+      "step": 1024
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8154749146218746e-05,
+      "loss": 0.8358,
+      "step": 1025
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8151191969734976e-05,
+      "loss": 0.5873,
+      "step": 1026
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.814763149952592e-05,
+      "loss": 0.7529,
+      "step": 1027
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8144067736098155e-05,
+      "loss": 0.8412,
+      "step": 1028
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8140500679958664e-05,
+      "loss": 0.6327,
+      "step": 1029
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.813693033161497e-05,
+      "loss": 0.9086,
+      "step": 1030
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8133356691575e-05,
+      "loss": 0.8191,
+      "step": 1031
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.812977976034718e-05,
+      "loss": 0.8268,
+      "step": 1032
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.812619953844042e-05,
+      "loss": 0.8367,
+      "step": 1033
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.812261602636406e-05,
+      "loss": 0.5667,
+      "step": 1034
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811902922462793e-05,
+      "loss": 0.788,
+      "step": 1035
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811543913374233e-05,
+      "loss": 0.6863,
+      "step": 1036
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.811184575421802e-05,
+      "loss": 0.6124,
+      "step": 1037
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810824908656623e-05,
+      "loss": 0.5898,
+      "step": 1038
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.810464913129866e-05,
+      "loss": 0.4922,
+      "step": 1039
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.8101045888927484e-05,
+      "loss": 0.8782,
+      "step": 1040
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.809743935996532e-05,
+      "loss": 0.8561,
+      "step": 1041
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.809382954492529e-05,
+      "loss": 0.6383,
+      "step": 1042
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.809021644432094e-05,
+      "loss": 1.0354,
+      "step": 1043
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.808660005866632e-05,
+      "loss": 0.6246,
+      "step": 1044
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8082980388475926e-05,
+      "loss": 0.5062,
+      "step": 1045
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.807935743426475e-05,
+      "loss": 0.7139,
+      "step": 1046
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.807573119654821e-05,
+      "loss": 0.5281,
+      "step": 1047
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.807210167584222e-05,
+      "loss": 0.6935,
+      "step": 1048
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.806846887266315e-05,
+      "loss": 0.7541,
+      "step": 1049
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8064832787527834e-05,
+      "loss": 0.6963,
+      "step": 1050
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.806119342095359e-05,
+      "loss": 1.038,
+      "step": 1051
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.805755077345819e-05,
+      "loss": 0.8916,
+      "step": 1052
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.805390484555987e-05,
+      "loss": 0.5105,
+      "step": 1053
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8050255637777334e-05,
+      "loss": 0.5748,
+      "step": 1054
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8046603150629756e-05,
+      "loss": 0.6761,
+      "step": 1055
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.804294738463678e-05,
+      "loss": 0.5219,
+      "step": 1056
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8039288340318526e-05,
+      "loss": 0.7798,
+      "step": 1057
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.803562601819554e-05,
+      "loss": 0.6982,
+      "step": 1058
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.803196041878888e-05,
+      "loss": 0.5231,
+      "step": 1059
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8028291542620044e-05,
+      "loss": 0.5778,
+      "step": 1060
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8024619390211e-05,
+      "loss": 0.4603,
+      "step": 1061
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.8020943962084196e-05,
+      "loss": 0.6802,
+      "step": 1062
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.801726525876253e-05,
+      "loss": 0.4814,
+      "step": 1063
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.801358328076938e-05,
+      "loss": 0.6588,
+      "step": 1064
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800989802862857e-05,
+      "loss": 0.7169,
+      "step": 1065
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800620950286442e-05,
+      "loss": 0.5791,
+      "step": 1066
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.800251770400167e-05,
+      "loss": 0.8467,
+      "step": 1067
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.799882263256558e-05,
+      "loss": 0.6709,
+      "step": 1068
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7995124289081824e-05,
+      "loss": 0.5956,
+      "step": 1069
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7991422674076595e-05,
+      "loss": 0.6873,
+      "step": 1070
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7987717788076494e-05,
+      "loss": 1.0767,
+      "step": 1071
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.798400963160863e-05,
+      "loss": 0.5312,
+      "step": 1072
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7980298205200555e-05,
+      "loss": 0.5714,
+      "step": 1073
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7976583509380316e-05,
+      "loss": 0.8619,
+      "step": 1074
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.797286554467637e-05,
+      "loss": 0.7446,
+      "step": 1075
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796914431161771e-05,
+      "loss": 0.6815,
+      "step": 1076
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.796541981073373e-05,
+      "loss": 0.7936,
+      "step": 1077
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7961692042554315e-05,
+      "loss": 0.6555,
+      "step": 1078
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.795796100760983e-05,
+      "loss": 0.8071,
+      "step": 1079
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.795422670643109e-05,
+      "loss": 0.6666,
+      "step": 1080
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.795048913954936e-05,
+      "loss": 0.6455,
+      "step": 1081
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.79467483074964e-05,
+      "loss": 0.5934,
+      "step": 1082
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.794300421080441e-05,
+      "loss": 0.6914,
+      "step": 1083
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793925685000606e-05,
+      "loss": 0.7624,
+      "step": 1084
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7935506225634505e-05,
+      "loss": 0.6435,
+      "step": 1085
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.793175233822333e-05,
+      "loss": 0.6462,
+      "step": 1086
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7927995188306604e-05,
+      "loss": 0.9321,
+      "step": 1087
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792423477641887e-05,
+      "loss": 0.7784,
+      "step": 1088
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.792047110309511e-05,
+      "loss": 0.5327,
+      "step": 1089
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.791670416887078e-05,
+      "loss": 0.3704,
+      "step": 1090
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7912933974281816e-05,
+      "loss": 0.4567,
+      "step": 1091
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790916051986459e-05,
+      "loss": 0.9652,
+      "step": 1092
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790538380615597e-05,
+      "loss": 0.7261,
+      "step": 1093
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7901603833693246e-05,
+      "loss": 0.5886,
+      "step": 1094
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.789782060301422e-05,
+      "loss": 0.7725,
+      "step": 1095
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7894034114657116e-05,
+      "loss": 0.6617,
+      "step": 1096
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7890244369160646e-05,
+      "loss": 0.6202,
+      "step": 1097
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788645136706398e-05,
+      "loss": 0.617,
+      "step": 1098
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.788265510890674e-05,
+      "loss": 0.682,
+      "step": 1099
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787885559522902e-05,
+      "loss": 0.8794,
+      "step": 1100
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7875052826571396e-05,
+      "loss": 0.6658,
+      "step": 1101
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787124680347487e-05,
+      "loss": 0.6093,
+      "step": 1102
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786743752648094e-05,
+      "loss": 0.6968,
+      "step": 1103
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786362499613154e-05,
+      "loss": 0.9153,
+      "step": 1104
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7859809212969084e-05,
+      "loss": 0.5411,
+      "step": 1105
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785599017753645e-05,
+      "loss": 0.6071,
+      "step": 1106
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.785216789037695e-05,
+      "loss": 0.5453,
+      "step": 1107
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784834235203442e-05,
+      "loss": 0.7086,
+      "step": 1108
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.78445135630531e-05,
+      "loss": 1.0032,
+      "step": 1109
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784068152397769e-05,
+      "loss": 0.7769,
+      "step": 1110
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783684623535342e-05,
+      "loss": 0.5905,
+      "step": 1111
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.783300769772591e-05,
+      "loss": 0.8186,
+      "step": 1112
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782916591164126e-05,
+      "loss": 0.5257,
+      "step": 1113
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782532087764607e-05,
+      "loss": 0.6926,
+      "step": 1114
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.782147259628736e-05,
+      "loss": 0.6613,
+      "step": 1115
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.781762106811263e-05,
+      "loss": 0.7904,
+      "step": 1116
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7813766293669826e-05,
+      "loss": 0.6694,
+      "step": 1117
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780990827350738e-05,
+      "loss": 0.5861,
+      "step": 1118
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780604700817416e-05,
+      "loss": 0.5831,
+      "step": 1119
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.780218249821952e-05,
+      "loss": 0.5717,
+      "step": 1120
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7798314744193265e-05,
+      "loss": 0.7767,
+      "step": 1121
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779444374664566e-05,
+      "loss": 0.5549,
+      "step": 1122
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.779056950612744e-05,
+      "loss": 1.0558,
+      "step": 1123
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.778669202318977e-05,
+      "loss": 0.7928,
+      "step": 1124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.778281129838432e-05,
+      "loss": 0.502,
+      "step": 1125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.77789273322632e-05,
+      "loss": 1.0536,
+      "step": 1126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.777504012537898e-05,
+      "loss": 0.8107,
+      "step": 1127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7771149678284694e-05,
+      "loss": 0.8033,
+      "step": 1128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776725599153383e-05,
+      "loss": 0.7519,
+      "step": 1129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.776335906568036e-05,
+      "loss": 0.964,
+      "step": 1130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775945890127869e-05,
+      "loss": 0.4795,
+      "step": 1131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7755555498883694e-05,
+      "loss": 0.3049,
+      "step": 1132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.775164885905072e-05,
+      "loss": 0.524,
+      "step": 1133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774773898233555e-05,
+      "loss": 0.88,
+      "step": 1134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.774382586929447e-05,
+      "loss": 0.7567,
+      "step": 1135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773990952048417e-05,
+      "loss": 0.6722,
+      "step": 1136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773598993646185e-05,
+      "loss": 0.679,
+      "step": 1137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.773206711778514e-05,
+      "loss": 0.8442,
+      "step": 1138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7728141065012144e-05,
+      "loss": 0.8884,
+      "step": 1139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772421177870142e-05,
+      "loss": 0.6957,
+      "step": 1140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.772027925941199e-05,
+      "loss": 0.9318,
+      "step": 1141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.771634350770333e-05,
+      "loss": 0.6603,
+      "step": 1142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7712404524135394e-05,
+      "loss": 0.6724,
+      "step": 1143
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7708462309268577e-05,
+      "loss": 0.6721,
+      "step": 1144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770451686366372e-05,
+      "loss": 0.758,
+      "step": 1145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770056818788217e-05,
+      "loss": 0.5341,
+      "step": 1146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769661628248568e-05,
+      "loss": 0.855,
+      "step": 1147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7692661148036516e-05,
+      "loss": 0.5566,
+      "step": 1148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768870278509736e-05,
+      "loss": 0.6461,
+      "step": 1149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.768474119423136e-05,
+      "loss": 0.6934,
+      "step": 1150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7680776376002154e-05,
+      "loss": 0.6442,
+      "step": 1151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76768083309738e-05,
+      "loss": 0.6775,
+      "step": 1152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.767283705971084e-05,
+      "loss": 0.769,
+      "step": 1153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7668862562778275e-05,
+      "loss": 0.7871,
+      "step": 1154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.766488484074153e-05,
+      "loss": 0.5865,
+      "step": 1155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7660903894166554e-05,
+      "loss": 0.4693,
+      "step": 1156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.76569197236197e-05,
+      "loss": 0.6697,
+      "step": 1157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7652932329667784e-05,
+      "loss": 0.6929,
+      "step": 1158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764894171287812e-05,
+      "loss": 0.8931,
+      "step": 1159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764494787381842e-05,
+      "loss": 0.4177,
+      "step": 1160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.764095081305692e-05,
+      "loss": 0.66,
+      "step": 1161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7636950531162274e-05,
+      "loss": 0.5983,
+      "step": 1162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7632947028703586e-05,
+      "loss": 0.6295,
+      "step": 1163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762894030625047e-05,
+      "loss": 0.8698,
+      "step": 1164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762493036437292e-05,
+      "loss": 0.8301,
+      "step": 1165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.762091720364147e-05,
+      "loss": 0.6765,
+      "step": 1166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761690082462705e-05,
+      "loss": 0.6592,
+      "step": 1167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.761288122790107e-05,
+      "loss": 0.8286,
+      "step": 1168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.760885841403542e-05,
+      "loss": 0.6869,
+      "step": 1169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.760483238360241e-05,
+      "loss": 0.631,
+      "step": 1170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.760080313717482e-05,
+      "loss": 0.4208,
+      "step": 1171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.759677067532591e-05,
+      "loss": 0.7198,
+      "step": 1172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.759273499862936e-05,
+      "loss": 0.6752,
+      "step": 1173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7588696107659334e-05,
+      "loss": 0.5636,
+      "step": 1174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.758465400299045e-05,
+      "loss": 0.5551,
+      "step": 1175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.758060868519778e-05,
+      "loss": 0.7852,
+      "step": 1176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.757656015485684e-05,
+      "loss": 0.7349,
+      "step": 1177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.757250841254364e-05,
+      "loss": 0.4962,
+      "step": 1178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.756845345883459e-05,
+      "loss": 0.4189,
+      "step": 1179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.756439529430661e-05,
+      "loss": 0.549,
+      "step": 1180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.756033391953706e-05,
+      "loss": 0.3438,
+      "step": 1181
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.755626933510373e-05,
+      "loss": 0.5739,
+      "step": 1182
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.755220154158491e-05,
+      "loss": 0.645,
+      "step": 1183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.754813053955932e-05,
+      "loss": 0.6418,
+      "step": 1184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7544056329606145e-05,
+      "loss": 0.6409,
+      "step": 1185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.753997891230502e-05,
+      "loss": 0.7168,
+      "step": 1186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.753589828823603e-05,
+      "loss": 0.6671,
+      "step": 1187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.753181445797975e-05,
+      "loss": 0.8655,
+      "step": 1188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7527727422117165e-05,
+      "loss": 0.7421,
+      "step": 1189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.752363718122975e-05,
+      "loss": 0.579,
+      "step": 1190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.751954373589942e-05,
+      "loss": 0.6408,
+      "step": 1191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.751544708670855e-05,
+      "loss": 0.7715,
+      "step": 1192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7511347234239984e-05,
+      "loss": 0.6715,
+      "step": 1193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7507244179076994e-05,
+      "loss": 0.5479,
+      "step": 1194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7503137921803334e-05,
+      "loss": 0.5221,
+      "step": 1195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.749902846300318e-05,
+      "loss": 0.7899,
+      "step": 1196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.749491580326121e-05,
+      "loss": 0.5535,
+      "step": 1197
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.749079994316252e-05,
+      "loss": 0.8116,
+      "step": 1198
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.748668088329269e-05,
+      "loss": 0.6267,
+      "step": 1199
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.748255862423772e-05,
+      "loss": 0.7336,
+      "step": 1200
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.747843316658409e-05,
+      "loss": 0.6174,
+      "step": 1201
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.747430451091873e-05,
+      "loss": 0.716,
+      "step": 1202
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.747017265782904e-05,
+      "loss": 0.9127,
+      "step": 1203
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.746603760790284e-05,
+      "loss": 0.8685,
+      "step": 1204
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7461899361728426e-05,
+      "loss": 0.5474,
+      "step": 1205
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7457757919894566e-05,
+      "loss": 0.7048,
+      "step": 1206
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.745361328299044e-05,
+      "loss": 0.5651,
+      "step": 1207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744946545160571e-05,
+      "loss": 0.5425,
+      "step": 1208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744531442633051e-05,
+      "loss": 0.7089,
+      "step": 1209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744116020775539e-05,
+      "loss": 0.9113,
+      "step": 1210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7437002796471374e-05,
+      "loss": 0.7607,
+      "step": 1211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.743284219306994e-05,
+      "loss": 0.8519,
+      "step": 1212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7428678398143014e-05,
+      "loss": 0.7289,
+      "step": 1213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.742451141228298e-05,
+      "loss": 0.6663,
+      "step": 1214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7420341236082675e-05,
+      "loss": 0.6423,
+      "step": 1215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.74161678701354e-05,
+      "loss": 0.7262,
+      "step": 1216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.74119913150349e-05,
+      "loss": 0.7783,
+      "step": 1217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7407811571375365e-05,
+      "loss": 0.7089,
+      "step": 1218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7403628639751454e-05,
+      "loss": 0.7534,
+      "step": 1219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.739944252075827e-05,
+      "loss": 0.8123,
+      "step": 1220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.739525321499138e-05,
+      "loss": 0.6941,
+      "step": 1221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.739106072304679e-05,
+      "loss": 0.6698,
+      "step": 1222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.738686504552097e-05,
+      "loss": 0.6881,
+      "step": 1223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7382666183010836e-05,
+      "loss": 0.6744,
+      "step": 1224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7378464136113765e-05,
+      "loss": 0.5807,
+      "step": 1225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7374258905427585e-05,
+      "loss": 0.6897,
+      "step": 1226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.737005049155058e-05,
+      "loss": 0.5333,
+      "step": 1227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.736583889508146e-05,
+      "loss": 0.7845,
+      "step": 1228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.736162411661944e-05,
+      "loss": 0.4344,
+      "step": 1229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7357406156764126e-05,
+      "loss": 0.9208,
+      "step": 1230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.735318501611564e-05,
+      "loss": 0.8882,
+      "step": 1231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.73489606952745e-05,
+      "loss": 0.6583,
+      "step": 1232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.734473319484172e-05,
+      "loss": 0.7351,
+      "step": 1233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7340502515418726e-05,
+      "loss": 0.7664,
+      "step": 1234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7336268657607434e-05,
+      "loss": 0.6319,
+      "step": 1235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.733203162201019e-05,
+      "loss": 0.65,
+      "step": 1236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.73277914092298e-05,
+      "loss": 0.6128,
+      "step": 1237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7323548019869516e-05,
+      "loss": 0.5698,
+      "step": 1238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7319301454533054e-05,
+      "loss": 0.5978,
+      "step": 1239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7315051713824565e-05,
+      "loss": 0.8982,
+      "step": 1240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.731079879834867e-05,
+      "loss": 0.4486,
+      "step": 1241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.730654270871042e-05,
+      "loss": 1.0771,
+      "step": 1242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.730228344551534e-05,
+      "loss": 0.8818,
+      "step": 1243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7298021009369395e-05,
+      "loss": 0.7599,
+      "step": 1244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.729375540087899e-05,
+      "loss": 0.5064,
+      "step": 1245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.728948662065102e-05,
+      "loss": 0.6606,
+      "step": 1246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.728521466929278e-05,
+      "loss": 0.517,
+      "step": 1247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7280939547412054e-05,
+      "loss": 0.7681,
+      "step": 1248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.727666125561706e-05,
+      "loss": 0.7098,
+      "step": 1249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.727237979451647e-05,
+      "loss": 0.651,
+      "step": 1250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.726809516471943e-05,
+      "loss": 0.6165,
+      "step": 1251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.726380736683548e-05,
+      "loss": 0.6795,
+      "step": 1252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.725951640147467e-05,
+      "loss": 0.6748,
+      "step": 1253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.725522226924747e-05,
+      "loss": 0.7593,
+      "step": 1254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.72509249707648e-05,
+      "loss": 0.5613,
+      "step": 1255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.724662450663805e-05,
+      "loss": 0.7067,
+      "step": 1256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.724232087747904e-05,
+      "loss": 0.507,
+      "step": 1257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.723801408390005e-05,
+      "loss": 0.758,
+      "step": 1258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.723370412651381e-05,
+      "loss": 0.7633,
+      "step": 1259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.722939100593351e-05,
+      "loss": 0.8865,
+      "step": 1260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.722507472277275e-05,
+      "loss": 0.6813,
+      "step": 1261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7220755277645634e-05,
+      "loss": 0.6325,
+      "step": 1262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.721643267116668e-05,
+      "loss": 0.603,
+      "step": 1263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.721210690395086e-05,
+      "loss": 0.9398,
+      "step": 1264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.720777797661362e-05,
+      "loss": 0.5811,
+      "step": 1265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.720344588977083e-05,
+      "loss": 0.7057,
+      "step": 1266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7199110644038805e-05,
+      "loss": 0.6057,
+      "step": 1267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7194772240034325e-05,
+      "loss": 0.7222,
+      "step": 1268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.719043067837463e-05,
+      "loss": 0.6713,
+      "step": 1269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.718608595967738e-05,
+      "loss": 0.6286,
+      "step": 1270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.718173808456071e-05,
+      "loss": 0.7517,
+      "step": 1271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.71773870536432e-05,
+      "loss": 0.6717,
+      "step": 1272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.717303286754384e-05,
+      "loss": 0.3792,
+      "step": 1273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.716867552688213e-05,
+      "loss": 0.5794,
+      "step": 1274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7164315032277986e-05,
+      "loss": 0.6768,
+      "step": 1275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7159951384351766e-05,
+      "loss": 0.7549,
+      "step": 1276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.71555845837243e-05,
+      "loss": 0.6124,
+      "step": 1277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.715121463101684e-05,
+      "loss": 0.6754,
+      "step": 1278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7146841526851105e-05,
+      "loss": 0.7996,
+      "step": 1279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.714246527184927e-05,
+      "loss": 0.7941,
+      "step": 1280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.713808586663392e-05,
+      "loss": 0.6092,
+      "step": 1281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.713370331182814e-05,
+      "loss": 0.659,
+      "step": 1282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7129317608055425e-05,
+      "loss": 0.6354,
+      "step": 1283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.712492875593973e-05,
+      "loss": 0.9788,
+      "step": 1284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.712053675610546e-05,
+      "loss": 0.8857,
+      "step": 1285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.711614160917747e-05,
+      "loss": 0.9481,
+      "step": 1286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.711174331578104e-05,
+      "loss": 0.4459,
+      "step": 1287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7107341876541936e-05,
+      "loss": 0.6326,
+      "step": 1288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.710293729208635e-05,
+      "loss": 0.5737,
+      "step": 1289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7098529563040915e-05,
+      "loss": 0.6678,
+      "step": 1290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.709411869003273e-05,
+      "loss": 0.7448,
+      "step": 1291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7089704673689317e-05,
+      "loss": 0.701,
+      "step": 1292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7085287514638664e-05,
+      "loss": 0.6943,
+      "step": 1293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.708086721350921e-05,
+      "loss": 0.65,
+      "step": 1294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.707644377092982e-05,
+      "loss": 0.6895,
+      "step": 1295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7072017187529834e-05,
+      "loss": 0.6257,
+      "step": 1296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.706758746393901e-05,
+      "loss": 0.9252,
+      "step": 1297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.706315460078756e-05,
+      "loss": 0.6838,
+      "step": 1298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7058718598706163e-05,
+      "loss": 0.9006,
+      "step": 1299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.705427945832592e-05,
+      "loss": 0.5074,
+      "step": 1300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.70498371802784e-05,
+      "loss": 0.6118,
+      "step": 1301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.70453917651956e-05,
+      "loss": 0.8058,
+      "step": 1302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7040943213709967e-05,
+      "loss": 0.5238,
+      "step": 1303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7036491526454394e-05,
+      "loss": 0.7684,
+      "step": 1304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.703203670406224e-05,
+      "loss": 0.7378,
+      "step": 1305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.702757874716728e-05,
+      "loss": 0.5126,
+      "step": 1306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.702311765640375e-05,
+      "loss": 0.7529,
+      "step": 1307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.701865343240633e-05,
+      "loss": 0.7552,
+      "step": 1308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.701418607581015e-05,
+      "loss": 0.6956,
+      "step": 1309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.700971558725078e-05,
+      "loss": 0.714,
+      "step": 1310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.700524196736424e-05,
+      "loss": 0.5204,
+      "step": 1311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7000765216786993e-05,
+      "loss": 0.7139,
+      "step": 1312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.699628533615593e-05,
+      "loss": 0.5033,
+      "step": 1313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6991802326108435e-05,
+      "loss": 0.8564,
+      "step": 1314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6987316187282285e-05,
+      "loss": 0.7187,
+      "step": 1315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6982826920315735e-05,
+      "loss": 0.8033,
+      "step": 1316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.697833452584747e-05,
+      "loss": 0.5696,
+      "step": 1317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.697383900451662e-05,
+      "loss": 0.8394,
+      "step": 1318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6969340356962764e-05,
+      "loss": 0.748,
+      "step": 1319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6964838583825935e-05,
+      "loss": 0.7205,
+      "step": 1320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.696033368574659e-05,
+      "loss": 0.7855,
+      "step": 1321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.695582566336564e-05,
+      "loss": 0.7654,
+      "step": 1322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.695131451732446e-05,
+      "loss": 0.6219,
+      "step": 1323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6946800248264844e-05,
+      "loss": 0.8229,
+      "step": 1324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.694228285682902e-05,
+      "loss": 0.9111,
+      "step": 1325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.693776234365971e-05,
+      "loss": 0.8739,
+      "step": 1326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6933238709400024e-05,
+      "loss": 0.7662,
+      "step": 1327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.692871195469355e-05,
+      "loss": 0.7626,
+      "step": 1328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6924182080184304e-05,
+      "loss": 0.9218,
+      "step": 1329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6919649086516756e-05,
+      "loss": 0.7755,
+      "step": 1330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6915112974335826e-05,
+      "loss": 0.7178,
+      "step": 1331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.691057374428686e-05,
+      "loss": 0.6621,
+      "step": 1332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6906031397015644e-05,
+      "loss": 0.6426,
+      "step": 1333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.690148593316843e-05,
+      "loss": 0.6976,
+      "step": 1334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.68969373533919e-05,
+      "loss": 0.7459,
+      "step": 1335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.689238565833319e-05,
+      "loss": 0.7935,
+      "step": 1336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.688783084863986e-05,
+      "loss": 0.5349,
+      "step": 1337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.688327292495992e-05,
+      "loss": 0.6231,
+      "step": 1338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.687871188794184e-05,
+      "loss": 0.7609,
+      "step": 1339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.687414773823451e-05,
+      "loss": 0.6069,
+      "step": 1340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.686958047648728e-05,
+      "loss": 0.6483,
+      "step": 1341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6865010103349926e-05,
+      "loss": 0.6083,
+      "step": 1342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.686043661947269e-05,
+      "loss": 0.6487,
+      "step": 1343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6855860025506226e-05,
+      "loss": 0.7141,
+      "step": 1344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6851280322101654e-05,
+      "loss": 0.6545,
+      "step": 1345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.684669750991053e-05,
+      "loss": 0.4736,
+      "step": 1346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.684211158958486e-05,
+      "loss": 0.5726,
+      "step": 1347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.683752256177707e-05,
+      "loss": 0.8787,
+      "step": 1348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.683293042714005e-05,
+      "loss": 0.7256,
+      "step": 1349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.682833518632712e-05,
+      "loss": 0.713,
+      "step": 1350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.682373683999205e-05,
+      "loss": 0.6577,
+      "step": 1351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.681913538878905e-05,
+      "loss": 0.5755,
+      "step": 1352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.681453083337276e-05,
+      "loss": 0.5442,
+      "step": 1353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.680992317439827e-05,
+      "loss": 0.7424,
+      "step": 1354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.680531241252113e-05,
+      "loss": 0.6743,
+      "step": 1355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6800698548397294e-05,
+      "loss": 0.5635,
+      "step": 1356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.679608158268319e-05,
+      "loss": 0.8035,
+      "step": 1357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6791461516035674e-05,
+      "loss": 0.6578,
+      "step": 1358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.678683834911204e-05,
+      "loss": 0.6973,
+      "step": 1359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.678221208257003e-05,
+      "loss": 1.1077,
+      "step": 1360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6777582717067825e-05,
+      "loss": 0.7316,
+      "step": 1361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6772950253264035e-05,
+      "loss": 0.7428,
+      "step": 1362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.676831469181773e-05,
+      "loss": 0.7462,
+      "step": 1363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6763676033388414e-05,
+      "loss": 0.7755,
+      "step": 1364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.675903427863604e-05,
+      "loss": 0.9293,
+      "step": 1365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.675438942822096e-05,
+      "loss": 0.7792,
+      "step": 1366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.674974148280403e-05,
+      "loss": 0.853,
+      "step": 1367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.67450904430465e-05,
+      "loss": 0.8734,
+      "step": 1368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6740436309610074e-05,
+      "loss": 0.5168,
+      "step": 1369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.67357790831569e-05,
+      "loss": 0.884,
+      "step": 1370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6731118764349555e-05,
+      "loss": 0.4589,
+      "step": 1371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.672645535385108e-05,
+      "loss": 0.6912,
+      "step": 1372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.672178885232493e-05,
+      "loss": 0.7494,
+      "step": 1373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.6717119260435003e-05,
+      "loss": 0.5574,
+      "step": 1374
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.671244657884565e-05,
+      "loss": 0.7382,
+      "step": 1375
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.670777080822165e-05,
+      "loss": 0.5829,
+      "step": 1376
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.670309194922824e-05,
+      "loss": 0.5378,
+      "step": 1377
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.669841000253106e-05,
+      "loss": 0.8612,
+      "step": 1378
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6693724968796226e-05,
+      "loss": 0.695,
+      "step": 1379
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.668903684869028e-05,
+      "loss": 0.6323,
+      "step": 1380
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.668434564288019e-05,
+      "loss": 0.6141,
+      "step": 1381
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.667965135203339e-05,
+      "loss": 0.4308,
+      "step": 1382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.667495397681773e-05,
+      "loss": 0.5299,
+      "step": 1383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.66702535179015e-05,
+      "loss": 0.7576,
+      "step": 1384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.666554997595345e-05,
+      "loss": 0.7444,
+      "step": 1385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.666084335164274e-05,
+      "loss": 0.7392,
+      "step": 1386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.665613364563899e-05,
+      "loss": 0.7404,
+      "step": 1387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6651420858612254e-05,
+      "loss": 0.7635,
+      "step": 1388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6646704991233015e-05,
+      "loss": 0.6133,
+      "step": 1389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6641986044172206e-05,
+      "loss": 0.5577,
+      "step": 1390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.663726401810118e-05,
+      "loss": 0.6967,
+      "step": 1391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.663253891369176e-05,
+      "loss": 0.6983,
+      "step": 1392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.662781073161617e-05,
+      "loss": 0.5395,
+      "step": 1393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.66230794725471e-05,
+      "loss": 0.5702,
+      "step": 1394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6618345137157664e-05,
+      "loss": 0.5723,
+      "step": 1395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6613607726121416e-05,
+      "loss": 0.6553,
+      "step": 1396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.660886724011235e-05,
+      "loss": 0.5268,
+      "step": 1397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.66041236798049e-05,
+      "loss": 0.637,
+      "step": 1398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6599377045873926e-05,
+      "loss": 0.5437,
+      "step": 1399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.659462733899474e-05,
+      "loss": 0.7454,
+      "step": 1400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6589874559843074e-05,
+      "loss": 0.7118,
+      "step": 1401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6585118709095114e-05,
+      "loss": 0.7791,
+      "step": 1402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6580359787427476e-05,
+      "loss": 0.6466,
+      "step": 1403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.657559779551721e-05,
+      "loss": 0.7675,
+      "step": 1404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6570832734041814e-05,
+      "loss": 0.6442,
+      "step": 1405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.65660646036792e-05,
+      "loss": 0.5355,
+      "step": 1406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.656129340510774e-05,
+      "loss": 0.8394,
+      "step": 1407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.655651913900623e-05,
+      "loss": 0.816,
+      "step": 1408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6551741806053916e-05,
+      "loss": 0.4254,
+      "step": 1409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.654696140693046e-05,
+      "loss": 0.8406,
+      "step": 1410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.654217794231597e-05,
+      "loss": 0.8754,
+      "step": 1411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.653739141289098e-05,
+      "loss": 0.7262,
+      "step": 1412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6532601819336506e-05,
+      "loss": 0.6818,
+      "step": 1413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6527809162333935e-05,
+      "loss": 0.772,
+      "step": 1414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6523013442565124e-05,
+      "loss": 0.6388,
+      "step": 1415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.651821466071236e-05,
+      "loss": 0.7134,
+      "step": 1416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.651341281745838e-05,
+      "loss": 0.5523,
+      "step": 1417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6508607913486324e-05,
+      "loss": 0.5181,
+      "step": 1418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.650379994947981e-05,
+      "loss": 0.5667,
+      "step": 1419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.649898892612284e-05,
+      "loss": 0.6675,
+      "step": 1420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6494174844099904e-05,
+      "loss": 0.5371,
+      "step": 1421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.648935770409589e-05,
+      "loss": 0.6185,
+      "step": 1422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6484537506796136e-05,
+      "loss": 0.6541,
+      "step": 1423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.647971425288641e-05,
+      "loss": 0.694,
+      "step": 1424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.647488794305292e-05,
+      "loss": 0.6806,
+      "step": 1425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6470058577982316e-05,
+      "loss": 0.7644,
+      "step": 1426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.646522615836166e-05,
+      "loss": 0.5704,
+      "step": 1427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.646039068487846e-05,
+      "loss": 0.8459,
+      "step": 1428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6455552158220664e-05,
+      "loss": 0.6434,
+      "step": 1429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.645071057907666e-05,
+      "loss": 0.6536,
+      "step": 1430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.644586594813524e-05,
+      "loss": 0.7014,
+      "step": 1431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.644101826608567e-05,
+      "loss": 0.86,
+      "step": 1432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.643616753361762e-05,
+      "loss": 0.7685,
+      "step": 1433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6431313751421216e-05,
+      "loss": 0.6172,
+      "step": 1434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.642645692018699e-05,
+      "loss": 0.5988,
+      "step": 1435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.642159704060593e-05,
+      "loss": 0.6705,
+      "step": 1436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6416734113369456e-05,
+      "loss": 0.4987,
+      "step": 1437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6411868139169425e-05,
+      "loss": 0.6448,
+      "step": 1438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6406999118698103e-05,
+      "loss": 0.807,
+      "step": 1439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.640212705264823e-05,
+      "loss": 0.6602,
+      "step": 1440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6397251941712924e-05,
+      "loss": 0.5302,
+      "step": 1441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6392373786585796e-05,
+      "loss": 0.838,
+      "step": 1442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.638749258796085e-05,
+      "loss": 0.8935,
+      "step": 1443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6382608346532536e-05,
+      "loss": 0.5974,
+      "step": 1444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6377721062995737e-05,
+      "loss": 0.6163,
+      "step": 1445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6372830738045766e-05,
+      "loss": 0.5242,
+      "step": 1446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.636793737237838e-05,
+      "loss": 0.5079,
+      "step": 1447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6363040966689754e-05,
+      "loss": 0.6135,
+      "step": 1448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.635814152167649e-05,
+      "loss": 0.5724,
+      "step": 1449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6353239038035644e-05,
+      "loss": 0.4623,
+      "step": 1450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.6348333516464694e-05,
+      "loss": 0.7189,
+      "step": 1451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.634342495766155e-05,
+      "loss": 0.6373,
+      "step": 1452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.633851336232454e-05,
+      "loss": 0.8038,
+      "step": 1453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.633359873115246e-05,
+      "loss": 0.4748,
+      "step": 1454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.63286810648445e-05,
+      "loss": 0.5817,
+      "step": 1455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.63237603641003e-05,
+      "loss": 0.7524,
+      "step": 1456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.631883662961993e-05,
+      "loss": 0.6829,
+      "step": 1457
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.63139098621039e-05,
+      "loss": 0.6277,
+      "step": 1458
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6308980062253125e-05,
+      "loss": 0.676,
+      "step": 1459
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.630404723076899e-05,
+      "loss": 0.795,
+      "step": 1460
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6299111368353265e-05,
+      "loss": 0.3866,
+      "step": 1461
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6294172475708196e-05,
+      "loss": 0.7448,
+      "step": 1462
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.628923055353643e-05,
+      "loss": 0.5144,
+      "step": 1463
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.628428560254107e-05,
+      "loss": 0.4974,
+      "step": 1464
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.627933762342561e-05,
+      "loss": 0.688,
+      "step": 1465
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6274386616894015e-05,
+      "loss": 0.5644,
+      "step": 1466
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.626943258365068e-05,
+      "loss": 0.464,
+      "step": 1467
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6264475524400384e-05,
+      "loss": 0.6967,
+      "step": 1468
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.62595154398484e-05,
+      "loss": 0.7133,
+      "step": 1469
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6254552330700377e-05,
+      "loss": 0.7783,
+      "step": 1470
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.624958619766243e-05,
+      "loss": 0.7644,
+      "step": 1471
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.624461704144109e-05,
+      "loss": 0.9272,
+      "step": 1472
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.623964486274332e-05,
+      "loss": 0.7384,
+      "step": 1473
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6234669662276516e-05,
+      "loss": 0.4962,
+      "step": 1474
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.622969144074849e-05,
+      "loss": 0.6389,
+      "step": 1475
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6224710198867505e-05,
+      "loss": 0.7794,
+      "step": 1476
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.621972593734224e-05,
+      "loss": 0.7468,
+      "step": 1477
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.62147386568818e-05,
+      "loss": 0.5986,
+      "step": 1478
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.620974835819575e-05,
+      "loss": 0.5773,
+      "step": 1479
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.620475504199404e-05,
+      "loss": 0.8043,
+      "step": 1480
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.619975870898707e-05,
+      "loss": 0.7991,
+      "step": 1481
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.619475935988567e-05,
+      "loss": 1.0215,
+      "step": 1482
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.618975699540111e-05,
+      "loss": 0.8391,
+      "step": 1483
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.618475161624507e-05,
+      "loss": 0.6588,
+      "step": 1484
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6179743223129666e-05,
+      "loss": 0.5671,
+      "step": 1485
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6174731816767445e-05,
+      "loss": 0.9575,
+      "step": 1486
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.616971739787138e-05,
+      "loss": 0.722,
+      "step": 1487
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.616469996715487e-05,
+      "loss": 0.8823,
+      "step": 1488
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.615967952533175e-05,
+      "loss": 0.8099,
+      "step": 1489
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6154656073116284e-05,
+      "loss": 0.8604,
+      "step": 1490
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6149629611223144e-05,
+      "loss": 0.611,
+      "step": 1491
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6144600140367466e-05,
+      "loss": 0.5627,
+      "step": 1492
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.613956766126478e-05,
+      "loss": 0.603,
+      "step": 1493
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.613453217463106e-05,
+      "loss": 0.4786,
+      "step": 1494
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.612949368118272e-05,
+      "loss": 0.5933,
+      "step": 1495
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6124452181636563e-05,
+      "loss": 0.5632,
+      "step": 1496
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.611940767670986e-05,
+      "loss": 0.5812,
+      "step": 1497
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6114360167120296e-05,
+      "loss": 0.4209,
+      "step": 1498
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.610930965358597e-05,
+      "loss": 0.6282,
+      "step": 1499
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.610425613682543e-05,
+      "loss": 0.507,
+      "step": 1500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.609919961755763e-05,
+      "loss": 0.8252,
+      "step": 1501
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6094140096501975e-05,
+      "loss": 0.6904,
+      "step": 1502
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.608907757437827e-05,
+      "loss": 0.6876,
+      "step": 1503
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6084012051906776e-05,
+      "loss": 0.9312,
+      "step": 1504
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.607894352980816e-05,
+      "loss": 0.6704,
+      "step": 1505
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.607387200880352e-05,
+      "loss": 0.4978,
+      "step": 1506
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6068797489614385e-05,
+      "loss": 0.6631,
+      "step": 1507
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.606371997296271e-05,
+      "loss": 0.8397,
+      "step": 1508
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.605863945957086e-05,
+      "loss": 0.6969,
+      "step": 1509
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6053555950161654e-05,
+      "loss": 0.3837,
+      "step": 1510
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.604846944545833e-05,
+      "loss": 0.3978,
+      "step": 1511
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6043379946184526e-05,
+      "loss": 0.9911,
+      "step": 1512
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6038287453064344e-05,
+      "loss": 0.4642,
+      "step": 1513
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6033191966822285e-05,
+      "loss": 0.6583,
+      "step": 1514
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6028093488183294e-05,
+      "loss": 0.6005,
+      "step": 1515
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.602299201787272e-05,
+      "loss": 0.375,
+      "step": 1516
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.601788755661636e-05,
+      "loss": 0.5919,
+      "step": 1517
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.601278010514043e-05,
+      "loss": 0.7737,
+      "step": 1518
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.600766966417156e-05,
+      "loss": 0.8375,
+      "step": 1519
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.600255623443681e-05,
+      "loss": 0.6005,
+      "step": 1520
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.599743981666368e-05,
+      "loss": 0.6231,
+      "step": 1521
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.599232041158007e-05,
+      "loss": 0.7734,
+      "step": 1522
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5987198019914334e-05,
+      "loss": 0.9153,
+      "step": 1523
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5982072642395233e-05,
+      "loss": 0.5269,
+      "step": 1524
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.597694427975194e-05,
+      "loss": 0.8812,
+      "step": 1525
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.597181293271409e-05,
+      "loss": 0.7158,
+      "step": 1526
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5966678602011714e-05,
+      "loss": 0.7789,
+      "step": 1527
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.596154128837526e-05,
+      "loss": 0.7281,
+      "step": 1528
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.595640099253564e-05,
+      "loss": 0.5745,
+      "step": 1529
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.595125771522414e-05,
+      "loss": 0.7034,
+      "step": 1530
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5946111457172515e-05,
+      "loss": 0.4681,
+      "step": 1531
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.594096221911291e-05,
+      "loss": 0.8019,
+      "step": 1532
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.593581000177791e-05,
+      "loss": 0.9843,
+      "step": 1533
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.593065480590054e-05,
+      "loss": 0.5473,
+      "step": 1534
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5925496632214206e-05,
+      "loss": 0.6478,
+      "step": 1535
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.592033548145277e-05,
+      "loss": 0.6273,
+      "step": 1536
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5915171354350526e-05,
+      "loss": 0.7332,
+      "step": 1537
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5910004251642155e-05,
+      "loss": 0.6866,
+      "step": 1538
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.590483417406279e-05,
+      "loss": 0.5759,
+      "step": 1539
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.5899661122347984e-05,
+      "loss": 0.8019,
+      "step": 1540
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5894485097233695e-05,
+      "loss": 0.7432,
+      "step": 1541
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.588930609945633e-05,
+      "loss": 0.6985,
+      "step": 1542
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5884124129752704e-05,
+      "loss": 0.4458,
+      "step": 1543
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.587893918886005e-05,
+      "loss": 0.917,
+      "step": 1544
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5873751277516034e-05,
+      "loss": 0.5372,
+      "step": 1545
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5868560396458745e-05,
+      "loss": 0.6626,
+      "step": 1546
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.586336654642669e-05,
+      "loss": 0.5729,
+      "step": 1547
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.585816972815881e-05,
+      "loss": 0.5745,
+      "step": 1548
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.585296994239442e-05,
+      "loss": 0.7286,
+      "step": 1549
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.584776718987334e-05,
+      "loss": 0.566,
+      "step": 1550
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5842561471335746e-05,
+      "loss": 0.6975,
+      "step": 1551
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.583735278752225e-05,
+      "loss": 0.849,
+      "step": 1552
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.58321411391739e-05,
+      "loss": 0.606,
+      "step": 1553
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.582692652703216e-05,
+      "loss": 0.7624,
+      "step": 1554
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5821708951838906e-05,
+      "loss": 0.4542,
+      "step": 1555
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.581648841433646e-05,
+      "loss": 0.8147,
+      "step": 1556
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.581126491526753e-05,
+      "loss": 0.6502,
+      "step": 1557
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5806038455375286e-05,
+      "loss": 0.691,
+      "step": 1558
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5800809035403275e-05,
+      "loss": 0.8172,
+      "step": 1559
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.57955766560955e-05,
+      "loss": 0.7943,
+      "step": 1560
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.579034131819637e-05,
+      "loss": 0.5534,
+      "step": 1561
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.578510302245073e-05,
+      "loss": 0.6009,
+      "step": 1562
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.577986176960381e-05,
+      "loss": 0.793,
+      "step": 1563
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.577461756040129e-05,
+      "loss": 0.5701,
+      "step": 1564
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.576937039558928e-05,
+      "loss": 0.8097,
+      "step": 1565
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.576412027591429e-05,
+      "loss": 0.9147,
+      "step": 1566
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.575886720212325e-05,
+      "loss": 0.6221,
+      "step": 1567
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5753611174963504e-05,
+      "loss": 0.6419,
+      "step": 1568
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.574835219518285e-05,
+      "loss": 0.8004,
+      "step": 1569
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5743090263529485e-05,
+      "loss": 0.548,
+      "step": 1570
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5737825380752006e-05,
+      "loss": 0.9083,
+      "step": 1571
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.573255754759946e-05,
+      "loss": 0.6525,
+      "step": 1572
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.572728676482131e-05,
+      "loss": 0.6254,
+      "step": 1573
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.572201303316742e-05,
+      "loss": 0.6964,
+      "step": 1574
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.571673635338809e-05,
+      "loss": 0.579,
+      "step": 1575
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.571145672623404e-05,
+      "loss": 0.8347,
+      "step": 1576
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5706174152456385e-05,
+      "loss": 0.6664,
+      "step": 1577
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5700888632806696e-05,
+      "loss": 0.8953,
+      "step": 1578
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5695600168036934e-05,
+      "loss": 0.6318,
+      "step": 1579
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.56903087588995e-05,
+      "loss": 0.5722,
+      "step": 1580
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5685014406147194e-05,
+      "loss": 0.5569,
+      "step": 1581
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5679717110533255e-05,
+      "loss": 0.7635,
+      "step": 1582
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.567441687281132e-05,
+      "loss": 0.9044,
+      "step": 1583
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.566911369373546e-05,
+      "loss": 0.5775,
+      "step": 1584
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.566380757406017e-05,
+      "loss": 0.8225,
+      "step": 1585
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.565849851454033e-05,
+      "loss": 0.6694,
+      "step": 1586
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.565318651593128e-05,
+      "loss": 0.5338,
+      "step": 1587
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5647871578988756e-05,
+      "loss": 0.6374,
+      "step": 1588
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.564255370446891e-05,
+      "loss": 0.6207,
+      "step": 1589
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.563723289312831e-05,
+      "loss": 0.6323,
+      "step": 1590
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5631909145723966e-05,
+      "loss": 0.8093,
+      "step": 1591
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.562658246301328e-05,
+      "loss": 0.6129,
+      "step": 1592
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.562125284575408e-05,
+      "loss": 0.537,
+      "step": 1593
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5615920294704615e-05,
+      "loss": 0.7059,
+      "step": 1594
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5610584810623544e-05,
+      "loss": 0.6946,
+      "step": 1595
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5605246394269965e-05,
+      "loss": 0.7323,
+      "step": 1596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.559990504640335e-05,
+      "loss": 0.6987,
+      "step": 1597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.559456076778361e-05,
+      "loss": 0.7454,
+      "step": 1598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.558921355917111e-05,
+      "loss": 0.8591,
+      "step": 1599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.558386342132657e-05,
+      "loss": 0.8541,
+      "step": 1600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.557851035501117e-05,
+      "loss": 0.5609,
+      "step": 1601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.557315436098649e-05,
+      "loss": 0.6068,
+      "step": 1602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5567795440014514e-05,
+      "loss": 0.6119,
+      "step": 1603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.556243359285768e-05,
+      "loss": 0.7445,
+      "step": 1604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5557068820278806e-05,
+      "loss": 0.5891,
+      "step": 1605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.555170112304113e-05,
+      "loss": 0.3894,
+      "step": 1606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.554633050190833e-05,
+      "loss": 0.8228,
+      "step": 1607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5540956957644496e-05,
+      "loss": 0.6411,
+      "step": 1608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.553558049101409e-05,
+      "loss": 0.5952,
+      "step": 1609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.553020110278206e-05,
+      "loss": 0.6755,
+      "step": 1610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.552481879371371e-05,
+      "loss": 0.7279,
+      "step": 1611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5519433564574775e-05,
+      "loss": 0.787,
+      "step": 1612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.551404541613143e-05,
+      "loss": 0.9438,
+      "step": 1613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5508654349150255e-05,
+      "loss": 0.7314,
+      "step": 1614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5503260364398215e-05,
+      "loss": 0.8096,
+      "step": 1615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.549786346264273e-05,
+      "loss": 0.5398,
+      "step": 1616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.549246364465162e-05,
+      "loss": 0.5594,
+      "step": 1617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.54870609111931e-05,
+      "loss": 0.584,
+      "step": 1618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.548165526303584e-05,
+      "loss": 0.7228,
+      "step": 1619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.5476246700948884e-05,
+      "loss": 0.6157,
+      "step": 1620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.547083522570172e-05,
+      "loss": 0.7935,
+      "step": 1621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.546542083806424e-05,
+      "loss": 0.7568,
+      "step": 1622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.546000353880674e-05,
+      "loss": 0.7155,
+      "step": 1623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.545458332869996e-05,
+      "loss": 0.5059,
+      "step": 1624
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.544916020851502e-05,
+      "loss": 0.6893,
+      "step": 1625
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.544373417902347e-05,
+      "loss": 0.7713,
+      "step": 1626
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.543830524099727e-05,
+      "loss": 0.4998,
+      "step": 1627
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5432873395208805e-05,
+      "loss": 0.6061,
+      "step": 1628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5427438642430853e-05,
+      "loss": 0.5527,
+      "step": 1629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.542200098343663e-05,
+      "loss": 0.4129,
+      "step": 1630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.541656041899975e-05,
+      "loss": 0.6783,
+      "step": 1631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5411116949894236e-05,
+      "loss": 0.4666,
+      "step": 1632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.540567057689453e-05,
+      "loss": 0.6091,
+      "step": 1633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.540022130077551e-05,
+      "loss": 0.7906,
+      "step": 1634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.539476912231242e-05,
+      "loss": 0.9504,
+      "step": 1635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.538931404228096e-05,
+      "loss": 0.8719,
+      "step": 1636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.538385606145721e-05,
+      "loss": 0.4401,
+      "step": 1637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.537839518061769e-05,
+      "loss": 0.8033,
+      "step": 1638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.537293140053933e-05,
+      "loss": 0.5386,
+      "step": 1639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5367464721999436e-05,
+      "loss": 0.6274,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.536199514577577e-05,
+      "loss": 0.8271,
+      "step": 1641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5356522672646494e-05,
+      "loss": 0.7202,
+      "step": 1642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.535104730339017e-05,
+      "loss": 0.5802,
+      "step": 1643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5345569038785785e-05,
+      "loss": 0.7005,
+      "step": 1644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.534008787961272e-05,
+      "loss": 0.582,
+      "step": 1645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5334603826650806e-05,
+      "loss": 0.5227,
+      "step": 1646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5329116880680237e-05,
+      "loss": 0.7233,
+      "step": 1647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.532362704248165e-05,
+      "loss": 0.8467,
+      "step": 1648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.531813431283609e-05,
+      "loss": 0.5154,
+      "step": 1649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5312638692525e-05,
+      "loss": 0.8185,
+      "step": 1650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5307140182330254e-05,
+      "loss": 0.6892,
+      "step": 1651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.530163878303412e-05,
+      "loss": 0.7693,
+      "step": 1652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.529613449541927e-05,
+      "loss": 0.5982,
+      "step": 1653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.529062732026883e-05,
+      "loss": 0.7859,
+      "step": 1654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.528511725836629e-05,
+      "loss": 0.3699,
+      "step": 1655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.527960431049556e-05,
+      "loss": 0.7009,
+      "step": 1656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.527408847744098e-05,
+      "loss": 0.576,
+      "step": 1657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.526856975998729e-05,
+      "loss": 0.8167,
+      "step": 1658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.526304815891963e-05,
+      "loss": 0.574,
+      "step": 1659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.525752367502357e-05,
+      "loss": 0.6977,
+      "step": 1660
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.525199630908506e-05,
+      "loss": 0.6232,
+      "step": 1661
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.524646606189051e-05,
+      "loss": 0.6609,
+      "step": 1662
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5240932934226685e-05,
+      "loss": 0.5819,
+      "step": 1663
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.523539692688079e-05,
+      "loss": 0.5919,
+      "step": 1664
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.522985804064044e-05,
+      "loss": 0.6293,
+      "step": 1665
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5224316276293654e-05,
+      "loss": 0.7125,
+      "step": 1666
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.521877163462885e-05,
+      "loss": 0.473,
+      "step": 1667
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.521322411643487e-05,
+      "loss": 0.5368,
+      "step": 1668
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.520767372250097e-05,
+      "loss": 0.7077,
+      "step": 1669
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.520212045361679e-05,
+      "loss": 0.5919,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.519656431057241e-05,
+      "loss": 0.4105,
+      "step": 1671
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5191005294158295e-05,
+      "loss": 0.5809,
+      "step": 1672
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.518544340516532e-05,
+      "loss": 0.6062,
+      "step": 1673
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.517987864438479e-05,
+      "loss": 0.7253,
+      "step": 1674
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5174311012608406e-05,
+      "loss": 0.7006,
+      "step": 1675
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5168740510628265e-05,
+      "loss": 0.5456,
+      "step": 1676
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5163167139236886e-05,
+      "loss": 0.7909,
+      "step": 1677
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.51575908992272e-05,
+      "loss": 0.4951,
+      "step": 1678
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.515201179139253e-05,
+      "loss": 0.5418,
+      "step": 1679
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5146429816526633e-05,
+      "loss": 0.562,
+      "step": 1680
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.514084497542365e-05,
+      "loss": 0.6979,
+      "step": 1681
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.513525726887813e-05,
+      "loss": 0.9518,
+      "step": 1682
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5129666697685036e-05,
+      "loss": 0.674,
+      "step": 1683
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.512407326263975e-05,
+      "loss": 0.96,
+      "step": 1684
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.511847696453806e-05,
+      "loss": 0.472,
+      "step": 1685
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5112877804176126e-05,
+      "loss": 0.7922,
+      "step": 1686
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.510727578235057e-05,
+      "loss": 0.6031,
+      "step": 1687
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.510167089985837e-05,
+      "loss": 0.4982,
+      "step": 1688
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.509606315749694e-05,
+      "loss": 0.5701,
+      "step": 1689
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5090452556064106e-05,
+      "loss": 0.7123,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.508483909635808e-05,
+      "loss": 0.8124,
+      "step": 1691
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.507922277917749e-05,
+      "loss": 0.7197,
+      "step": 1692
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5073603605321365e-05,
+      "loss": 0.4172,
+      "step": 1693
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.506798157558917e-05,
+      "loss": 0.5339,
+      "step": 1694
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.506235669078073e-05,
+      "loss": 0.5829,
+      "step": 1695
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.505672895169629e-05,
+      "loss": 0.5837,
+      "step": 1696
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.505109835913654e-05,
+      "loss": 0.8347,
+      "step": 1697
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.504546491390252e-05,
+      "loss": 0.5543,
+      "step": 1698
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.503982861679572e-05,
+      "loss": 0.6498,
+      "step": 1699
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5034189468618e-05,
+      "loss": 0.6233,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5028547470171664e-05,
+      "loss": 0.8804,
+      "step": 1701
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.5022902622259376e-05,
+      "loss": 0.687,
+      "step": 1702
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.501725492568425e-05,
+      "loss": 0.6409,
+      "step": 1703
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.501160438124978e-05,
+      "loss": 0.9021,
+      "step": 1704
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.500595098975986e-05,
+      "loss": 0.4749,
+      "step": 1705
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.50002947520188e-05,
+      "loss": 0.8359,
+      "step": 1706
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.499463566883133e-05,
+      "loss": 0.9551,
+      "step": 1707
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.498897374100256e-05,
+      "loss": 0.7059,
+      "step": 1708
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.498330896933801e-05,
+      "loss": 0.7709,
+      "step": 1709
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.497764135464361e-05,
+      "loss": 0.7242,
+      "step": 1710
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4971970897725694e-05,
+      "loss": 0.6089,
+      "step": 1711
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4966297599391e-05,
+      "loss": 0.9591,
+      "step": 1712
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.496062146044667e-05,
+      "loss": 0.6974,
+      "step": 1713
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.495494248170025e-05,
+      "loss": 0.6112,
+      "step": 1714
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.494926066395969e-05,
+      "loss": 0.8669,
+      "step": 1715
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.494357600803334e-05,
+      "loss": 0.5439,
+      "step": 1716
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.493788851472995e-05,
+      "loss": 0.4941,
+      "step": 1717
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4932198184858704e-05,
+      "loss": 0.6005,
+      "step": 1718
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.492650501922915e-05,
+      "loss": 0.6552,
+      "step": 1719
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.492080901865125e-05,
+      "loss": 0.5281,
+      "step": 1720
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.491511018393539e-05,
+      "loss": 0.8939,
+      "step": 1721
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.490940851589234e-05,
+      "loss": 0.9252,
+      "step": 1722
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.490370401533327e-05,
+      "loss": 0.8308,
+      "step": 1723
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.489799668306976e-05,
+      "loss": 0.7357,
+      "step": 1724
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.489228651991381e-05,
+      "loss": 0.7238,
+      "step": 1725
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.48865735266778e-05,
+      "loss": 0.7964,
+      "step": 1726
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4880857704174506e-05,
+      "loss": 0.5082,
+      "step": 1727
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.487513905321713e-05,
+      "loss": 0.595,
+      "step": 1728
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4869417574619265e-05,
+      "loss": 0.5151,
+      "step": 1729
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.486369326919491e-05,
+      "loss": 0.8732,
+      "step": 1730
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.485796613775845e-05,
+      "loss": 0.6757,
+      "step": 1731
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4852236181124696e-05,
+      "loss": 0.6287,
+      "step": 1732
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4846503400108854e-05,
+      "loss": 0.7723,
+      "step": 1733
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4840767795526516e-05,
+      "loss": 0.9247,
+      "step": 1734
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4835029368193695e-05,
+      "loss": 0.6174,
+      "step": 1735
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.48292881189268e-05,
+      "loss": 0.7294,
+      "step": 1736
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.482354404854264e-05,
+      "loss": 0.5693,
+      "step": 1737
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.481779715785842e-05,
+      "loss": 0.8182,
+      "step": 1738
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4812047447691755e-05,
+      "loss": 0.6997,
+      "step": 1739
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.480629491886066e-05,
+      "loss": 0.7647,
+      "step": 1740
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.480053957218354e-05,
+      "loss": 0.7523,
+      "step": 1741
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.479478140847922e-05,
+      "loss": 0.3161,
+      "step": 1742
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4789020428566907e-05,
+      "loss": 0.5616,
+      "step": 1743
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4783256633266214e-05,
+      "loss": 0.5647,
+      "step": 1744
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.477749002339717e-05,
+      "loss": 0.8467,
+      "step": 1745
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4771720599780186e-05,
+      "loss": 0.6783,
+      "step": 1746
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4765948363236075e-05,
+      "loss": 0.6241,
+      "step": 1747
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.476017331458606e-05,
+      "loss": 0.5095,
+      "step": 1748
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.475439545465176e-05,
+      "loss": 0.6732,
+      "step": 1749
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.474861478425518e-05,
+      "loss": 0.6769,
+      "step": 1750
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.474283130421875e-05,
+      "loss": 0.6136,
+      "step": 1751
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4737045015365285e-05,
+      "loss": 0.6455,
+      "step": 1752
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4731255918518e-05,
+      "loss": 0.5945,
+      "step": 1753
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4725464014500505e-05,
+      "loss": 0.4116,
+      "step": 1754
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.471966930413682e-05,
+      "loss": 0.5734,
+      "step": 1755
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4713871788251374e-05,
+      "loss": 0.7568,
+      "step": 1756
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4708071467668954e-05,
+      "loss": 0.7932,
+      "step": 1757
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4702268343214795e-05,
+      "loss": 0.5914,
+      "step": 1758
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.469646241571449e-05,
+      "loss": 0.8336,
+      "step": 1759
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4690653685994075e-05,
+      "loss": 0.7303,
+      "step": 1760
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.468484215487994e-05,
+      "loss": 0.6141,
+      "step": 1761
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.467902782319889e-05,
+      "loss": 0.6136,
+      "step": 1762
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.467321069177814e-05,
+      "loss": 0.7505,
+      "step": 1763
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4667390761445294e-05,
+      "loss": 0.6198,
+      "step": 1764
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4661568033028354e-05,
+      "loss": 0.6624,
+      "step": 1765
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.465574250735572e-05,
+      "loss": 0.4207,
+      "step": 1766
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4649914185256195e-05,
+      "loss": 0.5513,
+      "step": 1767
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.464408306755897e-05,
+      "loss": 0.8252,
+      "step": 1768
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4638249155093647e-05,
+      "loss": 0.7589,
+      "step": 1769
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.463241244869021e-05,
+      "loss": 0.6538,
+      "step": 1770
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.462657294917904e-05,
+      "loss": 0.6242,
+      "step": 1771
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.462073065739095e-05,
+      "loss": 0.5984,
+      "step": 1772
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.461488557415711e-05,
+      "loss": 0.9276,
+      "step": 1773
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4609037700309094e-05,
+      "loss": 0.7748,
+      "step": 1774
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.460318703667888e-05,
+      "loss": 0.5645,
+      "step": 1775
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.459733358409887e-05,
+      "loss": 0.6294,
+      "step": 1776
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.45914773434018e-05,
+      "loss": 0.7878,
+      "step": 1777
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.458561831542086e-05,
+      "loss": 0.6202,
+      "step": 1778
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.4579756500989606e-05,
+      "loss": 0.7523,
+      "step": 1779
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.457389190094201e-05,
+      "loss": 0.7036,
+      "step": 1780
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.456802451611242e-05,
+      "loss": 0.9543,
+      "step": 1781
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.45621543473356e-05,
+      "loss": 0.5488,
+      "step": 1782
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.455628139544668e-05,
+      "loss": 0.6647,
+      "step": 1783
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.455040566128123e-05,
+      "loss": 0.6902,
+      "step": 1784
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.454452714567518e-05,
+      "loss": 0.5597,
+      "step": 1785
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.453864584946487e-05,
+      "loss": 0.6401,
+      "step": 1786
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.453276177348704e-05,
+      "loss": 0.5952,
+      "step": 1787
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.45268749185788e-05,
+      "loss": 0.8141,
+      "step": 1788
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.45209852855777e-05,
+      "loss": 0.605,
+      "step": 1789
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.451509287532164e-05,
+      "loss": 0.5925,
+      "step": 1790
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.450919768864894e-05,
+      "loss": 0.6147,
+      "step": 1791
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.450329972639831e-05,
+      "loss": 0.5869,
+      "step": 1792
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4497398989408845e-05,
+      "loss": 0.9741,
+      "step": 1793
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.449149547852006e-05,
+      "loss": 0.6953,
+      "step": 1794
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4485589194571847e-05,
+      "loss": 0.565,
+      "step": 1795
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.447968013840449e-05,
+      "loss": 0.816,
+      "step": 1796
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4473768310858666e-05,
+      "loss": 0.4431,
+      "step": 1797
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.446785371277547e-05,
+      "loss": 0.6282,
+      "step": 1798
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4461936344996344e-05,
+      "loss": 0.6357,
+      "step": 1799
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.445601620836317e-05,
+      "loss": 0.647,
+      "step": 1800
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4450093303718216e-05,
+      "loss": 0.676,
+      "step": 1801
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.444416763190412e-05,
+      "loss": 0.7941,
+      "step": 1802
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4438239193763934e-05,
+      "loss": 0.551,
+      "step": 1803
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4432307990141106e-05,
+      "loss": 0.6861,
+      "step": 1804
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.442637402187946e-05,
+      "loss": 0.6859,
+      "step": 1805
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4420437289823226e-05,
+      "loss": 0.6017,
+      "step": 1806
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.441449779481702e-05,
+      "loss": 0.5684,
+      "step": 1807
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4408555537705857e-05,
+      "loss": 0.6271,
+      "step": 1808
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4402610519335146e-05,
+      "loss": 0.5355,
+      "step": 1809
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4396662740550685e-05,
+      "loss": 0.6425,
+      "step": 1810
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.439071220219867e-05,
+      "loss": 0.712,
+      "step": 1811
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4384758905125686e-05,
+      "loss": 0.4586,
+      "step": 1812
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4378802850178694e-05,
+      "loss": 0.5871,
+      "step": 1813
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4372844038205086e-05,
+      "loss": 0.7022,
+      "step": 1814
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.436688247005261e-05,
+      "loss": 0.6058,
+      "step": 1815
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4360918146569416e-05,
+      "loss": 0.5919,
+      "step": 1816
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4354951068604055e-05,
+      "loss": 0.8478,
+      "step": 1817
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.434898123700547e-05,
+      "loss": 0.7844,
+      "step": 1818
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.434300865262298e-05,
+      "loss": 0.8745,
+      "step": 1819
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.433703331630632e-05,
+      "loss": 0.4953,
+      "step": 1820
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.433105522890559e-05,
+      "loss": 0.7443,
+      "step": 1821
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4325074391271294e-05,
+      "loss": 0.548,
+      "step": 1822
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4319090804254334e-05,
+      "loss": 0.8133,
+      "step": 1823
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.431310446870599e-05,
+      "loss": 0.8678,
+      "step": 1824
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.430711538547795e-05,
+      "loss": 0.8961,
+      "step": 1825
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.430112355542226e-05,
+      "loss": 0.8923,
+      "step": 1826
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4295128979391405e-05,
+      "loss": 0.8589,
+      "step": 1827
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.428913165823822e-05,
+      "loss": 0.921,
+      "step": 1828
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4283131592815945e-05,
+      "loss": 0.5227,
+      "step": 1829
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4277128783978215e-05,
+      "loss": 0.3287,
+      "step": 1830
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.427112323257906e-05,
+      "loss": 0.8015,
+      "step": 1831
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4265114939472874e-05,
+      "loss": 0.5516,
+      "step": 1832
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.425910390551446e-05,
+      "loss": 0.6821,
+      "step": 1833
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4253090131559014e-05,
+      "loss": 0.6915,
+      "step": 1834
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4247073618462134e-05,
+      "loss": 0.7219,
+      "step": 1835
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.424105436707976e-05,
+      "loss": 0.5186,
+      "step": 1836
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.423503237826826e-05,
+      "loss": 0.577,
+      "step": 1837
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.42290076528844e-05,
+      "loss": 0.688,
+      "step": 1838
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4222980191785314e-05,
+      "loss": 0.6113,
+      "step": 1839
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4216949995828515e-05,
+      "loss": 0.524,
+      "step": 1840
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.421091706587194e-05,
+      "loss": 0.6857,
+      "step": 1841
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.420488140277388e-05,
+      "loss": 0.8892,
+      "step": 1842
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.419884300739303e-05,
+      "loss": 0.7394,
+      "step": 1843
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.419280188058849e-05,
+      "loss": 0.7095,
+      "step": 1844
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.418675802321972e-05,
+      "loss": 0.6251,
+      "step": 1845
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4180711436146574e-05,
+      "loss": 0.7348,
+      "step": 1846
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4174662120229325e-05,
+      "loss": 0.8851,
+      "step": 1847
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.416861007632859e-05,
+      "loss": 0.567,
+      "step": 1848
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4162555305305396e-05,
+      "loss": 0.6002,
+      "step": 1849
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4156497808021166e-05,
+      "loss": 0.5726,
+      "step": 1850
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.41504375853377e-05,
+      "loss": 0.6465,
+      "step": 1851
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.414437463811718e-05,
+      "loss": 0.7835,
+      "step": 1852
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4138308967222185e-05,
+      "loss": 0.8925,
+      "step": 1853
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4132240573515695e-05,
+      "loss": 0.8186,
+      "step": 1854
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.412616945786103e-05,
+      "loss": 0.6971,
+      "step": 1855
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.412009562112196e-05,
+      "loss": 0.7062,
+      "step": 1856
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4114019064162593e-05,
+      "loss": 0.7839,
+      "step": 1857
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.410793978784744e-05,
+      "loss": 0.6903,
+      "step": 1858
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.410185779304142e-05,
+      "loss": 0.5103,
+      "step": 1859
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4095773080609806e-05,
+      "loss": 0.4371,
+      "step": 1860
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.408968565141827e-05,
+      "loss": 0.6293,
+      "step": 1861
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4083595506332876e-05,
+      "loss": 0.8527,
+      "step": 1862
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.407750264622007e-05,
+      "loss": 0.4455,
+      "step": 1863
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4071407071946686e-05,
+      "loss": 0.5803,
+      "step": 1864
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.406530878437994e-05,
+      "loss": 0.8133,
+      "step": 1865
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.405920778438743e-05,
+      "loss": 0.3893,
+      "step": 1866
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.405310407283717e-05,
+      "loss": 0.8375,
+      "step": 1867
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.404699765059751e-05,
+      "loss": 0.7983,
+      "step": 1868
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.404088851853723e-05,
+      "loss": 0.5478,
+      "step": 1869
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.403477667752547e-05,
+      "loss": 0.4652,
+      "step": 1870
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4028662128431754e-05,
+      "loss": 0.6164,
+      "step": 1871
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4022544872126015e-05,
+      "loss": 0.8089,
+      "step": 1872
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.4016424909478554e-05,
+      "loss": 0.7029,
+      "step": 1873
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.401030224136005e-05,
+      "loss": 0.6754,
+      "step": 1874
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.400417686864159e-05,
+      "loss": 0.511,
+      "step": 1875
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3998048792194624e-05,
+      "loss": 0.9174,
+      "step": 1876
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.399191801289099e-05,
+      "loss": 0.6074,
+      "step": 1877
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3985784531602924e-05,
+      "loss": 0.7465,
+      "step": 1878
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.397964834920304e-05,
+      "loss": 0.7662,
+      "step": 1879
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.397350946656433e-05,
+      "loss": 0.6608,
+      "step": 1880
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.396736788456017e-05,
+      "loss": 0.8489,
+      "step": 1881
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.396122360406433e-05,
+      "loss": 0.4852,
+      "step": 1882
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.395507662595095e-05,
+      "loss": 0.5825,
+      "step": 1883
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.394892695109456e-05,
+      "loss": 0.7353,
+      "step": 1884
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.39427745803701e-05,
+      "loss": 0.5566,
+      "step": 1885
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.393661951465284e-05,
+      "loss": 0.6848,
+      "step": 1886
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3930461754818485e-05,
+      "loss": 0.7695,
+      "step": 1887
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3924301301743085e-05,
+      "loss": 0.7626,
+      "step": 1888
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3918138156303104e-05,
+      "loss": 0.4301,
+      "step": 1889
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.391197231937535e-05,
+      "loss": 0.6855,
+      "step": 1890
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3905803791837066e-05,
+      "loss": 0.5346,
+      "step": 1891
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.389963257456583e-05,
+      "loss": 0.4619,
+      "step": 1892
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.389345866843963e-05,
+      "loss": 0.7504,
+      "step": 1893
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.388728207433683e-05,
+      "loss": 0.7422,
+      "step": 1894
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3881102793136174e-05,
+      "loss": 0.7747,
+      "step": 1895
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.387492082571679e-05,
+      "loss": 0.6773,
+      "step": 1896
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.386873617295818e-05,
+      "loss": 0.7548,
+      "step": 1897
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.386254883574025e-05,
+      "loss": 0.964,
+      "step": 1898
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.385635881494327e-05,
+      "loss": 0.7106,
+      "step": 1899
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.385016611144789e-05,
+      "loss": 0.6912,
+      "step": 1900
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.384397072613515e-05,
+      "loss": 0.8568,
+      "step": 1901
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3837772659886465e-05,
+      "loss": 0.9275,
+      "step": 1902
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3831571913583644e-05,
+      "loss": 0.6872,
+      "step": 1903
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3825368488108856e-05,
+      "loss": 0.7643,
+      "step": 1904
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.381916238434468e-05,
+      "loss": 0.5746,
+      "step": 1905
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3812953603174046e-05,
+      "loss": 0.8411,
+      "step": 1906
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.38067421454803e-05,
+      "loss": 0.7277,
+      "step": 1907
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3800528012147115e-05,
+      "loss": 0.5219,
+      "step": 1908
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.379431120405859e-05,
+      "loss": 0.4415,
+      "step": 1909
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.37880917220992e-05,
+      "loss": 1.003,
+      "step": 1910
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.378186956715379e-05,
+      "loss": 0.6957,
+      "step": 1911
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.377564474010759e-05,
+      "loss": 0.6712,
+      "step": 1912
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.37694172418462e-05,
+      "loss": 0.7508,
+      "step": 1913
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3763187073255596e-05,
+      "loss": 0.8956,
+      "step": 1914
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.375695423522217e-05,
+      "loss": 0.4683,
+      "step": 1915
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3750718728632666e-05,
+      "loss": 0.4916,
+      "step": 1916
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.374448055437419e-05,
+      "loss": 0.5594,
+      "step": 1917
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.373823971333427e-05,
+      "loss": 0.6915,
+      "step": 1918
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3731996206400784e-05,
+      "loss": 0.8148,
+      "step": 1919
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3725750034461993e-05,
+      "loss": 0.8248,
+      "step": 1920
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.371950119840655e-05,
+      "loss": 0.7253,
+      "step": 1921
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.371324969912347e-05,
+      "loss": 0.6131,
+      "step": 1922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.370699553750216e-05,
+      "loss": 0.5313,
+      "step": 1923
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3700738714432396e-05,
+      "loss": 0.7969,
+      "step": 1924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3694479230804344e-05,
+      "loss": 0.6624,
+      "step": 1925
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3688217087508545e-05,
+      "loss": 0.6183,
+      "step": 1926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.36819522854359e-05,
+      "loss": 0.7032,
+      "step": 1927
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.367568482547772e-05,
+      "loss": 0.6625,
+      "step": 1928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.366941470852567e-05,
+      "loss": 0.6451,
+      "step": 1929
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.366314193547181e-05,
+      "loss": 0.656,
+      "step": 1930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.365686650720855e-05,
+      "loss": 0.5857,
+      "step": 1931
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3650588424628715e-05,
+      "loss": 0.8666,
+      "step": 1932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3644307688625485e-05,
+      "loss": 0.8217,
+      "step": 1933
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.363802430009241e-05,
+      "loss": 0.7593,
+      "step": 1934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3631738259923446e-05,
+      "loss": 0.7199,
+      "step": 1935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.36254495690129e-05,
+      "loss": 0.6517,
+      "step": 1936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3619158228255466e-05,
+      "loss": 0.4688,
+      "step": 1937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3612864238546215e-05,
+      "loss": 0.4619,
+      "step": 1938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.360656760078059e-05,
+      "loss": 0.6937,
+      "step": 1939
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.360026831585443e-05,
+      "loss": 0.7148,
+      "step": 1940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.359396638466392e-05,
+      "loss": 0.7736,
+      "step": 1941
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.358766180810564e-05,
+      "loss": 0.7817,
+      "step": 1942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3581354587076554e-05,
+      "loss": 0.6346,
+      "step": 1943
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3575044722473977e-05,
+      "loss": 0.8103,
+      "step": 1944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.356873221519562e-05,
+      "loss": 0.5526,
+      "step": 1945
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.356241706613957e-05,
+      "loss": 0.7663,
+      "step": 1946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.355609927620428e-05,
+      "loss": 0.5209,
+      "step": 1947
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.354977884628859e-05,
+      "loss": 0.9198,
+      "step": 1948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3543455777291694e-05,
+      "loss": 0.5565,
+      "step": 1949
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3537130070113195e-05,
+      "loss": 0.5222,
+      "step": 1950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.353080172565304e-05,
+      "loss": 0.7914,
+      "step": 1951
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.352447074481157e-05,
+      "loss": 0.8604,
+      "step": 1952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.351813712848949e-05,
+      "loss": 0.665,
+      "step": 1953
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.351180087758789e-05,
+      "loss": 0.7737,
+      "step": 1954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3505461993008226e-05,
+      "loss": 0.7741,
+      "step": 1955
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.349912047565235e-05,
+      "loss": 0.6396,
+      "step": 1956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.3492776326422434e-05,
+      "loss": 0.8132,
+      "step": 1957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3486429546221085e-05,
+      "loss": 0.8467,
+      "step": 1958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3480080135951264e-05,
+      "loss": 0.7467,
+      "step": 1959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3473728096516294e-05,
+      "loss": 0.6882,
+      "step": 1960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.34673734288199e-05,
+      "loss": 0.5886,
+      "step": 1961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.346101613376613e-05,
+      "loss": 0.5751,
+      "step": 1962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.345465621225946e-05,
+      "loss": 0.9865,
+      "step": 1963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.344829366520472e-05,
+      "loss": 0.6347,
+      "step": 1964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.344192849350709e-05,
+      "loss": 0.5032,
+      "step": 1965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3435560698072166e-05,
+      "loss": 0.5365,
+      "step": 1966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.342919027980588e-05,
+      "loss": 0.8989,
+      "step": 1967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.342281723961457e-05,
+      "loss": 0.8695,
+      "step": 1968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.341644157840491e-05,
+      "loss": 0.8136,
+      "step": 1969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.341006329708398e-05,
+      "loss": 0.5938,
+      "step": 1970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.340368239655922e-05,
+      "loss": 0.7209,
+      "step": 1971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3397298877738434e-05,
+      "loss": 0.7628,
+      "step": 1972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.339091274152981e-05,
+      "loss": 0.9218,
+      "step": 1973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.338452398884191e-05,
+      "loss": 0.8557,
+      "step": 1974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.337813262058366e-05,
+      "loss": 0.8363,
+      "step": 1975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.337173863766436e-05,
+      "loss": 0.5886,
+      "step": 1976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.336534204099368e-05,
+      "loss": 0.7337,
+      "step": 1977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.335894283148168e-05,
+      "loss": 0.8102,
+      "step": 1978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.335254101003875e-05,
+      "loss": 0.4185,
+      "step": 1979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.334613657757571e-05,
+      "loss": 0.7948,
+      "step": 1980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.333972953500369e-05,
+      "loss": 0.4582,
+      "step": 1981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.333331988323425e-05,
+      "loss": 0.6092,
+      "step": 1982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.332690762317927e-05,
+      "loss": 0.4473,
+      "step": 1983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3320492755751044e-05,
+      "loss": 0.5813,
+      "step": 1984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.331407528186219e-05,
+      "loss": 0.8765,
+      "step": 1985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3307655202425745e-05,
+      "loss": 0.8863,
+      "step": 1986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.330123251835509e-05,
+      "loss": 0.5545,
+      "step": 1987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3294807230563984e-05,
+      "loss": 0.599,
+      "step": 1988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3288379339966546e-05,
+      "loss": 0.6542,
+      "step": 1989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.328194884747728e-05,
+      "loss": 1.134,
+      "step": 1990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3275515754011054e-05,
+      "loss": 0.5551,
+      "step": 1991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.32690800604831e-05,
+      "loss": 0.7628,
+      "step": 1992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.326264176780902e-05,
+      "loss": 0.7822,
+      "step": 1993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.325620087690481e-05,
+      "loss": 0.8103,
+      "step": 1994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.324975738868681e-05,
+      "loss": 0.6332,
+      "step": 1995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.324331130407173e-05,
+      "loss": 0.6014,
+      "step": 1996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.323686262397666e-05,
+      "loss": 0.5034,
+      "step": 1997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.323041134931905e-05,
+      "loss": 0.4778,
+      "step": 1998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.322395748101674e-05,
+      "loss": 0.655,
+      "step": 1999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3217501019987904e-05,
+      "loss": 0.8301,
+      "step": 2000
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 0.75048828125,
+      "eval_runtime": 3.6048,
+      "eval_samples_per_second": 27.741,
+      "eval_steps_per_second": 4.716,
+      "step": 2000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.321104196715111e-05,
+      "loss": 0.6713,
+      "step": 2001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3204580323425296e-05,
+      "loss": 0.8718,
+      "step": 2002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.319811608972976e-05,
+      "loss": 0.6578,
+      "step": 2003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3191649266984164e-05,
+      "loss": 0.6699,
+      "step": 2004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.318517985610855e-05,
+      "loss": 0.476,
+      "step": 2005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.317870785802331e-05,
+      "loss": 0.8882,
+      "step": 2006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.317223327364923e-05,
+      "loss": 0.8036,
+      "step": 2007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.316575610390745e-05,
+      "loss": 0.8063,
+      "step": 2008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.315927634971948e-05,
+      "loss": 0.724,
+      "step": 2009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.315279401200719e-05,
+      "loss": 0.6727,
+      "step": 2010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3146309091692816e-05,
+      "loss": 0.7653,
+      "step": 2011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3139821589698985e-05,
+      "loss": 0.8763,
+      "step": 2012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.313333150694867e-05,
+      "loss": 0.6496,
+      "step": 2013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.312683884436521e-05,
+      "loss": 0.6473,
+      "step": 2014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3120343602872324e-05,
+      "loss": 0.367,
+      "step": 2015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.311384578339409e-05,
+      "loss": 0.5323,
+      "step": 2016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.310734538685496e-05,
+      "loss": 0.6335,
+      "step": 2017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.310084241417973e-05,
+      "loss": 0.7479,
+      "step": 2018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3094336866293606e-05,
+      "loss": 0.6846,
+      "step": 2019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3087828744122104e-05,
+      "loss": 0.6192,
+      "step": 2020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3081318048591156e-05,
+      "loss": 0.7292,
+      "step": 2021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.307480478062703e-05,
+      "loss": 0.6605,
+      "step": 2022
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.306828894115639e-05,
+      "loss": 0.7722,
+      "step": 2023
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.306177053110622e-05,
+      "loss": 0.778,
+      "step": 2024
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3055249551403895e-05,
+      "loss": 0.4865,
+      "step": 2025
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.304872600297718e-05,
+      "loss": 0.5792,
+      "step": 2026
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.304219988675417e-05,
+      "loss": 0.7191,
+      "step": 2027
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.303567120366333e-05,
+      "loss": 0.6098,
+      "step": 2028
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3029139954633505e-05,
+      "loss": 0.5415,
+      "step": 2029
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.30226061405939e-05,
+      "loss": 0.6473,
+      "step": 2030
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.301606976247407e-05,
+      "loss": 0.7358,
+      "step": 2031
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.3009530821203956e-05,
+      "loss": 0.6081,
+      "step": 2032
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.300298931771386e-05,
+      "loss": 0.4302,
+      "step": 2033
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.2996445252934437e-05,
+      "loss": 0.6424,
+      "step": 2034
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.298989862779671e-05,
+      "loss": 0.6826,
+      "step": 2035
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.2983349443232066e-05,
+      "loss": 0.6982,
+      "step": 2036
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.2976797700172264e-05,
+      "loss": 0.7054,
+      "step": 2037
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.297024339954943e-05,
+      "loss": 0.8681,
+      "step": 2038
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.296368654229603e-05,
+      "loss": 0.629,
+      "step": 2039
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.295712712934492e-05,
+      "loss": 0.7241,
+      "step": 2040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.29505651616293e-05,
+      "loss": 0.6604,
+      "step": 2041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.294400064008276e-05,
+      "loss": 0.9403,
+      "step": 2042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.293743356563922e-05,
+      "loss": 0.6039,
+      "step": 2043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2930863939232976e-05,
+      "loss": 0.92,
+      "step": 2044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2924291761798706e-05,
+      "loss": 0.4404,
+      "step": 2045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.291771703427143e-05,
+      "loss": 0.6128,
+      "step": 2046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2911139757586536e-05,
+      "loss": 1.0013,
+      "step": 2047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2904559932679766e-05,
+      "loss": 0.9681,
+      "step": 2048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.289797756048724e-05,
+      "loss": 0.4328,
+      "step": 2049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2891392641945426e-05,
+      "loss": 0.7516,
+      "step": 2050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.288480517799118e-05,
+      "loss": 0.7867,
+      "step": 2051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2878215169561684e-05,
+      "loss": 0.7596,
+      "step": 2052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2871622617594515e-05,
+      "loss": 0.7301,
+      "step": 2053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2865027523027576e-05,
+      "loss": 0.5823,
+      "step": 2054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.285842988679918e-05,
+      "loss": 0.8087,
+      "step": 2055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2851829709847946e-05,
+      "loss": 0.7204,
+      "step": 2056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.28452269931129e-05,
+      "loss": 0.8707,
+      "step": 2057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2838621737533404e-05,
+      "loss": 0.5825,
+      "step": 2058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2832013944049195e-05,
+      "loss": 0.6227,
+      "step": 2059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.282540361360037e-05,
+      "loss": 0.6497,
+      "step": 2060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2818790747127366e-05,
+      "loss": 1.0072,
+      "step": 2061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.281217534557101e-05,
+      "loss": 0.8104,
+      "step": 2062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.280555740987248e-05,
+      "loss": 0.6536,
+      "step": 2063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.27989369409733e-05,
+      "loss": 0.4926,
+      "step": 2064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.279231393981537e-05,
+      "loss": 0.8186,
+      "step": 2065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.278568840734094e-05,
+      "loss": 0.7145,
+      "step": 2066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.277906034449264e-05,
+      "loss": 0.7523,
+      "step": 2067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2772429752213426e-05,
+      "loss": 0.6756,
+      "step": 2068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2765796631446656e-05,
+      "loss": 0.7508,
+      "step": 2069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.275916098313602e-05,
+      "loss": 0.803,
+      "step": 2070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.275252280822556e-05,
+      "loss": 0.5919,
+      "step": 2071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2745882107659694e-05,
+      "loss": 0.5073,
+      "step": 2072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2739238882383206e-05,
+      "loss": 0.8359,
+      "step": 2073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2732593133341227e-05,
+      "loss": 0.613,
+      "step": 2074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.272594486147925e-05,
+      "loss": 0.8103,
+      "step": 2075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.271929406774311e-05,
+      "loss": 0.7899,
+      "step": 2076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2712640753079034e-05,
+      "loss": 0.9061,
+      "step": 2077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.270598491843359e-05,
+      "loss": 0.6216,
+      "step": 2078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2699326564753706e-05,
+      "loss": 0.6773,
+      "step": 2079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.269266569298665e-05,
+      "loss": 0.8018,
+      "step": 2080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.268600230408009e-05,
+      "loss": 0.6796,
+      "step": 2081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2679336398982e-05,
+      "loss": 0.6008,
+      "step": 2082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.267266797864077e-05,
+      "loss": 0.6829,
+      "step": 2083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.26659970440051e-05,
+      "loss": 0.882,
+      "step": 2084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.265932359602407e-05,
+      "loss": 0.917,
+      "step": 2085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.265264763564711e-05,
+      "loss": 0.6874,
+      "step": 2086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2645969163824005e-05,
+      "loss": 0.8999,
+      "step": 2087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.263928818150492e-05,
+      "loss": 0.8998,
+      "step": 2088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2632604689640355e-05,
+      "loss": 1.0046,
+      "step": 2089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.262591868918115e-05,
+      "loss": 0.7469,
+      "step": 2090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.261923018107855e-05,
+      "loss": 0.7886,
+      "step": 2091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.261253916628413e-05,
+      "loss": 0.5721,
+      "step": 2092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.26058456457498e-05,
+      "loss": 0.795,
+      "step": 2093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.259914962042786e-05,
+      "loss": 0.5859,
+      "step": 2094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.259245109127096e-05,
+      "loss": 0.5763,
+      "step": 2095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.25857500592321e-05,
+      "loss": 0.5454,
+      "step": 2096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.257904652526464e-05,
+      "loss": 0.8335,
+      "step": 2097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.257234049032228e-05,
+      "loss": 0.6387,
+      "step": 2098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2565631955359106e-05,
+      "loss": 0.544,
+      "step": 2099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2558920921329526e-05,
+      "loss": 0.6688,
+      "step": 2100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.255220738918833e-05,
+      "loss": 0.8273,
+      "step": 2101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2545491359890655e-05,
+      "loss": 0.587,
+      "step": 2102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2538772834391994e-05,
+      "loss": 0.6093,
+      "step": 2103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2532051813648185e-05,
+      "loss": 0.5896,
+      "step": 2104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.252532829861543e-05,
+      "loss": 0.6287,
+      "step": 2105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2518602290250295e-05,
+      "loss": 0.7556,
+      "step": 2106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2511873789509674e-05,
+      "loss": 0.8739,
+      "step": 2107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.250514279735085e-05,
+      "loss": 0.827,
+      "step": 2108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2498409314731427e-05,
+      "loss": 0.4242,
+      "step": 2109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2491673342609395e-05,
+      "loss": 0.6805,
+      "step": 2110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2484934881943075e-05,
+      "loss": 0.8064,
+      "step": 2111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2478193933691137e-05,
+      "loss": 0.5422,
+      "step": 2112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.247145049881264e-05,
+      "loss": 0.6019,
+      "step": 2113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.246470457826696e-05,
+      "loss": 0.8285,
+      "step": 2114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.245795617301384e-05,
+      "loss": 0.6542,
+      "step": 2115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.245120528401337e-05,
+      "loss": 0.6142,
+      "step": 2116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2444451912226026e-05,
+      "loss": 0.9938,
+      "step": 2117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.243769605861259e-05,
+      "loss": 0.6257,
+      "step": 2118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.243093772413423e-05,
+      "loss": 0.5182,
+      "step": 2119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.2424176909752444e-05,
+      "loss": 0.4578,
+      "step": 2120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.241741361642911e-05,
+      "loss": 0.4419,
+      "step": 2121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.241064784512643e-05,
+      "loss": 0.9402,
+      "step": 2122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.240387959680698e-05,
+      "loss": 0.8981,
+      "step": 2123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.239710887243367e-05,
+      "loss": 0.5536,
+      "step": 2124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2390335672969786e-05,
+      "loss": 0.7649,
+      "step": 2125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2383559999378944e-05,
+      "loss": 0.7604,
+      "step": 2126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.237678185262512e-05,
+      "loss": 0.5764,
+      "step": 2127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2370001233672645e-05,
+      "loss": 0.7555,
+      "step": 2128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2363218143486206e-05,
+      "loss": 0.6205,
+      "step": 2129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2356432583030824e-05,
+      "loss": 0.8335,
+      "step": 2130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2349644553271884e-05,
+      "loss": 0.5488,
+      "step": 2131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.234285405517512e-05,
+      "loss": 1.0094,
+      "step": 2132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2336061089706626e-05,
+      "loss": 0.8461,
+      "step": 2133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.232926565783283e-05,
+      "loss": 0.617,
+      "step": 2134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.232246776052052e-05,
+      "loss": 0.8666,
+      "step": 2135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2315667398736845e-05,
+      "loss": 0.6134,
+      "step": 2136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.230886457344929e-05,
+      "loss": 0.6765,
+      "step": 2137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.230205928562568e-05,
+      "loss": 0.4482,
+      "step": 2138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.229525153623422e-05,
+      "loss": 0.4881,
+      "step": 2139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2288441326243446e-05,
+      "loss": 0.8024,
+      "step": 2140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.228162865662224e-05,
+      "loss": 0.5537,
+      "step": 2141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.227481352833985e-05,
+      "loss": 0.903,
+      "step": 2142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.226799594236587e-05,
+      "loss": 0.6592,
+      "step": 2143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2261175899670244e-05,
+      "loss": 0.736,
+      "step": 2144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2254353401223235e-05,
+      "loss": 0.6634,
+      "step": 2145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.224752844799551e-05,
+      "loss": 0.6108,
+      "step": 2146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.224070104095804e-05,
+      "loss": 0.7458,
+      "step": 2147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.223387118108216e-05,
+      "loss": 0.5256,
+      "step": 2148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.222703886933957e-05,
+      "loss": 0.8729,
+      "step": 2149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.22202041067023e-05,
+      "loss": 0.6265,
+      "step": 2150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2213366894142736e-05,
+      "loss": 0.7515,
+      "step": 2151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.220652723263359e-05,
+      "loss": 0.8785,
+      "step": 2152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.219968512314796e-05,
+      "loss": 0.6022,
+      "step": 2153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.219284056665927e-05,
+      "loss": 0.9584,
+      "step": 2154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2185993564141303e-05,
+      "loss": 0.7137,
+      "step": 2155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.217914411656817e-05,
+      "loss": 0.7892,
+      "step": 2156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2172292224914353e-05,
+      "loss": 0.7497,
+      "step": 2157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.216543789015467e-05,
+      "loss": 0.7356,
+      "step": 2158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.21585811132643e-05,
+      "loss": 0.8006,
+      "step": 2159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.215172189521873e-05,
+      "loss": 0.6231,
+      "step": 2160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.214486023699384e-05,
+      "loss": 0.4614,
+      "step": 2161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.213799613956585e-05,
+      "loss": 0.5821,
+      "step": 2162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.21311296039113e-05,
+      "loss": 0.5162,
+      "step": 2163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2124260631007104e-05,
+      "loss": 0.6658,
+      "step": 2164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.21173892218305e-05,
+      "loss": 0.6655,
+      "step": 2165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2110515377359104e-05,
+      "loss": 0.7598,
+      "step": 2166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.210363909857084e-05,
+      "loss": 1.0285,
+      "step": 2167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.209676038644401e-05,
+      "loss": 0.7982,
+      "step": 2168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.208987924195724e-05,
+      "loss": 0.6087,
+      "step": 2169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.208299566608952e-05,
+      "loss": 0.6182,
+      "step": 2170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.207610965982019e-05,
+      "loss": 0.5861,
+      "step": 2171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2069221224128886e-05,
+      "loss": 0.7087,
+      "step": 2172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.206233035999566e-05,
+      "loss": 0.8751,
+      "step": 2173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.205543706840087e-05,
+      "loss": 0.6782,
+      "step": 2174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.204854135032522e-05,
+      "loss": 0.7047,
+      "step": 2175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.204164320674977e-05,
+      "loss": 0.5815,
+      "step": 2176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.2034742638655924e-05,
+      "loss": 0.7468,
+      "step": 2177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.202783964702542e-05,
+      "loss": 0.603,
+      "step": 2178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.202093423284035e-05,
+      "loss": 0.5554,
+      "step": 2179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.201402639708315e-05,
+      "loss": 0.7314,
+      "step": 2180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.200711614073659e-05,
+      "loss": 0.5334,
+      "step": 2181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.200020346478382e-05,
+      "loss": 0.8365,
+      "step": 2182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1993288370208276e-05,
+      "loss": 0.7863,
+      "step": 2183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.19863708579938e-05,
+      "loss": 0.5732,
+      "step": 2184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.197945092912452e-05,
+      "loss": 0.4169,
+      "step": 2185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.197252858458495e-05,
+      "loss": 0.8065,
+      "step": 2186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1965603825359934e-05,
+      "loss": 0.7363,
+      "step": 2187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.195867665243466e-05,
+      "loss": 0.6041,
+      "step": 2188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.195174706679464e-05,
+      "loss": 0.7363,
+      "step": 2189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.194481506942579e-05,
+      "loss": 0.6378,
+      "step": 2190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.193788066131428e-05,
+      "loss": 0.8286,
+      "step": 2191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1930943843446704e-05,
+      "loss": 0.8988,
+      "step": 2192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.192400461680994e-05,
+      "loss": 0.7857,
+      "step": 2193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1917062982391247e-05,
+      "loss": 0.6783,
+      "step": 2194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1910118941178206e-05,
+      "loss": 0.7021,
+      "step": 2195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1903172494158756e-05,
+      "loss": 0.6661,
+      "step": 2196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.189622364232117e-05,
+      "loss": 0.4936,
+      "step": 2197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.188927238665404e-05,
+      "loss": 0.6649,
+      "step": 2198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.188231872814635e-05,
+      "loss": 0.8074,
+      "step": 2199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.187536266778739e-05,
+      "loss": 0.7435,
+      "step": 2200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.1868404206566804e-05,
+      "loss": 0.7225,
+      "step": 2201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.186144334547456e-05,
+      "loss": 0.7722,
+      "step": 2202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.185448008550099e-05,
+      "loss": 0.9573,
+      "step": 2203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.184751442763677e-05,
+      "loss": 0.5062,
+      "step": 2204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.184054637287288e-05,
+      "loss": 0.6998,
+      "step": 2205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.183357592220069e-05,
+      "loss": 0.8,
+      "step": 2206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.182660307661187e-05,
+      "loss": 0.8086,
+      "step": 2207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1819627837098466e-05,
+      "loss": 0.7542,
+      "step": 2208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1812650204652833e-05,
+      "loss": 0.7258,
+      "step": 2209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.180567018026769e-05,
+      "loss": 0.5328,
+      "step": 2210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1798687764936076e-05,
+      "loss": 0.6716,
+      "step": 2211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1791702959651386e-05,
+      "loss": 0.6823,
+      "step": 2212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.178471576540736e-05,
+      "loss": 0.8137,
+      "step": 2213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1777726183198044e-05,
+      "loss": 0.7627,
+      "step": 2214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1770734214017866e-05,
+      "loss": 0.6804,
+      "step": 2215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1763739858861574e-05,
+      "loss": 0.6378,
+      "step": 2216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.175674311872425e-05,
+      "loss": 0.6399,
+      "step": 2217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.174974399460133e-05,
+      "loss": 0.7083,
+      "step": 2218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.174274248748857e-05,
+      "loss": 0.5811,
+      "step": 2219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.173573859838208e-05,
+      "loss": 0.9971,
+      "step": 2220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1728732328278305e-05,
+      "loss": 0.8562,
+      "step": 2221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.172172367817404e-05,
+      "loss": 0.6392,
+      "step": 2222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.171471264906639e-05,
+      "loss": 0.9115,
+      "step": 2223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1707699241952825e-05,
+      "loss": 0.8047,
+      "step": 2224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1700683457831146e-05,
+      "loss": 0.9575,
+      "step": 2225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.169366529769948e-05,
+      "loss": 0.9453,
+      "step": 2226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1686644762556314e-05,
+      "loss": 0.5287,
+      "step": 2227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.167962185340045e-05,
+      "loss": 0.6898,
+      "step": 2228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.167259657123105e-05,
+      "loss": 0.5454,
+      "step": 2229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1665568917047595e-05,
+      "loss": 0.4932,
+      "step": 2230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.165853889184991e-05,
+      "loss": 0.5137,
+      "step": 2231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1651506496638174e-05,
+      "loss": 0.8478,
+      "step": 2232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.164447173241286e-05,
+      "loss": 0.4354,
+      "step": 2233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.163743460017482e-05,
+      "loss": 0.6288,
+      "step": 2234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1630395100925236e-05,
+      "loss": 0.8006,
+      "step": 2235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.16233532356656e-05,
+      "loss": 0.564,
+      "step": 2236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.161630900539778e-05,
+      "loss": 0.6057,
+      "step": 2237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.160926241112395e-05,
+      "loss": 0.7856,
+      "step": 2238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.160221345384664e-05,
+      "loss": 0.4229,
+      "step": 2239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.159516213456869e-05,
+      "loss": 0.8425,
+      "step": 2240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1588108454293305e-05,
+      "loss": 0.7077,
+      "step": 2241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.158105241402401e-05,
+      "loss": 0.5044,
+      "step": 2242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.157399401476466e-05,
+      "loss": 0.6089,
+      "step": 2243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1566933257519484e-05,
+      "loss": 0.7016,
+      "step": 2244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1559870143292977e-05,
+      "loss": 0.8069,
+      "step": 2245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1552804673090046e-05,
+      "loss": 0.8969,
+      "step": 2246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.154573684791587e-05,
+      "loss": 0.4764,
+      "step": 2247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.153866666877602e-05,
+      "loss": 0.7346,
+      "step": 2248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.153159413667633e-05,
+      "loss": 0.5914,
+      "step": 2249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1524519252623054e-05,
+      "loss": 0.6361,
+      "step": 2250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1517442017622715e-05,
+      "loss": 0.5421,
+      "step": 2251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1510362432682194e-05,
+      "loss": 0.8745,
+      "step": 2252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.150328049880871e-05,
+      "loss": 0.9564,
+      "step": 2253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.14961962170098e-05,
+      "loss": 0.722,
+      "step": 2254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1489109588293364e-05,
+      "loss": 0.8788,
+      "step": 2255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1482020613667606e-05,
+      "loss": 0.5615,
+      "step": 2256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.147492929414107e-05,
+      "loss": 0.7012,
+      "step": 2257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.146783563072266e-05,
+      "loss": 0.5769,
+      "step": 2258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1460739624421583e-05,
+      "loss": 0.7341,
+      "step": 2259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1453641276247386e-05,
+      "loss": 0.7894,
+      "step": 2260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1446540587209954e-05,
+      "loss": 0.8419,
+      "step": 2261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.143943755831951e-05,
+      "loss": 0.4904,
+      "step": 2262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.143233219058659e-05,
+      "loss": 0.6753,
+      "step": 2263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.142522448502208e-05,
+      "loss": 0.9129,
+      "step": 2264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.14181144426372e-05,
+      "loss": 0.6019,
+      "step": 2265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1411002064443506e-05,
+      "loss": 0.8553,
+      "step": 2266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1403887351452865e-05,
+      "loss": 0.5594,
+      "step": 2267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.139677030467749e-05,
+      "loss": 0.6536,
+      "step": 2268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.138965092512993e-05,
+      "loss": 0.476,
+      "step": 2269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.138252921382305e-05,
+      "loss": 0.6782,
+      "step": 2270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.137540517177007e-05,
+      "loss": 0.5553,
+      "step": 2271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.136827879998453e-05,
+      "loss": 0.73,
+      "step": 2272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.136115009948028e-05,
+      "loss": 0.6866,
+      "step": 2273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1354019071271544e-05,
+      "loss": 1.1044,
+      "step": 2274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.134688571637285e-05,
+      "loss": 0.9385,
+      "step": 2275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.133975003579906e-05,
+      "loss": 0.8496,
+      "step": 2276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1332612030565364e-05,
+      "loss": 0.7386,
+      "step": 2277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.132547170168729e-05,
+      "loss": 0.7675,
+      "step": 2278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.13183290501807e-05,
+      "loss": 0.6406,
+      "step": 2279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.131118407706178e-05,
+      "loss": 0.7158,
+      "step": 2280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1304036783347035e-05,
+      "loss": 0.9187,
+      "step": 2281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.129688717005332e-05,
+      "loss": 0.5884,
+      "step": 2282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1289735238197816e-05,
+      "loss": 0.6398,
+      "step": 2283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.128258098879803e-05,
+      "loss": 0.7388,
+      "step": 2284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1275424422871794e-05,
+      "loss": 0.6894,
+      "step": 2285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.126826554143728e-05,
+      "loss": 0.842,
+      "step": 2286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.126110434551297e-05,
+      "loss": 0.6595,
+      "step": 2287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.12539408361177e-05,
+      "loss": 0.7384,
+      "step": 2288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1246775014270626e-05,
+      "loss": 0.7559,
+      "step": 2289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.1239606880991224e-05,
+      "loss": 0.7914,
+      "step": 2290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.123243643729931e-05,
+      "loss": 0.7299,
+      "step": 2291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1225263684215026e-05,
+      "loss": 0.5391,
+      "step": 2292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1218088622758835e-05,
+      "loss": 0.5778,
+      "step": 2293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.121091125395154e-05,
+      "loss": 0.5605,
+      "step": 2294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.120373157881427e-05,
+      "loss": 0.7271,
+      "step": 2295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1196549598368474e-05,
+      "loss": 0.6151,
+      "step": 2296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.118936531363593e-05,
+      "loss": 0.6545,
+      "step": 2297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1182178725638753e-05,
+      "loss": 0.6907,
+      "step": 2298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.117498983539938e-05,
+      "loss": 0.6434,
+      "step": 2299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.116779864394058e-05,
+      "loss": 0.7598,
+      "step": 2300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.116060515228544e-05,
+      "loss": 1.0082,
+      "step": 2301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1153409361457374e-05,
+      "loss": 0.7031,
+      "step": 2302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.114621127248014e-05,
+      "loss": 0.9389,
+      "step": 2303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1139010886377805e-05,
+      "loss": 0.9164,
+      "step": 2304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.113180820417478e-05,
+      "loss": 0.4728,
+      "step": 2305
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.112460322689578e-05,
+      "loss": 0.5668,
+      "step": 2306
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.111739595556586e-05,
+      "loss": 0.4822,
+      "step": 2307
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1110186391210407e-05,
+      "loss": 0.5546,
+      "step": 2308
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1102974534855124e-05,
+      "loss": 0.6858,
+      "step": 2309
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.109576038752604e-05,
+      "loss": 0.7088,
+      "step": 2310
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.108854395024951e-05,
+      "loss": 0.5797,
+      "step": 2311
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.108132522405223e-05,
+      "loss": 0.9745,
+      "step": 2312
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1074104209961203e-05,
+      "loss": 0.6511,
+      "step": 2313
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.106688090900377e-05,
+      "loss": 0.5168,
+      "step": 2314
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1059655322207585e-05,
+      "loss": 0.506,
+      "step": 2315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.105242745060063e-05,
+      "loss": 0.6996,
+      "step": 2316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.104519729521123e-05,
+      "loss": 0.5608,
+      "step": 2317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.103796485706801e-05,
+      "loss": 0.6276,
+      "step": 2318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.103073013719993e-05,
+      "loss": 0.6982,
+      "step": 2319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.102349313663628e-05,
+      "loss": 0.7343,
+      "step": 2320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.1016253856406675e-05,
+      "loss": 0.6958,
+      "step": 2321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.100901229754103e-05,
+      "loss": 0.8087,
+      "step": 2322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.100176846106962e-05,
+      "loss": 0.6954,
+      "step": 2323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0994522348023024e-05,
+      "loss": 0.7565,
+      "step": 2324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.098727395943215e-05,
+      "loss": 0.626,
+      "step": 2325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0980023296328216e-05,
+      "loss": 0.6993,
+      "step": 2326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0972770359742794e-05,
+      "loss": 0.4949,
+      "step": 2327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.096551515070775e-05,
+      "loss": 0.8011,
+      "step": 2328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.095825767025528e-05,
+      "loss": 0.5013,
+      "step": 2329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.095099791941791e-05,
+      "loss": 0.5731,
+      "step": 2330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0943735899228494e-05,
+      "loss": 0.6848,
+      "step": 2331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.093647161072019e-05,
+      "loss": 0.5931,
+      "step": 2332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0929205054926503e-05,
+      "loss": 1.0883,
+      "step": 2333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.092193623288124e-05,
+      "loss": 0.4493,
+      "step": 2334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.091466514561854e-05,
+      "loss": 0.7621,
+      "step": 2335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0907391794172856e-05,
+      "loss": 1.0511,
+      "step": 2336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0900116179578976e-05,
+      "loss": 0.8733,
+      "step": 2337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0892838302872005e-05,
+      "loss": 0.9802,
+      "step": 2338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.088555816508736e-05,
+      "loss": 0.6227,
+      "step": 2339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.087827576726079e-05,
+      "loss": 0.5497,
+      "step": 2340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0870991110428374e-05,
+      "loss": 0.8583,
+      "step": 2341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.086370419562649e-05,
+      "loss": 0.6377,
+      "step": 2342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.085641502389185e-05,
+      "loss": 0.5988,
+      "step": 2343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0849123596261496e-05,
+      "loss": 0.8696,
+      "step": 2344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0841829913772766e-05,
+      "loss": 0.8018,
+      "step": 2345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.083453397746334e-05,
+      "loss": 0.6954,
+      "step": 2346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.082723578837122e-05,
+      "loss": 0.6594,
+      "step": 2347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.081993534753471e-05,
+      "loss": 0.7285,
+      "step": 2348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0812632655992456e-05,
+      "loss": 0.6675,
+      "step": 2349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.080532771478341e-05,
+      "loss": 0.7227,
+      "step": 2350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.079802052494684e-05,
+      "loss": 0.7231,
+      "step": 2351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.079071108752234e-05,
+      "loss": 0.7473,
+      "step": 2352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0783399403549844e-05,
+      "loss": 0.6938,
+      "step": 2353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.077608547406957e-05,
+      "loss": 0.6644,
+      "step": 2354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0768769300122086e-05,
+      "loss": 0.6909,
+      "step": 2355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.076145088274826e-05,
+      "loss": 0.7047,
+      "step": 2356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.075413022298926e-05,
+      "loss": 0.8654,
+      "step": 2357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0746807321886635e-05,
+      "loss": 0.7686,
+      "step": 2358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0739482180482205e-05,
+      "loss": 0.8198,
+      "step": 2359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.07321547998181e-05,
+      "loss": 0.8607,
+      "step": 2360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0724825180936824e-05,
+      "loss": 0.5171,
+      "step": 2361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0717493324881125e-05,
+      "loss": 0.5667,
+      "step": 2362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0710159232694137e-05,
+      "loss": 0.6598,
+      "step": 2363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.070282290541927e-05,
+      "loss": 0.6106,
+      "step": 2364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0695484344100264e-05,
+      "loss": 0.9557,
+      "step": 2365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.068814354978118e-05,
+      "loss": 0.6279,
+      "step": 2366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.06808005235064e-05,
+      "loss": 0.5427,
+      "step": 2367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0673455266320616e-05,
+      "loss": 0.8429,
+      "step": 2368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.066610777926883e-05,
+      "loss": 0.5919,
+      "step": 2369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.0658758063396374e-05,
+      "loss": 0.472,
+      "step": 2370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.06514061197489e-05,
+      "loss": 0.6539,
+      "step": 2371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.064405194937237e-05,
+      "loss": 0.5109,
+      "step": 2372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.063669555331306e-05,
+      "loss": 0.4944,
+      "step": 2373
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.062933693261757e-05,
+      "loss": 0.584,
+      "step": 2374
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.06219760883328e-05,
+      "loss": 0.7492,
+      "step": 2375
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0614613021505995e-05,
+      "loss": 0.7253,
+      "step": 2376
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0607247733184694e-05,
+      "loss": 0.7206,
+      "step": 2377
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.059988022441676e-05,
+      "loss": 0.7834,
+      "step": 2378
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0592510496250355e-05,
+      "loss": 0.5912,
+      "step": 2379
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0585138549734e-05,
+      "loss": 0.7281,
+      "step": 2380
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.057776438591648e-05,
+      "loss": 0.7195,
+      "step": 2381
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.057038800584694e-05,
+      "loss": 0.7958,
+      "step": 2382
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0563009410574795e-05,
+      "loss": 0.7228,
+      "step": 2383
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.055562860114982e-05,
+      "loss": 0.5682,
+      "step": 2384
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0548245578622066e-05,
+      "loss": 0.6832,
+      "step": 2385
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0540860344041934e-05,
+      "loss": 0.5139,
+      "step": 2386
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.053347289846011e-05,
+      "loss": 0.9336,
+      "step": 2387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.052608324292762e-05,
+      "loss": 0.7382,
+      "step": 2388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0518691378495776e-05,
+      "loss": 0.4302,
+      "step": 2389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.051129730621623e-05,
+      "loss": 0.4995,
+      "step": 2390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.050390102714095e-05,
+      "loss": 0.6058,
+      "step": 2391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.049650254232218e-05,
+      "loss": 0.6842,
+      "step": 2392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0489101852812525e-05,
+      "loss": 0.8248,
+      "step": 2393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0481698959664873e-05,
+      "loss": 0.6428,
+      "step": 2394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.047429386393244e-05,
+      "loss": 0.7139,
+      "step": 2395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.046688656666875e-05,
+      "loss": 0.7722,
+      "step": 2396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.045947706892764e-05,
+      "loss": 0.8364,
+      "step": 2397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.045206537176326e-05,
+      "loss": 0.7249,
+      "step": 2398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.044465147623007e-05,
+      "loss": 0.7334,
+      "step": 2399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0437235383382855e-05,
+      "loss": 0.6302,
+      "step": 2400
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.04298170942767e-05,
+      "loss": 0.5319,
+      "step": 2401
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0422396609967e-05,
+      "loss": 0.9197,
+      "step": 2402
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.041497393150949e-05,
+      "loss": 0.563,
+      "step": 2403
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.040754905996017e-05,
+      "loss": 0.8293,
+      "step": 2404
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.04001219963754e-05,
+      "loss": 0.8947,
+      "step": 2405
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.039269274181181e-05,
+      "loss": 0.5416,
+      "step": 2406
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0385261297326374e-05,
+      "loss": 0.7449,
+      "step": 2407
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0377827663976366e-05,
+      "loss": 0.9165,
+      "step": 2408
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.037039184281937e-05,
+      "loss": 0.7625,
+      "step": 2409
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.036295383491327e-05,
+      "loss": 0.7902,
+      "step": 2410
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.03555136413163e-05,
+      "loss": 0.7645,
+      "step": 2411
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0348071263086944e-05,
+      "loss": 0.9807,
+      "step": 2412
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.034062670128406e-05,
+      "loss": 0.4153,
+      "step": 2413
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.033317995696676e-05,
+      "loss": 0.5171,
+      "step": 2414
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.032573103119452e-05,
+      "loss": 0.5203,
+      "step": 2415
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.031827992502708e-05,
+      "loss": 0.8772,
+      "step": 2416
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.031082663952454e-05,
+      "loss": 0.7986,
+      "step": 2417
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.030337117574724e-05,
+      "loss": 0.7586,
+      "step": 2418
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0295913534755905e-05,
+      "loss": 0.5145,
+      "step": 2419
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.028845371761151e-05,
+      "loss": 0.6337,
+      "step": 2420
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.028099172537537e-05,
+      "loss": 0.651,
+      "step": 2421
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.027352755910913e-05,
+      "loss": 0.5739,
+      "step": 2422
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.026606121987469e-05,
+      "loss": 0.6128,
+      "step": 2423
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.025859270873429e-05,
+      "loss": 0.983,
+      "step": 2424
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.025112202675049e-05,
+      "loss": 0.5428,
+      "step": 2425
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.024364917498614e-05,
+      "loss": 0.669,
+      "step": 2426
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.02361741545044e-05,
+      "loss": 0.8566,
+      "step": 2427
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.022869696636875e-05,
+      "loss": 0.968,
+      "step": 2428
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.022121761164297e-05,
+      "loss": 0.4995,
+      "step": 2429
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.021373609139114e-05,
+      "loss": 0.7646,
+      "step": 2430
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.020625240667767e-05,
+      "loss": 0.722,
+      "step": 2431
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.019876655856726e-05,
+      "loss": 0.8613,
+      "step": 2432
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.019127854812492e-05,
+      "loss": 0.703,
+      "step": 2433
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0183788376415974e-05,
+      "loss": 0.9763,
+      "step": 2434
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.017629604450606e-05,
+      "loss": 0.7824,
+      "step": 2435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.016880155346109e-05,
+      "loss": 0.5277,
+      "step": 2436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.016130490434733e-05,
+      "loss": 0.7919,
+      "step": 2437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.015380609823132e-05,
+      "loss": 0.6571,
+      "step": 2438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.014630513617992e-05,
+      "loss": 1.0636,
+      "step": 2439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.013880201926028e-05,
+      "loss": 0.5263,
+      "step": 2440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.013129674853989e-05,
+      "loss": 0.8123,
+      "step": 2441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.012378932508652e-05,
+      "loss": 0.7144,
+      "step": 2442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0116279749968254e-05,
+      "loss": 0.6352,
+      "step": 2443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.010876802425347e-05,
+      "loss": 0.6587,
+      "step": 2444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.010125414901088e-05,
+      "loss": 0.477,
+      "step": 2445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0093738125309474e-05,
+      "loss": 0.6186,
+      "step": 2446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.008621995421855e-05,
+      "loss": 0.7679,
+      "step": 2447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0078699636807735e-05,
+      "loss": 0.7163,
+      "step": 2448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0071177174146946e-05,
+      "loss": 1.0651,
+      "step": 2449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.00636525673064e-05,
+      "loss": 0.5456,
+      "step": 2450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.005612581735662e-05,
+      "loss": 0.619,
+      "step": 2451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.0048596925368455e-05,
+      "loss": 0.5295,
+      "step": 2452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.004106589241302e-05,
+      "loss": 0.6306,
+      "step": 2453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.003353271956176e-05,
+      "loss": 0.7592,
+      "step": 2454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.002599740788644e-05,
+      "loss": 0.7863,
+      "step": 2455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.00184599584591e-05,
+      "loss": 0.5084,
+      "step": 2456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.001092037235209e-05,
+      "loss": 0.775,
+      "step": 2457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.0003378650638076e-05,
+      "loss": 0.649,
+      "step": 2458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.999583479439001e-05,
+      "loss": 0.6172,
+      "step": 2459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9988288804681176e-05,
+      "loss": 0.809,
+      "step": 2460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9980740682585123e-05,
+      "loss": 0.7835,
+      "step": 2461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9973190429175736e-05,
+      "loss": 0.5746,
+      "step": 2462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.996563804552719e-05,
+      "loss": 0.8763,
+      "step": 2463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.995808353271396e-05,
+      "loss": 0.6637,
+      "step": 2464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.995052689181084e-05,
+      "loss": 0.6263,
+      "step": 2465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.99429681238929e-05,
+      "loss": 0.6414,
+      "step": 2466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9935407230035524e-05,
+      "loss": 0.3524,
+      "step": 2467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.992784421131443e-05,
+      "loss": 0.9221,
+      "step": 2468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9920279068805585e-05,
+      "loss": 0.7236,
+      "step": 2469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.991271180358529e-05,
+      "loss": 0.6216,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.990514241673015e-05,
+      "loss": 0.7122,
+      "step": 2471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.989757090931704e-05,
+      "loss": 0.6535,
+      "step": 2472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9889997282423185e-05,
+      "loss": 0.7056,
+      "step": 2473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.988242153712608e-05,
+      "loss": 0.6202,
+      "step": 2474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.987484367450353e-05,
+      "loss": 0.7981,
+      "step": 2475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.986726369563362e-05,
+      "loss": 0.7479,
+      "step": 2476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.985968160159478e-05,
+      "loss": 0.5164,
+      "step": 2477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9852097393465706e-05,
+      "loss": 0.7108,
+      "step": 2478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.98445110723254e-05,
+      "loss": 0.6525,
+      "step": 2479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9836922639253175e-05,
+      "loss": 0.5942,
+      "step": 2480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.982933209532864e-05,
+      "loss": 0.5889,
+      "step": 2481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.982173944163171e-05,
+      "loss": 0.7882,
+      "step": 2482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9814144679242574e-05,
+      "loss": 0.7197,
+      "step": 2483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.980654780924176e-05,
+      "loss": 0.8161,
+      "step": 2484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.979894883271007e-05,
+      "loss": 0.4958,
+      "step": 2485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.97913477507286e-05,
+      "loss": 0.6147,
+      "step": 2486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.978374456437878e-05,
+      "loss": 0.5544,
+      "step": 2487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.97761392747423e-05,
+      "loss": 0.5563,
+      "step": 2488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.976853188290117e-05,
+      "loss": 0.6049,
+      "step": 2489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.97609223899377e-05,
+      "loss": 0.7897,
+      "step": 2490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.97533107969345e-05,
+      "loss": 0.506,
+      "step": 2491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.974569710497445e-05,
+      "loss": 1.03,
+      "step": 2492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9738081315140776e-05,
+      "loss": 0.698,
+      "step": 2493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9730463428516964e-05,
+      "loss": 1.0293,
+      "step": 2494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.972284344618682e-05,
+      "loss": 0.6562,
+      "step": 2495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9715221369234435e-05,
+      "loss": 0.6425,
+      "step": 2496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.97075971987442e-05,
+      "loss": 0.6772,
+      "step": 2497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.969997093580082e-05,
+      "loss": 0.625,
+      "step": 2498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9692342581489285e-05,
+      "loss": 0.4758,
+      "step": 2499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9684712136894865e-05,
+      "loss": 0.6754,
+      "step": 2500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9677079603103155e-05,
+      "loss": 0.7104,
+      "step": 2501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.966944498120004e-05,
+      "loss": 0.635,
+      "step": 2502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.96618082722717e-05,
+      "loss": 0.4742,
+      "step": 2503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9654169477404614e-05,
+      "loss": 0.618,
+      "step": 2504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9646528597685554e-05,
+      "loss": 0.7185,
+      "step": 2505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.963888563420157e-05,
+      "loss": 0.7195,
+      "step": 2506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.963124058804005e-05,
+      "loss": 0.8871,
+      "step": 2507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.962359346028865e-05,
+      "loss": 0.5861,
+      "step": 2508
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.961594425203534e-05,
+      "loss": 0.6723,
+      "step": 2509
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.960829296436835e-05,
+      "loss": 0.7822,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9600639598376256e-05,
+      "loss": 0.8887,
+      "step": 2511
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.959298415514788e-05,
+      "loss": 0.6992,
+      "step": 2512
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.958532663577238e-05,
+      "loss": 0.5806,
+      "step": 2513
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9577667041339194e-05,
+      "loss": 0.7094,
+      "step": 2514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.957000537293804e-05,
+      "loss": 0.525,
+      "step": 2515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.956234163165896e-05,
+      "loss": 0.7882,
+      "step": 2516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9554675818592265e-05,
+      "loss": 0.4594,
+      "step": 2517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.954700793482858e-05,
+      "loss": 0.6924,
+      "step": 2518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.953933798145882e-05,
+      "loss": 0.6615,
+      "step": 2519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.953166595957418e-05,
+      "loss": 0.7736,
+      "step": 2520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9523991870266165e-05,
+      "loss": 0.5715,
+      "step": 2521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9516315714626574e-05,
+      "loss": 0.7061,
+      "step": 2522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.950863749374748e-05,
+      "loss": 0.6099,
+      "step": 2523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.950095720872129e-05,
+      "loss": 0.6954,
+      "step": 2524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.949327486064065e-05,
+      "loss": 0.6324,
+      "step": 2525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.948559045059856e-05,
+      "loss": 0.6621,
+      "step": 2526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9477903979688266e-05,
+      "loss": 0.6295,
+      "step": 2527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.947021544900331e-05,
+      "loss": 0.4116,
+      "step": 2528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.946252485963757e-05,
+      "loss": 0.7061,
+      "step": 2529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.945483221268518e-05,
+      "loss": 0.683,
+      "step": 2530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.944713750924057e-05,
+      "loss": 0.875,
+      "step": 2531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.943944075039846e-05,
+      "loss": 0.9017,
+      "step": 2532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.943174193725387e-05,
+      "loss": 0.764,
+      "step": 2533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.942404107090213e-05,
+      "loss": 0.7569,
+      "step": 2534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.941633815243883e-05,
+      "loss": 0.8578,
+      "step": 2535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.9408633182959866e-05,
+      "loss": 0.5188,
+      "step": 2536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.940092616356143e-05,
+      "loss": 0.6326,
+      "step": 2537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.939321709533999e-05,
+      "loss": 0.7724,
+      "step": 2538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.938550597939233e-05,
+      "loss": 0.6108,
+      "step": 2539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.937779281681552e-05,
+      "loss": 0.6147,
+      "step": 2540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.937007760870689e-05,
+      "loss": 0.5741,
+      "step": 2541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.93623603561641e-05,
+      "loss": 0.6517,
+      "step": 2542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.935464106028508e-05,
+      "loss": 0.4843,
+      "step": 2543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.934691972216805e-05,
+      "loss": 0.7802,
+      "step": 2544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9339196342911545e-05,
+      "loss": 0.6172,
+      "step": 2545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9331470923614354e-05,
+      "loss": 0.742,
+      "step": 2546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.932374346537559e-05,
+      "loss": 0.6406,
+      "step": 2547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.931601396929462e-05,
+      "loss": 0.9194,
+      "step": 2548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.930828243647113e-05,
+      "loss": 0.5681,
+      "step": 2549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.93005488680051e-05,
+      "loss": 0.5702,
+      "step": 2550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.929281326499678e-05,
+      "loss": 0.5389,
+      "step": 2551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.92850756285467e-05,
+      "loss": 0.6436,
+      "step": 2552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.927733595975571e-05,
+      "loss": 0.6431,
+      "step": 2553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.926959425972493e-05,
+      "loss": 0.4589,
+      "step": 2554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.926185052955578e-05,
+      "loss": 0.6913,
+      "step": 2555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.925410477034995e-05,
+      "loss": 0.6344,
+      "step": 2556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9246356983209435e-05,
+      "loss": 0.7958,
+      "step": 2557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.923860716923653e-05,
+      "loss": 0.4288,
+      "step": 2558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.923085532953378e-05,
+      "loss": 0.528,
+      "step": 2559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.922310146520406e-05,
+      "loss": 0.493,
+      "step": 2560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.921534557735049e-05,
+      "loss": 0.6804,
+      "step": 2561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.920758766707653e-05,
+      "loss": 0.8572,
+      "step": 2562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9199827735485875e-05,
+      "loss": 0.783,
+      "step": 2563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.919206578368255e-05,
+      "loss": 0.588,
+      "step": 2564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.918430181277085e-05,
+      "loss": 0.4423,
+      "step": 2565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.917653582385534e-05,
+      "loss": 0.5013,
+      "step": 2566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.916876781804091e-05,
+      "loss": 0.3223,
+      "step": 2567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9160997796432695e-05,
+      "loss": 0.6645,
+      "step": 2568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.915322576013615e-05,
+      "loss": 0.6383,
+      "step": 2569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.914545171025701e-05,
+      "loss": 0.5737,
+      "step": 2570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9137675647901276e-05,
+      "loss": 0.4673,
+      "step": 2571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9129897574175254e-05,
+      "loss": 0.705,
+      "step": 2572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.912211749018554e-05,
+      "loss": 0.7328,
+      "step": 2573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9114335397039005e-05,
+      "loss": 0.5746,
+      "step": 2574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.91065512958428e-05,
+      "loss": 0.6533,
+      "step": 2575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.909876518770438e-05,
+      "loss": 0.7614,
+      "step": 2576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.909097707373148e-05,
+      "loss": 0.97,
+      "step": 2577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9083186955032115e-05,
+      "loss": 0.6569,
+      "step": 2578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9075394832714586e-05,
+      "loss": 0.8568,
+      "step": 2579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.906760070788746e-05,
+      "loss": 0.6383,
+      "step": 2580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.9059804581659645e-05,
+      "loss": 0.9014,
+      "step": 2581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.905200645514028e-05,
+      "loss": 0.7275,
+      "step": 2582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.904420632943879e-05,
+      "loss": 0.548,
+      "step": 2583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.903640420566492e-05,
+      "loss": 0.7095,
+      "step": 2584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.902860008492869e-05,
+      "loss": 0.5638,
+      "step": 2585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.902079396834036e-05,
+      "loss": 0.6216,
+      "step": 2586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.901298585701054e-05,
+      "loss": 0.531,
+      "step": 2587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.900517575205008e-05,
+      "loss": 0.7057,
+      "step": 2588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.899736365457013e-05,
+      "loss": 0.7872,
+      "step": 2589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8989549565682095e-05,
+      "loss": 0.841,
+      "step": 2590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.898173348649772e-05,
+      "loss": 0.7718,
+      "step": 2591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.897391541812898e-05,
+      "loss": 0.618,
+      "step": 2592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.896609536168817e-05,
+      "loss": 0.6082,
+      "step": 2593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.895827331828783e-05,
+      "loss": 0.578,
+      "step": 2594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8950449289040825e-05,
+      "loss": 0.493,
+      "step": 2595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.894262327506027e-05,
+      "loss": 0.4481,
+      "step": 2596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.893479527745956e-05,
+      "loss": 0.7713,
+      "step": 2597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.892696529735241e-05,
+      "loss": 0.5155,
+      "step": 2598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.891913333585279e-05,
+      "loss": 0.7483,
+      "step": 2599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.891129939407494e-05,
+      "loss": 0.8358,
+      "step": 2600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.890346347313341e-05,
+      "loss": 0.6003,
+      "step": 2601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.889562557414302e-05,
+      "loss": 0.6088,
+      "step": 2602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.888778569821885e-05,
+      "loss": 1.0859,
+      "step": 2603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.88799438464763e-05,
+      "loss": 0.7432,
+      "step": 2604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.887210002003103e-05,
+      "loss": 0.7042,
+      "step": 2605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8864254219998974e-05,
+      "loss": 0.9741,
+      "step": 2606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.885640644749637e-05,
+      "loss": 1.056,
+      "step": 2607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.884855670363971e-05,
+      "loss": 0.6169,
+      "step": 2608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.884070498954577e-05,
+      "loss": 0.5625,
+      "step": 2609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.883285130633164e-05,
+      "loss": 0.6396,
+      "step": 2610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.882499565511465e-05,
+      "loss": 0.5973,
+      "step": 2611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.881713803701242e-05,
+      "loss": 0.8805,
+      "step": 2612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8809278453142875e-05,
+      "loss": 0.6791,
+      "step": 2613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.880141690462419e-05,
+      "loss": 0.6414,
+      "step": 2614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.879355339257481e-05,
+      "loss": 0.8059,
+      "step": 2615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8785687918113514e-05,
+      "loss": 0.707,
+      "step": 2616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.87778204823593e-05,
+      "loss": 0.5765,
+      "step": 2617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.876995108643147e-05,
+      "loss": 0.8853,
+      "step": 2618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.876207973144961e-05,
+      "loss": 0.5136,
+      "step": 2619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.875420641853359e-05,
+      "loss": 0.8297,
+      "step": 2620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.874633114880353e-05,
+      "loss": 0.6517,
+      "step": 2621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.8738453923379855e-05,
+      "loss": 0.8294,
+      "step": 2622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.873057474338327e-05,
+      "loss": 0.5648,
+      "step": 2623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8722693609934726e-05,
+      "loss": 0.4699,
+      "step": 2624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8714810524155485e-05,
+      "loss": 0.6408,
+      "step": 2625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8706925487167085e-05,
+      "loss": 0.4747,
+      "step": 2626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.869903850009131e-05,
+      "loss": 0.6672,
+      "step": 2627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8691149564050265e-05,
+      "loss": 0.6965,
+      "step": 2628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.86832586801663e-05,
+      "loss": 0.4947,
+      "step": 2629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.867536584956205e-05,
+      "loss": 0.461,
+      "step": 2630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.866747107336044e-05,
+      "loss": 0.9373,
+      "step": 2631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8659574352684645e-05,
+      "loss": 0.6423,
+      "step": 2632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.865167568865815e-05,
+      "loss": 0.8575,
+      "step": 2633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.86437750824047e-05,
+      "loss": 0.6346,
+      "step": 2634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.86358725350483e-05,
+      "loss": 0.869,
+      "step": 2635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.862796804771326e-05,
+      "loss": 0.6812,
+      "step": 2636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.862006162152414e-05,
+      "loss": 0.4408,
+      "step": 2637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.861215325760581e-05,
+      "loss": 0.5984,
+      "step": 2638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.860424295708338e-05,
+      "loss": 0.5109,
+      "step": 2639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8596330721082255e-05,
+      "loss": 0.5807,
+      "step": 2640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.858841655072811e-05,
+      "loss": 0.9098,
+      "step": 2641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.858050044714689e-05,
+      "loss": 0.5787,
+      "step": 2642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8572582411464836e-05,
+      "loss": 0.7354,
+      "step": 2643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.856466244480843e-05,
+      "loss": 0.798,
+      "step": 2644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8556740548304456e-05,
+      "loss": 0.6522,
+      "step": 2645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.854881672307997e-05,
+      "loss": 0.4611,
+      "step": 2646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8540890970262284e-05,
+      "loss": 0.6899,
+      "step": 2647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.853296329097901e-05,
+      "loss": 1.1266,
+      "step": 2648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8525033686358005e-05,
+      "loss": 0.559,
+      "step": 2649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.851710215752743e-05,
+      "loss": 0.469,
+      "step": 2650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8509168705615694e-05,
+      "loss": 0.754,
+      "step": 2651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8501233331751506e-05,
+      "loss": 0.5589,
+      "step": 2652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.849329603706382e-05,
+      "loss": 0.7191,
+      "step": 2653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8485356822681884e-05,
+      "loss": 0.5906,
+      "step": 2654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.84774156897352e-05,
+      "loss": 0.5211,
+      "step": 2655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.846947263935358e-05,
+      "loss": 0.5771,
+      "step": 2656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.846152767266705e-05,
+      "loss": 0.8008,
+      "step": 2657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.845358079080596e-05,
+      "loss": 0.489,
+      "step": 2658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.844563199490092e-05,
+      "loss": 0.7632,
+      "step": 2659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8437681286082804e-05,
+      "loss": 1.0475,
+      "step": 2660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.842972866548275e-05,
+      "loss": 0.5656,
+      "step": 2661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.842177413423219e-05,
+      "loss": 0.8419,
+      "step": 2662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.841381769346282e-05,
+      "loss": 0.555,
+      "step": 2663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.840585934430659e-05,
+      "loss": 0.6777,
+      "step": 2664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8397899087895746e-05,
+      "loss": 0.8156,
+      "step": 2665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.83899369253628e-05,
+      "loss": 0.5925,
+      "step": 2666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.838197285784052e-05,
+      "loss": 0.6559,
+      "step": 2667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8374006886461964e-05,
+      "loss": 0.6079,
+      "step": 2668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8366039012360445e-05,
+      "loss": 0.5466,
+      "step": 2669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.835806923666957e-05,
+      "loss": 0.8868,
+      "step": 2670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.835009756052317e-05,
+      "loss": 0.8468,
+      "step": 2671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8342123985055404e-05,
+      "loss": 0.7559,
+      "step": 2672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.833414851140067e-05,
+      "loss": 0.8433,
+      "step": 2673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8326171140693636e-05,
+      "loss": 0.7211,
+      "step": 2674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.831819187406924e-05,
+      "loss": 0.7914,
+      "step": 2675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.831021071266271e-05,
+      "loss": 0.6204,
+      "step": 2676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8302227657609514e-05,
+      "loss": 0.389,
+      "step": 2677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.82942427100454e-05,
+      "loss": 0.8285,
+      "step": 2678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.828625587110639e-05,
+      "loss": 0.6431,
+      "step": 2679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.827826714192879e-05,
+      "loss": 0.4994,
+      "step": 2680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.827027652364914e-05,
+      "loss": 0.7472,
+      "step": 2681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.826228401740428e-05,
+      "loss": 0.6873,
+      "step": 2682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.825428962433129e-05,
+      "loss": 0.7043,
+      "step": 2683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.824629334556755e-05,
+      "loss": 0.7087,
+      "step": 2684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8238295182250674e-05,
+      "loss": 0.652,
+      "step": 2685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.823029513551858e-05,
+      "loss": 0.7394,
+      "step": 2686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.822229320650944e-05,
+      "loss": 0.7734,
+      "step": 2687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.821428939636167e-05,
+      "loss": 0.6036,
+      "step": 2688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8206283706213985e-05,
+      "loss": 0.601,
+      "step": 2689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8198276137205356e-05,
+      "loss": 0.9773,
+      "step": 2690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.819026669047503e-05,
+      "loss": 0.6387,
+      "step": 2691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.818225536716248e-05,
+      "loss": 0.5377,
+      "step": 2692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.817424216840752e-05,
+      "loss": 0.7297,
+      "step": 2693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.816622709535017e-05,
+      "loss": 0.7809,
+      "step": 2694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8158210149130746e-05,
+      "loss": 0.5213,
+      "step": 2695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8150191330889804e-05,
+      "loss": 0.459,
+      "step": 2696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8142170641768194e-05,
+      "loss": 0.7204,
+      "step": 2697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8134148082907025e-05,
+      "loss": 0.6685,
+      "step": 2698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.812612365544765e-05,
+      "loss": 0.518,
+      "step": 2699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.811809736053172e-05,
+      "loss": 0.6083,
+      "step": 2700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8110069199301155e-05,
+      "loss": 0.6919,
+      "step": 2701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.810203917289809e-05,
+      "loss": 0.7114,
+      "step": 2702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.809400728246498e-05,
+      "loss": 0.9591,
+      "step": 2703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.808597352914451e-05,
+      "loss": 0.5937,
+      "step": 2704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.807793791407966e-05,
+      "loss": 0.8738,
+      "step": 2705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.806990043841364e-05,
+      "loss": 0.6562,
+      "step": 2706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.8061861103289965e-05,
+      "loss": 0.6571,
+      "step": 2707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8053819909852384e-05,
+      "loss": 0.4506,
+      "step": 2708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.804577685924491e-05,
+      "loss": 0.7006,
+      "step": 2709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.803773195261183e-05,
+      "loss": 0.5556,
+      "step": 2710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8029685191097706e-05,
+      "loss": 0.6713,
+      "step": 2711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8021636575847354e-05,
+      "loss": 0.7214,
+      "step": 2712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.801358610800584e-05,
+      "loss": 0.7687,
+      "step": 2713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.8005533788718514e-05,
+      "loss": 0.768,
+      "step": 2714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.799747961913098e-05,
+      "loss": 0.8995,
+      "step": 2715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7989423600389105e-05,
+      "loss": 0.5768,
+      "step": 2716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7981365733639027e-05,
+      "loss": 0.4163,
+      "step": 2717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.797330602002712e-05,
+      "loss": 0.4951,
+      "step": 2718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.796524446070007e-05,
+      "loss": 0.6784,
+      "step": 2719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.795718105680478e-05,
+      "loss": 0.8068,
+      "step": 2720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7949115809488435e-05,
+      "loss": 0.7086,
+      "step": 2721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.794104871989848e-05,
+      "loss": 0.4314,
+      "step": 2722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.793297978918262e-05,
+      "loss": 0.662,
+      "step": 2723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.792490901848882e-05,
+      "loss": 0.5063,
+      "step": 2724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.791683640896533e-05,
+      "loss": 0.5847,
+      "step": 2725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.790876196176061e-05,
+      "loss": 0.5283,
+      "step": 2726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.790068567802344e-05,
+      "loss": 0.6434,
+      "step": 2727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.789260755890283e-05,
+      "loss": 0.5435,
+      "step": 2728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.788452760554805e-05,
+      "loss": 0.9434,
+      "step": 2729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.787644581910863e-05,
+      "loss": 0.6299,
+      "step": 2730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.786836220073439e-05,
+      "loss": 0.8817,
+      "step": 2731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.786027675157535e-05,
+      "loss": 0.7778,
+      "step": 2732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7852189472781875e-05,
+      "loss": 0.7773,
+      "step": 2733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.784410036550451e-05,
+      "loss": 0.577,
+      "step": 2734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.783600943089411e-05,
+      "loss": 0.4221,
+      "step": 2735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.782791667010177e-05,
+      "loss": 0.6835,
+      "step": 2736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.781982208427885e-05,
+      "loss": 0.8547,
+      "step": 2737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.781172567457696e-05,
+      "loss": 0.5363,
+      "step": 2738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7803627442147985e-05,
+      "loss": 0.605,
+      "step": 2739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.779552738814407e-05,
+      "loss": 0.9648,
+      "step": 2740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.77874255137176e-05,
+      "loss": 0.8953,
+      "step": 2741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.777932182002124e-05,
+      "loss": 0.4243,
+      "step": 2742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.77712163082079e-05,
+      "loss": 0.7248,
+      "step": 2743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.776310897943074e-05,
+      "loss": 0.6855,
+      "step": 2744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.775499983484321e-05,
+      "loss": 0.9333,
+      "step": 2745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7746888875598994e-05,
+      "loss": 0.6081,
+      "step": 2746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.773877610285204e-05,
+      "loss": 0.5649,
+      "step": 2747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.773066151775655e-05,
+      "loss": 0.7648,
+      "step": 2748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.772254512146699e-05,
+      "loss": 0.499,
+      "step": 2749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.771442691513809e-05,
+      "loss": 0.5646,
+      "step": 2750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.770630689992481e-05,
+      "loss": 0.7344,
+      "step": 2751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.769818507698241e-05,
+      "loss": 0.7385,
+      "step": 2752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7690061447466364e-05,
+      "loss": 0.6704,
+      "step": 2753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7681936012532436e-05,
+      "loss": 0.707,
+      "step": 2754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7673808773336616e-05,
+      "loss": 0.6812,
+      "step": 2755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7665679731035184e-05,
+      "loss": 0.9701,
+      "step": 2756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7657548886784665e-05,
+      "loss": 0.6479,
+      "step": 2757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7649416241741824e-05,
+      "loss": 0.8295,
+      "step": 2758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.76412817970637e-05,
+      "loss": 0.7617,
+      "step": 2759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.763314555390757e-05,
+      "loss": 0.8988,
+      "step": 2760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.762500751343099e-05,
+      "loss": 0.6256,
+      "step": 2761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7616867676791765e-05,
+      "loss": 0.6653,
+      "step": 2762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.760872604514795e-05,
+      "loss": 0.7506,
+      "step": 2763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.760058261965784e-05,
+      "loss": 0.5078,
+      "step": 2764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.759243740148003e-05,
+      "loss": 0.816,
+      "step": 2765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.758429039177332e-05,
+      "loss": 0.7321,
+      "step": 2766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.757614159169679e-05,
+      "loss": 0.8909,
+      "step": 2767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.756799100240979e-05,
+      "loss": 0.5751,
+      "step": 2768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7559838625071876e-05,
+      "loss": 0.9646,
+      "step": 2769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.755168446084291e-05,
+      "loss": 0.6004,
+      "step": 2770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7543528510882976e-05,
+      "loss": 0.4773,
+      "step": 2771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.753537077635244e-05,
+      "loss": 0.5948,
+      "step": 2772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7527211258411875e-05,
+      "loss": 0.6647,
+      "step": 2773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.751904995822216e-05,
+      "loss": 0.6585,
+      "step": 2774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.751088687694442e-05,
+      "loss": 0.7532,
+      "step": 2775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7502722015739976e-05,
+      "loss": 0.7076,
+      "step": 2776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.749455537577048e-05,
+      "loss": 0.7511,
+      "step": 2777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.748638695819778e-05,
+      "loss": 0.832,
+      "step": 2778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.747821676418402e-05,
+      "loss": 0.7925,
+      "step": 2779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.747004479489154e-05,
+      "loss": 0.852,
+      "step": 2780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7461871051483e-05,
+      "loss": 0.8973,
+      "step": 2781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.745369553512128e-05,
+      "loss": 0.7876,
+      "step": 2782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.74455182469695e-05,
+      "loss": 0.686,
+      "step": 2783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7437339188191044e-05,
+      "loss": 0.6577,
+      "step": 2784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7429158359949554e-05,
+      "loss": 0.5717,
+      "step": 2785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.742097576340892e-05,
+      "loss": 0.5419,
+      "step": 2786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.741279139973327e-05,
+      "loss": 0.5724,
+      "step": 2787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.740460527008701e-05,
+      "loss": 0.5962,
+      "step": 2788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7396417375634785e-05,
+      "loss": 0.5967,
+      "step": 2789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.738822771754147e-05,
+      "loss": 0.6352,
+      "step": 2790
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.738003629697222e-05,
+      "loss": 0.8307,
+      "step": 2791
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.737184311509243e-05,
+      "loss": 0.6812,
+      "step": 2792
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.736364817306775e-05,
+      "loss": 0.7958,
+      "step": 2793
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.735545147206406e-05,
+      "loss": 0.4678,
+      "step": 2794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7347253013247526e-05,
+      "loss": 0.6414,
+      "step": 2795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7339052797784546e-05,
+      "loss": 0.6887,
+      "step": 2796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7330850826841744e-05,
+      "loss": 0.7626,
+      "step": 2797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.732264710158603e-05,
+      "loss": 0.7043,
+      "step": 2798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.731444162318456e-05,
+      "loss": 0.7516,
+      "step": 2799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.730623439280472e-05,
+      "loss": 0.7308,
+      "step": 2800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.729802541161413e-05,
+      "loss": 0.5345,
+      "step": 2801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7289814680780724e-05,
+      "loss": 0.6216,
+      "step": 2802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.728160220147263e-05,
+      "loss": 0.4969,
+      "step": 2803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7273387974858234e-05,
+      "loss": 0.556,
+      "step": 2804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.726517200210617e-05,
+      "loss": 0.8325,
+      "step": 2805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.725695428438535e-05,
+      "loss": 0.5924,
+      "step": 2806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.724873482286488e-05,
+      "loss": 1.0293,
+      "step": 2807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7240513618714165e-05,
+      "loss": 0.9606,
+      "step": 2808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7232290673102835e-05,
+      "loss": 0.7468,
+      "step": 2809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.722406598720077e-05,
+      "loss": 0.6295,
+      "step": 2810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7215839562178095e-05,
+      "loss": 0.651,
+      "step": 2811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.720761139920518e-05,
+      "loss": 0.9893,
+      "step": 2812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.719938149945266e-05,
+      "loss": 0.9181,
+      "step": 2813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7191149864091396e-05,
+      "loss": 0.7479,
+      "step": 2814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.718291649429251e-05,
+      "loss": 0.8657,
+      "step": 2815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.717468139122736e-05,
+      "loss": 0.7119,
+      "step": 2816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.716644455606756e-05,
+      "loss": 0.7604,
+      "step": 2817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7158205989984956e-05,
+      "loss": 0.5936,
+      "step": 2818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7149965694151665e-05,
+      "loss": 0.7683,
+      "step": 2819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7141723669740034e-05,
+      "loss": 0.855,
+      "step": 2820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.713347991792264e-05,
+      "loss": 0.8133,
+      "step": 2821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7125234439872356e-05,
+      "loss": 0.7094,
+      "step": 2822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.711698723676224e-05,
+      "loss": 0.9487,
+      "step": 2823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.710873830976563e-05,
+      "loss": 0.8025,
+      "step": 2824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.710048766005611e-05,
+      "loss": 0.5738,
+      "step": 2825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.709223528880749e-05,
+      "loss": 0.6536,
+      "step": 2826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.708398119719385e-05,
+      "loss": 0.8846,
+      "step": 2827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.707572538638949e-05,
+      "loss": 0.5031,
+      "step": 2828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.706746785756898e-05,
+      "loss": 0.6017,
+      "step": 2829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7059208611907105e-05,
+      "loss": 0.9059,
+      "step": 2830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.705094765057892e-05,
+      "loss": 0.6933,
+      "step": 2831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.704268497475971e-05,
+      "loss": 0.676,
+      "step": 2832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.703442058562501e-05,
+      "loss": 0.7788,
+      "step": 2833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7026154484350585e-05,
+      "loss": 0.7392,
+      "step": 2834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.701788667211247e-05,
+      "loss": 0.6644,
+      "step": 2835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7009617150086925e-05,
+      "loss": 0.6099,
+      "step": 2836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.700134591945045e-05,
+      "loss": 0.6745,
+      "step": 2837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6993072981379806e-05,
+      "loss": 0.6029,
+      "step": 2838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.698479833705198e-05,
+      "loss": 0.5898,
+      "step": 2839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.69765219876442e-05,
+      "loss": 0.6628,
+      "step": 2840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.696824393433395e-05,
+      "loss": 0.7465,
+      "step": 2841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.695996417829896e-05,
+      "loss": 0.5885,
+      "step": 2842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6951682720717185e-05,
+      "loss": 0.8649,
+      "step": 2843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.694339956276682e-05,
+      "loss": 0.8485,
+      "step": 2844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.693511470562633e-05,
+      "loss": 0.7074,
+      "step": 2845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.692682815047439e-05,
+      "loss": 0.6281,
+      "step": 2846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.691853989848994e-05,
+      "loss": 0.776,
+      "step": 2847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.691024995085213e-05,
+      "loss": 0.8448,
+      "step": 2848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6901958308740404e-05,
+      "loss": 0.7408,
+      "step": 2849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.689366497333439e-05,
+      "loss": 0.5466,
+      "step": 2850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6885369945814e-05,
+      "loss": 0.6278,
+      "step": 2851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.687707322735936e-05,
+      "loss": 0.772,
+      "step": 2852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6868774819150856e-05,
+      "loss": 0.678,
+      "step": 2853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.686047472236909e-05,
+      "loss": 0.5637,
+      "step": 2854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.685217293819491e-05,
+      "loss": 0.6482,
+      "step": 2855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.684386946780945e-05,
+      "loss": 0.933,
+      "step": 2856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.683556431239402e-05,
+      "loss": 0.6661,
+      "step": 2857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6827257473130197e-05,
+      "loss": 0.6484,
+      "step": 2858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6818948951199804e-05,
+      "loss": 0.9359,
+      "step": 2859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.68106387477849e-05,
+      "loss": 0.722,
+      "step": 2860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.680232686406776e-05,
+      "loss": 0.7545,
+      "step": 2861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.679401330123093e-05,
+      "loss": 0.8303,
+      "step": 2862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.67856980604572e-05,
+      "loss": 0.7428,
+      "step": 2863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6777381142929545e-05,
+      "loss": 0.7196,
+      "step": 2864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.676906254983124e-05,
+      "loss": 0.6098,
+      "step": 2865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6760742282345765e-05,
+      "loss": 0.7495,
+      "step": 2866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.675242034165686e-05,
+      "loss": 0.6038,
+      "step": 2867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6744096728948465e-05,
+      "loss": 0.9131,
+      "step": 2868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.673577144540479e-05,
+      "loss": 0.9127,
+      "step": 2869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.6727444492210294e-05,
+      "loss": 0.7257,
+      "step": 2870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.671911587054964e-05,
+      "loss": 0.5077,
+      "step": 2871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.671078558160773e-05,
+      "loss": 0.8535,
+      "step": 2872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.670245362656974e-05,
+      "loss": 0.6142,
+      "step": 2873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6694120006621046e-05,
+      "loss": 0.5792,
+      "step": 2874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.668578472294727e-05,
+      "loss": 0.5895,
+      "step": 2875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6677447776734285e-05,
+      "loss": 0.8948,
+      "step": 2876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.666910916916819e-05,
+      "loss": 0.7324,
+      "step": 2877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.666076890143531e-05,
+      "loss": 0.8491,
+      "step": 2878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.665242697472223e-05,
+      "loss": 0.8911,
+      "step": 2879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.664408339021574e-05,
+      "loss": 0.8235,
+      "step": 2880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.663573814910291e-05,
+      "loss": 0.4988,
+      "step": 2881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.662739125257099e-05,
+      "loss": 0.6412,
+      "step": 2882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.661904270180751e-05,
+      "loss": 0.7839,
+      "step": 2883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.661069249800024e-05,
+      "loss": 0.6426,
+      "step": 2884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6602340642337125e-05,
+      "loss": 0.7134,
+      "step": 2885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.659398713600641e-05,
+      "loss": 0.4237,
+      "step": 2886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6585631980196544e-05,
+      "loss": 0.8333,
+      "step": 2887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.657727517609622e-05,
+      "loss": 0.7745,
+      "step": 2888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.656891672489435e-05,
+      "loss": 0.9662,
+      "step": 2889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.656055662778012e-05,
+      "loss": 0.8806,
+      "step": 2890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6552194885942906e-05,
+      "loss": 0.8057,
+      "step": 2891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.654383150057232e-05,
+      "loss": 0.791,
+      "step": 2892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6535466472858247e-05,
+      "loss": 0.7636,
+      "step": 2893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.652709980399077e-05,
+      "loss": 0.817,
+      "step": 2894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.651873149516022e-05,
+      "loss": 0.8285,
+      "step": 2895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6510361547557146e-05,
+      "loss": 0.7599,
+      "step": 2896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.650198996237237e-05,
+      "loss": 0.8304,
+      "step": 2897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.64936167407969e-05,
+      "loss": 0.8739,
+      "step": 2898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6485241884021995e-05,
+      "loss": 0.7522,
+      "step": 2899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6476865393239146e-05,
+      "loss": 0.7962,
+      "step": 2900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.646848726964009e-05,
+      "loss": 0.7927,
+      "step": 2901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.646010751441678e-05,
+      "loss": 0.9263,
+      "step": 2902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.64517261287614e-05,
+      "loss": 0.6122,
+      "step": 2903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6443343113866376e-05,
+      "loss": 0.8286,
+      "step": 2904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.643495847092436e-05,
+      "loss": 0.659,
+      "step": 2905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.642657220112824e-05,
+      "loss": 0.7726,
+      "step": 2906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.641818430567113e-05,
+      "loss": 0.704,
+      "step": 2907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.640979478574638e-05,
+      "loss": 0.6997,
+      "step": 2908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6401403642547574e-05,
+      "loss": 0.8721,
+      "step": 2909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6393010877268496e-05,
+      "loss": 0.9562,
+      "step": 2910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.638461649110322e-05,
+      "loss": 0.7317,
+      "step": 2911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6376220485246006e-05,
+      "loss": 0.9703,
+      "step": 2912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.636782286089134e-05,
+      "loss": 0.8137,
+      "step": 2913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.635942361923397e-05,
+      "loss": 0.8864,
+      "step": 2914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6351022761468855e-05,
+      "loss": 0.7638,
+      "step": 2915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.634262028879119e-05,
+      "loss": 0.9533,
+      "step": 2916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.633421620239639e-05,
+      "loss": 1.0607,
+      "step": 2917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.632581050348011e-05,
+      "loss": 0.8084,
+      "step": 2918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.631740319323823e-05,
+      "loss": 0.9572,
+      "step": 2919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.630899427286686e-05,
+      "loss": 0.7981,
+      "step": 2920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.630058374356233e-05,
+      "loss": 1.2733,
+      "step": 2921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.629217160652123e-05,
+      "loss": 0.6949,
+      "step": 2922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.628375786294034e-05,
+      "loss": 0.8522,
+      "step": 2923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.627534251401668e-05,
+      "loss": 0.7594,
+      "step": 2924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.626692556094752e-05,
+      "loss": 0.7682,
+      "step": 2925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.625850700493034e-05,
+      "loss": 0.795,
+      "step": 2926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.625008684716283e-05,
+      "loss": 0.962,
+      "step": 2927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.624166508884295e-05,
+      "loss": 0.835,
+      "step": 2928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.623324173116885e-05,
+      "loss": 0.6631,
+      "step": 2929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.622481677533893e-05,
+      "loss": 0.9072,
+      "step": 2930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.621639022255181e-05,
+      "loss": 0.7007,
+      "step": 2931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.620796207400634e-05,
+      "loss": 0.8173,
+      "step": 2932
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6199532330901605e-05,
+      "loss": 0.8747,
+      "step": 2933
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.619110099443688e-05,
+      "loss": 0.7843,
+      "step": 2934
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.618266806581171e-05,
+      "loss": 1.1499,
+      "step": 2935
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6174233546225846e-05,
+      "loss": 0.9476,
+      "step": 2936
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.616579743687927e-05,
+      "loss": 0.9727,
+      "step": 2937
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.615735973897218e-05,
+      "loss": 0.7455,
+      "step": 2938
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.614892045370503e-05,
+      "loss": 0.6454,
+      "step": 2939
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.614047958227846e-05,
+      "loss": 1.0161,
+      "step": 2940
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.613203712589337e-05,
+      "loss": 0.8823,
+      "step": 2941
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.612359308575085e-05,
+      "loss": 0.8778,
+      "step": 2942
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6115147463052256e-05,
+      "loss": 0.6868,
+      "step": 2943
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.610670025899913e-05,
+      "loss": 1.2584,
+      "step": 2944
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.609825147479328e-05,
+      "loss": 1.1538,
+      "step": 2945
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.60898011116367e-05,
+      "loss": 1.0522,
+      "step": 2946
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6081349170731624e-05,
+      "loss": 0.7538,
+      "step": 2947
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.607289565328052e-05,
+      "loss": 0.9155,
+      "step": 2948
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.606444056048607e-05,
+      "loss": 0.6066,
+      "step": 2949
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.605598389355117e-05,
+      "loss": 0.8431,
+      "step": 2950
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.604752565367897e-05,
+      "loss": 0.884,
+      "step": 2951
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.603906584207282e-05,
+      "loss": 0.8853,
+      "step": 2952
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.603060445993629e-05,
+      "loss": 0.7566,
+      "step": 2953
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.60221415084732e-05,
+      "loss": 0.9189,
+      "step": 2954
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.6013676988887566e-05,
+      "loss": 0.7329,
+      "step": 2955
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.600521090238363e-05,
+      "loss": 0.7664,
+      "step": 2956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5996743250165866e-05,
+      "loss": 0.9304,
+      "step": 2957
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.598827403343898e-05,
+      "loss": 0.9553,
+      "step": 2958
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.597980325340788e-05,
+      "loss": 0.778,
+      "step": 2959
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.597133091127772e-05,
+      "loss": 0.9108,
+      "step": 2960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5962857008253845e-05,
+      "loss": 1.0444,
+      "step": 2961
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5954381545541846e-05,
+      "loss": 0.9401,
+      "step": 2962
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5945904524347515e-05,
+      "loss": 0.8757,
+      "step": 2963
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.59374259458769e-05,
+      "loss": 0.8511,
+      "step": 2964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.592894581133625e-05,
+      "loss": 0.7577,
+      "step": 2965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.592046412193202e-05,
+      "loss": 0.6393,
+      "step": 2966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.591198087887092e-05,
+      "loss": 0.9246,
+      "step": 2967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.590349608335984e-05,
+      "loss": 0.8555,
+      "step": 2968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5895009736605945e-05,
+      "loss": 1.1133,
+      "step": 2969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.588652183981656e-05,
+      "loss": 1.0718,
+      "step": 2970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5878032394199266e-05,
+      "loss": 0.6818,
+      "step": 2971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5869541400961875e-05,
+      "loss": 0.8507,
+      "step": 2972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.586104886131239e-05,
+      "loss": 0.9318,
+      "step": 2973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.585255477645905e-05,
+      "loss": 1.0208,
+      "step": 2974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5844059147610296e-05,
+      "loss": 0.9224,
+      "step": 2975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.583556197597483e-05,
+      "loss": 0.9913,
+      "step": 2976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.582706326276152e-05,
+      "loss": 0.9751,
+      "step": 2977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.581856300917949e-05,
+      "loss": 0.759,
+      "step": 2978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5810061216438085e-05,
+      "loss": 0.759,
+      "step": 2979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.580155788574684e-05,
+      "loss": 0.7811,
+      "step": 2980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5793053018315535e-05,
+      "loss": 0.843,
+      "step": 2981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.578454661535415e-05,
+      "loss": 1.1183,
+      "step": 2982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5776038678072915e-05,
+      "loss": 1.1526,
+      "step": 2983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5767529207682225e-05,
+      "loss": 0.8531,
+      "step": 2984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.575901820539275e-05,
+      "loss": 1.0773,
+      "step": 2985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.575050567241533e-05,
+      "loss": 0.7228,
+      "step": 2986
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5741991609961076e-05,
+      "loss": 1.2865,
+      "step": 2987
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5733476019241255e-05,
+      "loss": 0.8711,
+      "step": 2988
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5724958901467384e-05,
+      "loss": 0.7528,
+      "step": 2989
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.571644025785123e-05,
+      "loss": 1.0881,
+      "step": 2990
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5707920089604694e-05,
+      "loss": 0.8553,
+      "step": 2991
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5699398397939975e-05,
+      "loss": 1.26,
+      "step": 2992
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.569087518406945e-05,
+      "loss": 1.8535,
+      "step": 2993
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.568235044920571e-05,
+      "loss": 1.1105,
+      "step": 2994
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.567382419456158e-05,
+      "loss": 1.8252,
+      "step": 2995
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.566529642135008e-05,
+      "loss": 1.0342,
+      "step": 2996
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.565676713078447e-05,
+      "loss": 1.4631,
+      "step": 2997
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5648236324078215e-05,
+      "loss": 1.7129,
+      "step": 2998
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5639704002444986e-05,
+      "loss": 1.3633,
+      "step": 2999
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5631170167098685e-05,
+      "loss": 1.3077,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "eval_loss": 1.25390625,
+      "eval_runtime": 3.6148,
+      "eval_samples_per_second": 27.664,
+      "eval_steps_per_second": 4.703,
+      "step": 3000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.562263481925342e-05,
+      "loss": 1.2651,
+      "step": 3001
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5614097960123515e-05,
+      "loss": 1.1921,
+      "step": 3002
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5605559590923514e-05,
+      "loss": 0.983,
+      "step": 3003
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.559701971286818e-05,
+      "loss": 0.9602,
+      "step": 3004
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.558847832717246e-05,
+      "loss": 0.9904,
+      "step": 3005
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.557993543505156e-05,
+      "loss": 0.8749,
+      "step": 3006
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.557139103772088e-05,
+      "loss": 0.9993,
+      "step": 3007
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5562845136396016e-05,
+      "loss": 0.9342,
+      "step": 3008
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.555429773229281e-05,
+      "loss": 1.0283,
+      "step": 3009
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.55457488266273e-05,
+      "loss": 1.1064,
+      "step": 3010
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.553719842061575e-05,
+      "loss": 1.2313,
+      "step": 3011
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5528646515474595e-05,
+      "loss": 1.1102,
+      "step": 3012
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.552009311242055e-05,
+      "loss": 1.1108,
+      "step": 3013
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.55115382126705e-05,
+      "loss": 1.2074,
+      "step": 3014
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.550298181744155e-05,
+      "loss": 1.1061,
+      "step": 3015
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.549442392795103e-05,
+      "loss": 1.0972,
+      "step": 3016
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.548586454541645e-05,
+      "loss": 0.8463,
+      "step": 3017
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.547730367105557e-05,
+      "loss": 0.9817,
+      "step": 3018
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5468741306086354e-05,
+      "loss": 0.783,
+      "step": 3019
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.546017745172697e-05,
+      "loss": 0.9034,
+      "step": 3020
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5451612109195785e-05,
+      "loss": 1.3298,
+      "step": 3021
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.54430452797114e-05,
+      "loss": 0.9449,
+      "step": 3022
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.543447696449263e-05,
+      "loss": 1.0977,
+      "step": 3023
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5425907164758474e-05,
+      "loss": 1.0264,
+      "step": 3024
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5417335881728175e-05,
+      "loss": 1.0498,
+      "step": 3025
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.540876311662116e-05,
+      "loss": 0.9529,
+      "step": 3026
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.540018887065708e-05,
+      "loss": 1.0171,
+      "step": 3027
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.53916131450558e-05,
+      "loss": 0.9866,
+      "step": 3028
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.538303594103739e-05,
+      "loss": 0.8958,
+      "step": 3029
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5374457259822126e-05,
+      "loss": 1.0568,
+      "step": 3030
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.53658771026305e-05,
+      "loss": 0.8662,
+      "step": 3031
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.535729547068321e-05,
+      "loss": 0.5858,
+      "step": 3032
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.534871236520118e-05,
+      "loss": 0.8822,
+      "step": 3033
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.534012778740551e-05,
+      "loss": 0.8235,
+      "step": 3034
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5331541738517555e-05,
+      "loss": 0.7435,
+      "step": 3035
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.532295421975883e-05,
+      "loss": 0.995,
+      "step": 3036
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.53143652323511e-05,
+      "loss": 1.0499,
+      "step": 3037
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.53057747775163e-05,
+      "loss": 1.0566,
+      "step": 3038
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.529718285647663e-05,
+      "loss": 0.9664,
+      "step": 3039
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.5288589470454435e-05,
+      "loss": 1.0265,
+      "step": 3040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.527999462067231e-05,
+      "loss": 0.8812,
+      "step": 3041
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5271398308353044e-05,
+      "loss": 0.8707,
+      "step": 3042
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5262800534719645e-05,
+      "loss": 0.8804,
+      "step": 3043
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.525420130099531e-05,
+      "loss": 0.9821,
+      "step": 3044
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.524560060840345e-05,
+      "loss": 0.7888,
+      "step": 3045
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.52369984581677e-05,
+      "loss": 0.9249,
+      "step": 3046
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5228394851511874e-05,
+      "loss": 0.8761,
+      "step": 3047
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.521978978966003e-05,
+      "loss": 1.0317,
+      "step": 3048
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.521118327383639e-05,
+      "loss": 1.1056,
+      "step": 3049
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.520257530526543e-05,
+      "loss": 0.8632,
+      "step": 3050
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5193965885171785e-05,
+      "loss": 1.1349,
+      "step": 3051
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5185355014780326e-05,
+      "loss": 1.0449,
+      "step": 3052
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5176742695316126e-05,
+      "loss": 1.0115,
+      "step": 3053
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.516812892800446e-05,
+      "loss": 1.2393,
+      "step": 3054
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5159513714070814e-05,
+      "loss": 0.9308,
+      "step": 3055
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5150897054740874e-05,
+      "loss": 0.9916,
+      "step": 3056
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5142278951240535e-05,
+      "loss": 0.7913,
+      "step": 3057
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5133659404795893e-05,
+      "loss": 0.9471,
+      "step": 3058
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.512503841663325e-05,
+      "loss": 1.0369,
+      "step": 3059
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5116415987979124e-05,
+      "loss": 1.0134,
+      "step": 3060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5107792120060226e-05,
+      "loss": 1.0964,
+      "step": 3061
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.509916681410348e-05,
+      "loss": 0.837,
+      "step": 3062
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5090540071336005e-05,
+      "loss": 0.921,
+      "step": 3063
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.508191189298514e-05,
+      "loss": 0.9186,
+      "step": 3064
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.50732822802784e-05,
+      "loss": 1.0705,
+      "step": 3065
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5064651234443516e-05,
+      "loss": 0.9279,
+      "step": 3066
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5056018756708464e-05,
+      "loss": 1.0398,
+      "step": 3067
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.504738484830137e-05,
+      "loss": 0.9363,
+      "step": 3068
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.503874951045057e-05,
+      "loss": 1.006,
+      "step": 3069
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5030112744384624e-05,
+      "loss": 0.8528,
+      "step": 3070
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.50214745513323e-05,
+      "loss": 0.9296,
+      "step": 3071
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.5012834932522544e-05,
+      "loss": 1.2097,
+      "step": 3072
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.50041938891845e-05,
+      "loss": 1.1501,
+      "step": 3073
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.499555142254757e-05,
+      "loss": 0.806,
+      "step": 3074
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4986907533841276e-05,
+      "loss": 0.9613,
+      "step": 3075
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.497826222429542e-05,
+      "loss": 0.8134,
+      "step": 3076
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.496961549513996e-05,
+      "loss": 1.2629,
+      "step": 3077
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4960967347605065e-05,
+      "loss": 0.9215,
+      "step": 3078
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.495231778292112e-05,
+      "loss": 0.8282,
+      "step": 3079
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.494366680231868e-05,
+      "loss": 0.8331,
+      "step": 3080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.493501440702854e-05,
+      "loss": 0.7889,
+      "step": 3081
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4926360598281673e-05,
+      "loss": 0.7826,
+      "step": 3082
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4917705377309254e-05,
+      "loss": 0.7341,
+      "step": 3083
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.490904874534266e-05,
+      "loss": 0.856,
+      "step": 3084
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.490039070361349e-05,
+      "loss": 1.0364,
+      "step": 3085
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.489173125335351e-05,
+      "loss": 0.8938,
+      "step": 3086
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.488307039579471e-05,
+      "loss": 0.8805,
+      "step": 3087
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.487440813216927e-05,
+      "loss": 0.865,
+      "step": 3088
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4865744463709564e-05,
+      "loss": 1.1622,
+      "step": 3089
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4857079391648185e-05,
+      "loss": 0.973,
+      "step": 3090
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.484841291721791e-05,
+      "loss": 0.7705,
+      "step": 3091
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.483974504165172e-05,
+      "loss": 0.8385,
+      "step": 3092
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4831075766182796e-05,
+      "loss": 1.0634,
+      "step": 3093
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4822405092044517e-05,
+      "loss": 1.066,
+      "step": 3094
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.481373302047047e-05,
+      "loss": 0.8917,
+      "step": 3095
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.480505955269442e-05,
+      "loss": 1.1034,
+      "step": 3096
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.479638468995034e-05,
+      "loss": 1.0283,
+      "step": 3097
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4787708433472425e-05,
+      "loss": 1.2119,
+      "step": 3098
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.477903078449504e-05,
+      "loss": 1.1414,
+      "step": 3099
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.477035174425274e-05,
+      "loss": 0.959,
+      "step": 3100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.476167131398031e-05,
+      "loss": 1.1335,
+      "step": 3101
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.475298949491272e-05,
+      "loss": 1.0171,
+      "step": 3102
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.474430628828512e-05,
+      "loss": 1.1672,
+      "step": 3103
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4735621695332874e-05,
+      "loss": 0.8104,
+      "step": 3104
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.472693571729156e-05,
+      "loss": 0.9619,
+      "step": 3105
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.471824835539691e-05,
+      "loss": 1.2253,
+      "step": 3106
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.470955961088489e-05,
+      "loss": 0.881,
+      "step": 3107
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.470086948499165e-05,
+      "loss": 0.7446,
+      "step": 3108
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4692177978953535e-05,
+      "loss": 1.0498,
+      "step": 3109
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.468348509400707e-05,
+      "loss": 0.9143,
+      "step": 3110
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.467479083138902e-05,
+      "loss": 0.8232,
+      "step": 3111
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.466609519233631e-05,
+      "loss": 1.3232,
+      "step": 3112
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4657398178086074e-05,
+      "loss": 0.7429,
+      "step": 3113
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.464869978987562e-05,
+      "loss": 0.8853,
+      "step": 3114
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.464000002894249e-05,
+      "loss": 0.772,
+      "step": 3115
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.46312988965244e-05,
+      "loss": 1.105,
+      "step": 3116
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.462259639385925e-05,
+      "loss": 0.6878,
+      "step": 3117
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4613892522185155e-05,
+      "loss": 1.0181,
+      "step": 3118
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.4605187282740414e-05,
+      "loss": 0.7074,
+      "step": 3119
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.459648067676352e-05,
+      "loss": 1.1953,
+      "step": 3120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.458777270549317e-05,
+      "loss": 1.1139,
+      "step": 3121
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.457906337016825e-05,
+      "loss": 0.9805,
+      "step": 3122
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.457035267202784e-05,
+      "loss": 1.1305,
+      "step": 3123
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.456164061231121e-05,
+      "loss": 1.0354,
+      "step": 3124
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.455292719225782e-05,
+      "loss": 0.8849,
+      "step": 3125
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.454421241310733e-05,
+      "loss": 0.8975,
+      "step": 3126
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.453549627609961e-05,
+      "loss": 1.0474,
+      "step": 3127
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.452677878247469e-05,
+      "loss": 0.9652,
+      "step": 3128
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.451805993347282e-05,
+      "loss": 0.9657,
+      "step": 3129
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4509339730334425e-05,
+      "loss": 1.1661,
+      "step": 3130
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4500618174300145e-05,
+      "loss": 1.1003,
+      "step": 3131
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4491895266610776e-05,
+      "loss": 1.0981,
+      "step": 3132
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.448317100850734e-05,
+      "loss": 1.0635,
+      "step": 3133
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4474445401231036e-05,
+      "loss": 0.9908,
+      "step": 3134
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4465718446023264e-05,
+      "loss": 0.8503,
+      "step": 3135
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.445699014412561e-05,
+      "loss": 0.9492,
+      "step": 3136
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.444826049677984e-05,
+      "loss": 0.8606,
+      "step": 3137
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.443952950522793e-05,
+      "loss": 0.8278,
+      "step": 3138
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.443079717071205e-05,
+      "loss": 0.9534,
+      "step": 3139
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4422063494474534e-05,
+      "loss": 0.9027,
+      "step": 3140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.441332847775793e-05,
+      "loss": 0.8552,
+      "step": 3141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4404592121804975e-05,
+      "loss": 0.8843,
+      "step": 3142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.439585442785859e-05,
+      "loss": 0.9961,
+      "step": 3143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4387115397161886e-05,
+      "loss": 1.0366,
+      "step": 3144
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4378375030958174e-05,
+      "loss": 0.9065,
+      "step": 3145
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.436963333049094e-05,
+      "loss": 0.9806,
+      "step": 3146
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4360890297003874e-05,
+      "loss": 0.858,
+      "step": 3147
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4352145931740846e-05,
+      "loss": 0.8585,
+      "step": 3148
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4343400235945914e-05,
+      "loss": 1.0148,
+      "step": 3149
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4334653210863336e-05,
+      "loss": 1.3516,
+      "step": 3150
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4325904857737554e-05,
+      "loss": 1.2405,
+      "step": 3151
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.43171551778132e-05,
+      "loss": 0.9816,
+      "step": 3152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.430840417233509e-05,
+      "loss": 1.0313,
+      "step": 3153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.429965184254822e-05,
+      "loss": 1.1155,
+      "step": 3154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4290898189697814e-05,
+      "loss": 1.0145,
+      "step": 3155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4282143215029227e-05,
+      "loss": 1.1783,
+      "step": 3156
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.427338691978804e-05,
+      "loss": 1.2393,
+      "step": 3157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4264629305220025e-05,
+      "loss": 1.2056,
+      "step": 3158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4255870372571124e-05,
+      "loss": 1.103,
+      "step": 3159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.424711012308746e-05,
+      "loss": 1.1205,
+      "step": 3160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.423834855801536e-05,
+      "loss": 0.8103,
+      "step": 3161
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4229585678601354e-05,
+      "loss": 0.9752,
+      "step": 3162
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.422082148609212e-05,
+      "loss": 0.7925,
+      "step": 3163
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4212055981734545e-05,
+      "loss": 0.9645,
+      "step": 3164
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.420328916677569e-05,
+      "loss": 1.1152,
+      "step": 3165
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4194521042462836e-05,
+      "loss": 0.9937,
+      "step": 3166
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.41857516100434e-05,
+      "loss": 1.1731,
+      "step": 3167
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.417698087076502e-05,
+      "loss": 1.2592,
+      "step": 3168
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.416820882587553e-05,
+      "loss": 0.7372,
+      "step": 3169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4159435476622904e-05,
+      "loss": 1.0404,
+      "step": 3170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.415066082425534e-05,
+      "loss": 1.2384,
+      "step": 3171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.414188487002121e-05,
+      "loss": 0.8865,
+      "step": 3172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.413310761516907e-05,
+      "loss": 1.1256,
+      "step": 3173
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.412432906094766e-05,
+      "loss": 0.9379,
+      "step": 3174
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4115549208605895e-05,
+      "loss": 0.8979,
+      "step": 3175
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4106768059392916e-05,
+      "loss": 1.0161,
+      "step": 3176
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4097985614558e-05,
+      "loss": 0.8153,
+      "step": 3177
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.408920187535063e-05,
+      "loss": 1.0732,
+      "step": 3178
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.408041684302047e-05,
+      "loss": 1.0896,
+      "step": 3179
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.407163051881737e-05,
+      "loss": 1.0206,
+      "step": 3180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.406284290399136e-05,
+      "loss": 0.7603,
+      "step": 3181
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.405405399979266e-05,
+      "loss": 1.0811,
+      "step": 3182
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4045263807471667e-05,
+      "loss": 0.9117,
+      "step": 3183
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4036472328278964e-05,
+      "loss": 0.7555,
+      "step": 3184
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.4027679563465306e-05,
+      "loss": 1.1436,
+      "step": 3185
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.401888551428167e-05,
+      "loss": 1.0198,
+      "step": 3186
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.401009018197916e-05,
+      "loss": 0.8312,
+      "step": 3187
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.40012935678091e-05,
+      "loss": 1.2344,
+      "step": 3188
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.399249567302298e-05,
+      "loss": 0.9269,
+      "step": 3189
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.398369649887249e-05,
+      "loss": 1.1454,
+      "step": 3190
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.397489604660949e-05,
+      "loss": 0.8583,
+      "step": 3191
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.396609431748601e-05,
+      "loss": 0.9031,
+      "step": 3192
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3957291312754274e-05,
+      "loss": 0.9972,
+      "step": 3193
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3948487033666707e-05,
+      "loss": 1.0349,
+      "step": 3194
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.393968148147587e-05,
+      "loss": 0.8635,
+      "step": 3195
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3930874657434556e-05,
+      "loss": 0.8875,
+      "step": 3196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.39220665627957e-05,
+      "loss": 0.7014,
+      "step": 3197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3913257198812424e-05,
+      "loss": 1.059,
+      "step": 3198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3904446566738055e-05,
+      "loss": 0.9573,
+      "step": 3199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.389563466782608e-05,
+      "loss": 0.9132,
+      "step": 3200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3886821503330165e-05,
+      "loss": 1.0073,
+      "step": 3201
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3878007074504146e-05,
+      "loss": 0.9056,
+      "step": 3202
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3869191382602085e-05,
+      "loss": 1.031,
+      "step": 3203
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.386037442887817e-05,
+      "loss": 0.9535,
+      "step": 3204
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.3851556214586804e-05,
+      "loss": 0.9719,
+      "step": 3205
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.384273674098255e-05,
+      "loss": 0.9846,
+      "step": 3206
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.383391600932015e-05,
+      "loss": 0.7863,
+      "step": 3207
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3825094020854554e-05,
+      "loss": 1.037,
+      "step": 3208
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.381627077684083e-05,
+      "loss": 0.6566,
+      "step": 3209
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.38074462785343e-05,
+      "loss": 0.9353,
+      "step": 3210
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.379862052719041e-05,
+      "loss": 1.3372,
+      "step": 3211
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.378979352406481e-05,
+      "loss": 0.984,
+      "step": 3212
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.37809652704133e-05,
+      "loss": 0.9279,
+      "step": 3213
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.37721357674919e-05,
+      "loss": 1.0721,
+      "step": 3214
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3763305016556786e-05,
+      "loss": 1.0864,
+      "step": 3215
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.375447301886429e-05,
+      "loss": 1.2081,
+      "step": 3216
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3745639775670954e-05,
+      "loss": 1.1687,
+      "step": 3217
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.373680528823349e-05,
+      "loss": 1.058,
+      "step": 3218
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.372796955780878e-05,
+      "loss": 0.8815,
+      "step": 3219
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.371913258565388e-05,
+      "loss": 0.9774,
+      "step": 3220
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.371029437302604e-05,
+      "loss": 0.999,
+      "step": 3221
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.370145492118266e-05,
+      "loss": 0.9473,
+      "step": 3222
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.369261423138133e-05,
+      "loss": 1.0896,
+      "step": 3223
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.368377230487983e-05,
+      "loss": 0.9039,
+      "step": 3224
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.36749291429361e-05,
+      "loss": 0.7506,
+      "step": 3225
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.366608474680826e-05,
+      "loss": 1.207,
+      "step": 3226
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.365723911775459e-05,
+      "loss": 1.071,
+      "step": 3227
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.364839225703357e-05,
+      "loss": 0.9178,
+      "step": 3228
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.363954416590384e-05,
+      "loss": 1.0225,
+      "step": 3229
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3630694845624237e-05,
+      "loss": 0.9524,
+      "step": 3230
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.362184429745373e-05,
+      "loss": 1.0239,
+      "step": 3231
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.361299252265151e-05,
+      "loss": 0.8834,
+      "step": 3232
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.360413952247691e-05,
+      "loss": 1.1503,
+      "step": 3233
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.359528529818945e-05,
+      "loss": 0.9722,
+      "step": 3234
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3586429851048825e-05,
+      "loss": 0.962,
+      "step": 3235
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.35775731823149e-05,
+      "loss": 0.9685,
+      "step": 3236
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.356871529324771e-05,
+      "loss": 1.2308,
+      "step": 3237
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.355985618510747e-05,
+      "loss": 0.8407,
+      "step": 3238
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3550995859154576e-05,
+      "loss": 1.1459,
+      "step": 3239
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.354213431664959e-05,
+      "loss": 0.9587,
+      "step": 3240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3533271558853237e-05,
+      "loss": 0.9712,
+      "step": 3241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3524407587026426e-05,
+      "loss": 1.002,
+      "step": 3242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.351554240243023e-05,
+      "loss": 1.0071,
+      "step": 3243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.350667600632591e-05,
+      "loss": 1.1108,
+      "step": 3244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3497808399974885e-05,
+      "loss": 1.2904,
+      "step": 3245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.348893958463875e-05,
+      "loss": 0.9637,
+      "step": 3246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3480069561579287e-05,
+      "loss": 0.9352,
+      "step": 3247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3471198332058417e-05,
+      "loss": 1.1155,
+      "step": 3248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.346232589733826e-05,
+      "loss": 0.9259,
+      "step": 3249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.345345225868111e-05,
+      "loss": 1.1042,
+      "step": 3250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.34445774173494e-05,
+      "loss": 1.0493,
+      "step": 3251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3435701374605774e-05,
+      "loss": 0.8984,
+      "step": 3252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.342682413171302e-05,
+      "loss": 1.0975,
+      "step": 3253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.341794568993411e-05,
+      "loss": 0.9667,
+      "step": 3254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.340906605053218e-05,
+      "loss": 0.9116,
+      "step": 3255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.340018521477053e-05,
+      "loss": 1.0198,
+      "step": 3256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.339130318391265e-05,
+      "loss": 0.9852,
+      "step": 3257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3382419959222176e-05,
+      "loss": 1.0349,
+      "step": 3258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.337353554196294e-05,
+      "loss": 0.8854,
+      "step": 3259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.336464993339893e-05,
+      "loss": 1.1971,
+      "step": 3260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.33557631347943e-05,
+      "loss": 0.7941,
+      "step": 3261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.334687514741336e-05,
+      "loss": 1.0969,
+      "step": 3262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.333798597252063e-05,
+      "loss": 0.9366,
+      "step": 3263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.332909561138077e-05,
+      "loss": 1.0104,
+      "step": 3264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.332020406525861e-05,
+      "loss": 0.8654,
+      "step": 3265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.331131133541914e-05,
+      "loss": 0.7203,
+      "step": 3266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.330241742312755e-05,
+      "loss": 1.0981,
+      "step": 3267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.329352232964918e-05,
+      "loss": 0.9152,
+      "step": 3268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.328462605624952e-05,
+      "loss": 0.9274,
+      "step": 3269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3275728604194246e-05,
+      "loss": 1.0886,
+      "step": 3270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3266829974749216e-05,
+      "loss": 1.1125,
+      "step": 3271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3257930169180426e-05,
+      "loss": 1.1144,
+      "step": 3272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.324902918875407e-05,
+      "loss": 0.9872,
+      "step": 3273
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.324012703473647e-05,
+      "loss": 0.8939,
+      "step": 3274
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3231223708394156e-05,
+      "loss": 1.3325,
+      "step": 3275
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3222319210993794e-05,
+      "loss": 1.0826,
+      "step": 3276
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.321341354380223e-05,
+      "loss": 0.9249,
+      "step": 3277
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.320450670808649e-05,
+      "loss": 1.1104,
+      "step": 3278
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.319559870511374e-05,
+      "loss": 0.8529,
+      "step": 3279
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.318668953615132e-05,
+      "loss": 1.1571,
+      "step": 3280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.317777920246674e-05,
+      "loss": 0.9078,
+      "step": 3281
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3168867705327686e-05,
+      "loss": 1.0066,
+      "step": 3282
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.315995504600199e-05,
+      "loss": 1.0062,
+      "step": 3283
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.315104122575766e-05,
+      "loss": 1.038,
+      "step": 3284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.314212624586287e-05,
+      "loss": 1.123,
+      "step": 3285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.313321010758594e-05,
+      "loss": 0.9158,
+      "step": 3286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.312429281219539e-05,
+      "loss": 0.7271,
+      "step": 3287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3115374360959886e-05,
+      "loss": 1.2102,
+      "step": 3288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.3106454755148256e-05,
+      "loss": 0.9275,
+      "step": 3289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.309753399602948e-05,
+      "loss": 1.1267,
+      "step": 3290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.308861208487272e-05,
+      "loss": 1.1643,
+      "step": 3291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3079689022947326e-05,
+      "loss": 1.0905,
+      "step": 3292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.307076481152275e-05,
+      "loss": 0.8503,
+      "step": 3293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.306183945186865e-05,
+      "loss": 0.9559,
+      "step": 3294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.305291294525484e-05,
+      "loss": 1.4348,
+      "step": 3295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.3043985292951315e-05,
+      "loss": 1.0853,
+      "step": 3296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.303505649622819e-05,
+      "loss": 1.1268,
+      "step": 3297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.302612655635577e-05,
+      "loss": 1.0378,
+      "step": 3298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.301719547460454e-05,
+      "loss": 0.8353,
+      "step": 3299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.30082632522451e-05,
+      "loss": 1.2102,
+      "step": 3300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2999329890548256e-05,
+      "loss": 0.9846,
+      "step": 3301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2990395390784954e-05,
+      "loss": 0.8801,
+      "step": 3302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.298145975422631e-05,
+      "loss": 0.7673,
+      "step": 3303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.297252298214359e-05,
+      "loss": 0.8286,
+      "step": 3304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.296358507580824e-05,
+      "loss": 1.2351,
+      "step": 3305
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.295464603649186e-05,
+      "loss": 1.2449,
+      "step": 3306
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.29457058654662e-05,
+      "loss": 0.9058,
+      "step": 3307
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2936764564003184e-05,
+      "loss": 0.8163,
+      "step": 3308
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.29278221333749e-05,
+      "loss": 1.2991,
+      "step": 3309
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.291887857485358e-05,
+      "loss": 1.0087,
+      "step": 3310
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.290993388971162e-05,
+      "loss": 1.1509,
+      "step": 3311
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.29009880792216e-05,
+      "loss": 1.162,
+      "step": 3312
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.289204114465624e-05,
+      "loss": 1.2163,
+      "step": 3313
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.288309308728841e-05,
+      "loss": 1.0084,
+      "step": 3314
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2874143908391155e-05,
+      "loss": 1.3577,
+      "step": 3315
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.286519360923769e-05,
+      "loss": 0.9912,
+      "step": 3316
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.285624219110136e-05,
+      "loss": 1.4199,
+      "step": 3317
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2847289655255685e-05,
+      "loss": 0.8345,
+      "step": 3318
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.283833600297436e-05,
+      "loss": 1.1074,
+      "step": 3319
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2829381235531215e-05,
+      "loss": 1.0647,
+      "step": 3320
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.282042535420025e-05,
+      "loss": 1.1731,
+      "step": 3321
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.28114683602556e-05,
+      "loss": 0.9763,
+      "step": 3322
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2802510254971605e-05,
+      "loss": 0.8406,
+      "step": 3323
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2793551039622734e-05,
+      "loss": 0.9836,
+      "step": 3324
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.278459071548359e-05,
+      "loss": 1.121,
+      "step": 3325
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.277562928382899e-05,
+      "loss": 1.1801,
+      "step": 3326
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.276666674593387e-05,
+      "loss": 1.045,
+      "step": 3327
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2757703103073327e-05,
+      "loss": 1.3197,
+      "step": 3328
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.274873835652262e-05,
+      "loss": 1.1078,
+      "step": 3329
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2739772507557176e-05,
+      "loss": 0.9451,
+      "step": 3330
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.273080555745256e-05,
+      "loss": 1.2249,
+      "step": 3331
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.272183750748451e-05,
+      "loss": 1.0405,
+      "step": 3332
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.271286835892889e-05,
+      "loss": 1.0374,
+      "step": 3333
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.270389811306177e-05,
+      "loss": 1.1362,
+      "step": 3334
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.269492677115934e-05,
+      "loss": 1.3838,
+      "step": 3335
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2685954334497945e-05,
+      "loss": 0.8868,
+      "step": 3336
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.267698080435411e-05,
+      "loss": 0.8271,
+      "step": 3337
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.266800618200449e-05,
+      "loss": 0.9839,
+      "step": 3338
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.265903046872592e-05,
+      "loss": 1.1505,
+      "step": 3339
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.265005366579535e-05,
+      "loss": 0.9535,
+      "step": 3340
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.264107577448994e-05,
+      "loss": 1.0266,
+      "step": 3341
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2632096796086976e-05,
+      "loss": 1.457,
+      "step": 3342
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2623116731863876e-05,
+      "loss": 1.0524,
+      "step": 3343
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2614135583098246e-05,
+      "loss": 0.9058,
+      "step": 3344
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2605153351067844e-05,
+      "loss": 0.7711,
+      "step": 3345
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.259617003705057e-05,
+      "loss": 0.9948,
+      "step": 3346
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.258718564232448e-05,
+      "loss": 1.0145,
+      "step": 3347
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.257820016816779e-05,
+      "loss": 1.1609,
+      "step": 3348
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.256921361585887e-05,
+      "loss": 1.0922,
+      "step": 3349
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.256022598667622e-05,
+      "loss": 0.8402,
+      "step": 3350
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.255123728189853e-05,
+      "loss": 0.9937,
+      "step": 3351
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2542247502804616e-05,
+      "loss": 0.9498,
+      "step": 3352
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2533256650673446e-05,
+      "loss": 1.14,
+      "step": 3353
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.252426472678418e-05,
+      "loss": 1.2832,
+      "step": 3354
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.251527173241607e-05,
+      "loss": 1.0946,
+      "step": 3355
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2506277668848584e-05,
+      "loss": 0.9378,
+      "step": 3356
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.249728253736128e-05,
+      "loss": 1.1044,
+      "step": 3357
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2488286339233904e-05,
+      "loss": 1.2308,
+      "step": 3358
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.247928907574636e-05,
+      "loss": 1.379,
+      "step": 3359
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.247029074817867e-05,
+      "loss": 1.2679,
+      "step": 3360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.246129135781105e-05,
+      "loss": 0.8358,
+      "step": 3361
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2452290905923835e-05,
+      "loss": 1.0994,
+      "step": 3362
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.244328939379752e-05,
+      "loss": 1.4196,
+      "step": 3363
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.243428682271276e-05,
+      "loss": 0.8246,
+      "step": 3364
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.242528319395033e-05,
+      "loss": 0.8807,
+      "step": 3365
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.24162785087912e-05,
+      "loss": 0.9889,
+      "step": 3366
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.240727276851647e-05,
+      "loss": 1.1071,
+      "step": 3367
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2398265974407385e-05,
+      "loss": 1.0052,
+      "step": 3368
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.238925812774533e-05,
+      "loss": 1.0325,
+      "step": 3369
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2380249229811876e-05,
+      "loss": 1.4033,
+      "step": 3370
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.23712392818887e-05,
+      "loss": 1.0861,
+      "step": 3371
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2362228285257654e-05,
+      "loss": 1.2378,
+      "step": 3372
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.2353216241200754e-05,
+      "loss": 1.0449,
+      "step": 3373
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.234420315100012e-05,
+      "loss": 1.066,
+      "step": 3374
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.233518901593806e-05,
+      "loss": 1.1874,
+      "step": 3375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.232617383729702e-05,
+      "loss": 1.2169,
+      "step": 3376
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2317157616359576e-05,
+      "loss": 0.984,
+      "step": 3377
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.230814035440848e-05,
+      "loss": 1.2756,
+      "step": 3378
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2299122052726606e-05,
+      "loss": 1.0181,
+      "step": 3379
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.229010271259702e-05,
+      "loss": 0.9058,
+      "step": 3380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2281082335302874e-05,
+      "loss": 1.1565,
+      "step": 3381
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.227206092212751e-05,
+      "loss": 0.8693,
+      "step": 3382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.226303847435441e-05,
+      "loss": 0.9222,
+      "step": 3383
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2254014993267214e-05,
+      "loss": 0.8853,
+      "step": 3384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.224499048014966e-05,
+      "loss": 0.6307,
+      "step": 3385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.223596493628568e-05,
+      "loss": 0.8936,
+      "step": 3386
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.222693836295936e-05,
+      "loss": 1.072,
+      "step": 3387
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2217910761454894e-05,
+      "loss": 1.3269,
+      "step": 3388
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2208882133056645e-05,
+      "loss": 0.8141,
+      "step": 3389
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.219985247904912e-05,
+      "loss": 1.0209,
+      "step": 3390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.219082180071697e-05,
+      "loss": 0.8635,
+      "step": 3391
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.218179009934498e-05,
+      "loss": 1.0581,
+      "step": 3392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.217275737621811e-05,
+      "loss": 1.0039,
+      "step": 3393
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.216372363262144e-05,
+      "loss": 1.1274,
+      "step": 3394
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2154688869840197e-05,
+      "loss": 1.2101,
+      "step": 3395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.214565308915976e-05,
+      "loss": 0.8309,
+      "step": 3396
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.213661629186567e-05,
+      "loss": 1.2839,
+      "step": 3397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.212757847924358e-05,
+      "loss": 0.7502,
+      "step": 3398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.211853965257929e-05,
+      "loss": 1.3175,
+      "step": 3399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2109499813158763e-05,
+      "loss": 0.9464,
+      "step": 3400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.210045896226812e-05,
+      "loss": 0.8407,
+      "step": 3401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.209141710119358e-05,
+      "loss": 1.0735,
+      "step": 3402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2082374231221535e-05,
+      "loss": 1.1814,
+      "step": 3403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2073330353638524e-05,
+      "loss": 1.2256,
+      "step": 3404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.206428546973122e-05,
+      "loss": 1.3743,
+      "step": 3405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2055239580786434e-05,
+      "loss": 0.7726,
+      "step": 3406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.204619268809113e-05,
+      "loss": 1.4116,
+      "step": 3407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2037144792932417e-05,
+      "loss": 1.0824,
+      "step": 3408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.202809589659753e-05,
+      "loss": 1.1005,
+      "step": 3409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.2019046000373865e-05,
+      "loss": 1.0068,
+      "step": 3410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.200999510554895e-05,
+      "loss": 1.0695,
+      "step": 3411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.200094321341047e-05,
+      "loss": 0.8446,
+      "step": 3412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.199189032524622e-05,
+      "loss": 0.9729,
+      "step": 3413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.198283644234416e-05,
+      "loss": 1.1608,
+      "step": 3414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1973781565992406e-05,
+      "loss": 1.2532,
+      "step": 3415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.196472569747918e-05,
+      "loss": 1.0773,
+      "step": 3416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.195566883809286e-05,
+      "loss": 0.9718,
+      "step": 3417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.194661098912198e-05,
+      "loss": 1.1251,
+      "step": 3418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1937552151855206e-05,
+      "loss": 1.0928,
+      "step": 3419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1928492327581314e-05,
+      "loss": 1.0858,
+      "step": 3420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.191943151758927e-05,
+      "loss": 1.1556,
+      "step": 3421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.191036972316816e-05,
+      "loss": 0.8477,
+      "step": 3422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.19013069456072e-05,
+      "loss": 0.8439,
+      "step": 3423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.189224318619575e-05,
+      "loss": 1.1383,
+      "step": 3424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.188317844622331e-05,
+      "loss": 1.0139,
+      "step": 3425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1874112726979545e-05,
+      "loss": 1.0249,
+      "step": 3426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.18650460297542e-05,
+      "loss": 0.9459,
+      "step": 3427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.185597835583722e-05,
+      "loss": 1.2422,
+      "step": 3428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.184690970651867e-05,
+      "loss": 1.0466,
+      "step": 3429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1837840083088746e-05,
+      "loss": 0.9593,
+      "step": 3430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1828769486837775e-05,
+      "loss": 0.9768,
+      "step": 3431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1819697919056235e-05,
+      "loss": 1.4492,
+      "step": 3432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.181062538103474e-05,
+      "loss": 1.1964,
+      "step": 3433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.180155187406405e-05,
+      "loss": 1.1521,
+      "step": 3434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.179247739943505e-05,
+      "loss": 1.0122,
+      "step": 3435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.178340195843876e-05,
+      "loss": 0.926,
+      "step": 3436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.177432555236636e-05,
+      "loss": 0.9553,
+      "step": 3437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.176524818250914e-05,
+      "loss": 1.2129,
+      "step": 3438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1756169850158554e-05,
+      "loss": 0.749,
+      "step": 3439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.174709055660615e-05,
+      "loss": 1.2058,
+      "step": 3440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1738010303143674e-05,
+      "loss": 1.0497,
+      "step": 3441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.172892909106295e-05,
+      "loss": 1.2673,
+      "step": 3442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.171984692165599e-05,
+      "loss": 1.1548,
+      "step": 3443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1710763796214894e-05,
+      "loss": 1.0043,
+      "step": 3444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.170167971603193e-05,
+      "loss": 0.9973,
+      "step": 3445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.169259468239948e-05,
+      "loss": 1.1936,
+      "step": 3446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.168350869661009e-05,
+      "loss": 1.0178,
+      "step": 3447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.167442175995642e-05,
+      "loss": 1.0394,
+      "step": 3448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.166533387373127e-05,
+      "loss": 1.1239,
+      "step": 3449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.165624503922757e-05,
+      "loss": 1.252,
+      "step": 3450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.164715525773841e-05,
+      "loss": 1.1306,
+      "step": 3451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.163806453055697e-05,
+      "loss": 1.0282,
+      "step": 3452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.16289728589766e-05,
+      "loss": 0.9833,
+      "step": 3453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.161988024429077e-05,
+      "loss": 1.0684,
+      "step": 3454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1610786687793105e-05,
+      "loss": 1.0845,
+      "step": 3455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.160169219077733e-05,
+      "loss": 0.9238,
+      "step": 3456
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.159259675453732e-05,
+      "loss": 1.0665,
+      "step": 3457
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1583500380367104e-05,
+      "loss": 0.9608,
+      "step": 3458
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.15744030695608e-05,
+      "loss": 1.3804,
+      "step": 3459
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.156530482341271e-05,
+      "loss": 0.8229,
+      "step": 3460
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.155620564321722e-05,
+      "loss": 1.2024,
+      "step": 3461
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1547105530268885e-05,
+      "loss": 0.9961,
+      "step": 3462
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1538004485862374e-05,
+      "loss": 0.8966,
+      "step": 3463
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.152890251129251e-05,
+      "loss": 0.8441,
+      "step": 3464
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1519799607854214e-05,
+      "loss": 0.9312,
+      "step": 3465
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.151069577684257e-05,
+      "loss": 1.0897,
+      "step": 3466
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.150159101955277e-05,
+      "loss": 0.8441,
+      "step": 3467
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.149248533728017e-05,
+      "loss": 0.924,
+      "step": 3468
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1483378731320214e-05,
+      "loss": 1.1843,
+      "step": 3469
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1474271202968513e-05,
+      "loss": 1.0972,
+      "step": 3470
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.14651627535208e-05,
+      "loss": 1.0247,
+      "step": 3471
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.145605338427293e-05,
+      "loss": 1.061,
+      "step": 3472
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1446943096520905e-05,
+      "loss": 0.9836,
+      "step": 3473
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.143783189156082e-05,
+      "loss": 1.0664,
+      "step": 3474
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.142871977068897e-05,
+      "loss": 0.828,
+      "step": 3475
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1419606735201704e-05,
+      "loss": 0.8949,
+      "step": 3476
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.141049278639555e-05,
+      "loss": 0.9657,
+      "step": 3477
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1401377925567156e-05,
+      "loss": 0.9769,
+      "step": 3478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.139226215401329e-05,
+      "loss": 0.9994,
+      "step": 3479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1383145473030846e-05,
+      "loss": 0.8364,
+      "step": 3480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.137402788391686e-05,
+      "loss": 0.9369,
+      "step": 3481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.136490938796851e-05,
+      "loss": 0.8234,
+      "step": 3482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.135578998648306e-05,
+      "loss": 1.2921,
+      "step": 3483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.134666968075796e-05,
+      "loss": 0.866,
+      "step": 3484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.133754847209073e-05,
+      "loss": 0.983,
+      "step": 3485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.132842636177906e-05,
+      "loss": 0.8597,
+      "step": 3486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1319303351120756e-05,
+      "loss": 1.137,
+      "step": 3487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.131017944141373e-05,
+      "loss": 1.0034,
+      "step": 3488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.130105463395608e-05,
+      "loss": 1.1633,
+      "step": 3489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.129192893004597e-05,
+      "loss": 0.8856,
+      "step": 3490
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.128280233098172e-05,
+      "loss": 0.9786,
+      "step": 3491
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1273674838061764e-05,
+      "loss": 1.1123,
+      "step": 3492
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.12645464525847e-05,
+      "loss": 0.8788,
+      "step": 3493
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1255417175849195e-05,
+      "loss": 1.1243,
+      "step": 3494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.124628700915408e-05,
+      "loss": 1.1161,
+      "step": 3495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.123715595379832e-05,
+      "loss": 0.6515,
+      "step": 3496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1228024011080985e-05,
+      "loss": 0.7899,
+      "step": 3497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.121889118230127e-05,
+      "loss": 1.063,
+      "step": 3498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.120975746875852e-05,
+      "loss": 0.9474,
+      "step": 3499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.120062287175218e-05,
+      "loss": 1.0502,
+      "step": 3500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.119148739258183e-05,
+      "loss": 0.8682,
+      "step": 3501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.118235103254718e-05,
+      "loss": 1.17,
+      "step": 3502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.117321379294807e-05,
+      "loss": 1.2444,
+      "step": 3503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.116407567508444e-05,
+      "loss": 0.7946,
+      "step": 3504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.115493668025639e-05,
+      "loss": 1.1445,
+      "step": 3505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.114579680976411e-05,
+      "loss": 0.8745,
+      "step": 3506
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.113665606490795e-05,
+      "loss": 0.9041,
+      "step": 3507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.112751444698834e-05,
+      "loss": 0.8575,
+      "step": 3508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.111837195730589e-05,
+      "loss": 0.9216,
+      "step": 3509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1109228597161284e-05,
+      "loss": 1.2305,
+      "step": 3510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.110008436785535e-05,
+      "loss": 0.9807,
+      "step": 3511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1090939270689054e-05,
+      "loss": 0.9303,
+      "step": 3512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.108179330696346e-05,
+      "loss": 1.2156,
+      "step": 3513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1072646477979774e-05,
+      "loss": 0.8423,
+      "step": 3514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.106349878503929e-05,
+      "loss": 1.0278,
+      "step": 3515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1054350229443494e-05,
+      "loss": 1.3256,
+      "step": 3516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.104520081249393e-05,
+      "loss": 0.9514,
+      "step": 3517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.103605053549229e-05,
+      "loss": 1.0315,
+      "step": 3518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1026899399740384e-05,
+      "loss": 0.9001,
+      "step": 3519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1017747406540146e-05,
+      "loss": 1.16,
+      "step": 3520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1008594557193645e-05,
+      "loss": 0.9526,
+      "step": 3521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0999440853003035e-05,
+      "loss": 0.9082,
+      "step": 3522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.099028629527064e-05,
+      "loss": 0.955,
+      "step": 3523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.098113088529887e-05,
+      "loss": 0.7886,
+      "step": 3524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0971974624390266e-05,
+      "loss": 0.8359,
+      "step": 3525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.096281751384749e-05,
+      "loss": 1.1282,
+      "step": 3526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.095365955497333e-05,
+      "loss": 0.9662,
+      "step": 3527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0944500749070695e-05,
+      "loss": 1.1666,
+      "step": 3528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0935341097442596e-05,
+      "loss": 1.0697,
+      "step": 3529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0926180601392194e-05,
+      "loss": 1.1144,
+      "step": 3530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0917019262222746e-05,
+      "loss": 0.9288,
+      "step": 3531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0907857081237654e-05,
+      "loss": 1.2463,
+      "step": 3532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.089869405974039e-05,
+      "loss": 1.3595,
+      "step": 3533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.088953019903461e-05,
+      "loss": 0.9243,
+      "step": 3534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0880365500424044e-05,
+      "loss": 1.2301,
+      "step": 3535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0871199965212564e-05,
+      "loss": 0.9189,
+      "step": 3536
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.086203359470415e-05,
+      "loss": 1.1399,
+      "step": 3537
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0852866390202896e-05,
+      "loss": 0.9854,
+      "step": 3538
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.0843698353013034e-05,
+      "loss": 0.9554,
+      "step": 3539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0834529484438895e-05,
+      "loss": 1.1909,
+      "step": 3540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0825359785784936e-05,
+      "loss": 1.1669,
+      "step": 3541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.081618925835573e-05,
+      "loss": 1.4126,
+      "step": 3542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.080701790345598e-05,
+      "loss": 1.0565,
+      "step": 3543
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.079784572239049e-05,
+      "loss": 1.2504,
+      "step": 3544
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.078867271646418e-05,
+      "loss": 1.0883,
+      "step": 3545
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0779498886982125e-05,
+      "loss": 0.9871,
+      "step": 3546
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.077032423524945e-05,
+      "loss": 1.4697,
+      "step": 3547
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0761148762571454e-05,
+      "loss": 1.1,
+      "step": 3548
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.075197247025354e-05,
+      "loss": 1.1744,
+      "step": 3549
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.07427953596012e-05,
+      "loss": 1.0016,
+      "step": 3550
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0733617431920084e-05,
+      "loss": 1.1031,
+      "step": 3551
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0724438688515926e-05,
+      "loss": 1.3726,
+      "step": 3552
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0715259130694604e-05,
+      "loss": 0.8987,
+      "step": 3553
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.070607875976207e-05,
+      "loss": 0.8871,
+      "step": 3554
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.069689757702444e-05,
+      "loss": 1.3872,
+      "step": 3555
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.068771558378791e-05,
+      "loss": 1.01,
+      "step": 3556
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0678532781358805e-05,
+      "loss": 1.1819,
+      "step": 3557
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0669349171043574e-05,
+      "loss": 1.2607,
+      "step": 3558
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0660164754148754e-05,
+      "loss": 0.9574,
+      "step": 3559
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0650979531981046e-05,
+      "loss": 1.0919,
+      "step": 3560
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.06417935058472e-05,
+      "loss": 0.915,
+      "step": 3561
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.063260667705413e-05,
+      "loss": 1.1252,
+      "step": 3562
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0623419046908835e-05,
+      "loss": 1.2227,
+      "step": 3563
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.061423061671846e-05,
+      "loss": 0.9321,
+      "step": 3564
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.060504138779023e-05,
+      "loss": 1.047,
+      "step": 3565
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.059585136143151e-05,
+      "loss": 1.0612,
+      "step": 3566
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.058666053894977e-05,
+      "loss": 0.7708,
+      "step": 3567
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.057746892165257e-05,
+      "loss": 1.1392,
+      "step": 3568
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0568276510847615e-05,
+      "loss": 1.0138,
+      "step": 3569
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.055908330784272e-05,
+      "loss": 1.1605,
+      "step": 3570
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.054988931394579e-05,
+      "loss": 1.2971,
+      "step": 3571
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.054069453046487e-05,
+      "loss": 1.0656,
+      "step": 3572
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.053149895870809e-05,
+      "loss": 1.0438,
+      "step": 3573
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0522302599983716e-05,
+      "loss": 1.0865,
+      "step": 3574
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.051310545560011e-05,
+      "loss": 1.0488,
+      "step": 3575
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.050390752686575e-05,
+      "loss": 1.0951,
+      "step": 3576
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.049470881508924e-05,
+      "loss": 1.1909,
+      "step": 3577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0485509321579277e-05,
+      "loss": 0.964,
+      "step": 3578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0476309047644664e-05,
+      "loss": 1.0365,
+      "step": 3579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0467107994594334e-05,
+      "loss": 1.141,
+      "step": 3580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.045790616373734e-05,
+      "loss": 1.22,
+      "step": 3581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0448703556382797e-05,
+      "loss": 1.2141,
+      "step": 3582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0439500173839976e-05,
+      "loss": 1.0541,
+      "step": 3583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.043029601741825e-05,
+      "loss": 1.0282,
+      "step": 3584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0421091088427093e-05,
+      "loss": 0.9094,
+      "step": 3585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0411885388176086e-05,
+      "loss": 1.1293,
+      "step": 3586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0402678917974937e-05,
+      "loss": 1.2383,
+      "step": 3587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.039347167913345e-05,
+      "loss": 0.9452,
+      "step": 3588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0384263672961532e-05,
+      "loss": 0.9497,
+      "step": 3589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0375054900769206e-05,
+      "loss": 1.2115,
+      "step": 3590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.036584536386663e-05,
+      "loss": 1.011,
+      "step": 3591
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.035663506356402e-05,
+      "loss": 1.0327,
+      "step": 3592
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0347424001171744e-05,
+      "loss": 1.1423,
+      "step": 3593
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0338212178000254e-05,
+      "loss": 1.0011,
+      "step": 3594
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.032899959536012e-05,
+      "loss": 1.2998,
+      "step": 3595
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0319786254562023e-05,
+      "loss": 1.2556,
+      "step": 3596
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0310572156916727e-05,
+      "loss": 1.2246,
+      "step": 3597
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0301357303735156e-05,
+      "loss": 1.1062,
+      "step": 3598
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0292141696328286e-05,
+      "loss": 1.1057,
+      "step": 3599
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.028292533600723e-05,
+      "loss": 1.0714,
+      "step": 3600
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0273708224083203e-05,
+      "loss": 0.8538,
+      "step": 3601
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0264490361867526e-05,
+      "loss": 0.9752,
+      "step": 3602
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0255271750671622e-05,
+      "loss": 1.4552,
+      "step": 3603
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0246052391807023e-05,
+      "loss": 0.9337,
+      "step": 3604
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.023683228658538e-05,
+      "loss": 1.1316,
+      "step": 3605
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0227611436318437e-05,
+      "loss": 1.058,
+      "step": 3606
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0218389842318035e-05,
+      "loss": 1.1398,
+      "step": 3607
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0209167505896142e-05,
+      "loss": 1.2767,
+      "step": 3608
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0199944428364825e-05,
+      "loss": 1.1256,
+      "step": 3609
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.019072061103624e-05,
+      "loss": 1.1337,
+      "step": 3610
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.018149605522267e-05,
+      "loss": 1.4116,
+      "step": 3611
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.01722707622365e-05,
+      "loss": 1.313,
+      "step": 3612
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0163044733390205e-05,
+      "loss": 1.0319,
+      "step": 3613
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0153817969996377e-05,
+      "loss": 1.1003,
+      "step": 3614
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0144590473367712e-05,
+      "loss": 1.2653,
+      "step": 3615
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.013536224481701e-05,
+      "loss": 1.4169,
+      "step": 3616
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0126133285657165e-05,
+      "loss": 1.0509,
+      "step": 3617
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.011690359720119e-05,
+      "loss": 1.0226,
+      "step": 3618
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0107673180762197e-05,
+      "loss": 1.1544,
+      "step": 3619
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.009844203765339e-05,
+      "loss": 1.021,
+      "step": 3620
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.0089210169188086e-05,
+      "loss": 1.1896,
+      "step": 3621
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.007997757667972e-05,
+      "loss": 0.8922,
+      "step": 3622
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.00707442614418e-05,
+      "loss": 1.0027,
+      "step": 3623
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0061510224787952e-05,
+      "loss": 1.1337,
+      "step": 3624
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0052275468031916e-05,
+      "loss": 1.4041,
+      "step": 3625
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0043039992487515e-05,
+      "loss": 1.2042,
+      "step": 3626
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.003380379946869e-05,
+      "loss": 1.2334,
+      "step": 3627
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0024566890289464e-05,
+      "loss": 1.4309,
+      "step": 3628
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0015329266263977e-05,
+      "loss": 1.0472,
+      "step": 3629
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.0006090928706476e-05,
+      "loss": 1.45,
+      "step": 3630
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9996851878931286e-05,
+      "loss": 1.061,
+      "step": 3631
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.998761211825286e-05,
+      "loss": 0.9049,
+      "step": 3632
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.997837164798575e-05,
+      "loss": 1.0878,
+      "step": 3633
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9969130469444585e-05,
+      "loss": 0.9218,
+      "step": 3634
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9959888583944112e-05,
+      "loss": 1.2385,
+      "step": 3635
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9950645992799176e-05,
+      "loss": 1.0623,
+      "step": 3636
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.994140269732473e-05,
+      "loss": 1.0721,
+      "step": 3637
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9932158698835805e-05,
+      "loss": 1.0703,
+      "step": 3638
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9922913998647562e-05,
+      "loss": 0.8921,
+      "step": 3639
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.991366859807524e-05,
+      "loss": 0.9855,
+      "step": 3640
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9904422498434188e-05,
+      "loss": 1.4873,
+      "step": 3641
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.989517570103984e-05,
+      "loss": 1.2759,
+      "step": 3642
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9885928207207753e-05,
+      "loss": 1.1412,
+      "step": 3643
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.987668001825356e-05,
+      "loss": 1.0714,
+      "step": 3644
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9867431135493013e-05,
+      "loss": 1.0687,
+      "step": 3645
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.985818156024194e-05,
+      "loss": 1.321,
+      "step": 3646
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.984893129381629e-05,
+      "loss": 0.8772,
+      "step": 3647
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.98396803375321e-05,
+      "loss": 1.0118,
+      "step": 3648
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9830428692705504e-05,
+      "loss": 0.9733,
+      "step": 3649
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.982117636065273e-05,
+      "loss": 1.224,
+      "step": 3650
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.981192334269011e-05,
+      "loss": 1.1473,
+      "step": 3651
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9802669640134085e-05,
+      "loss": 0.9597,
+      "step": 3652
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.979341525430117e-05,
+      "loss": 1.0858,
+      "step": 3653
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.978416018650799e-05,
+      "loss": 1.155,
+      "step": 3654
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9774904438071277e-05,
+      "loss": 1.0708,
+      "step": 3655
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9765648010307828e-05,
+      "loss": 1.1895,
+      "step": 3656
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.975639090453457e-05,
+      "loss": 0.9669,
+      "step": 3657
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9747133122068515e-05,
+      "loss": 1.3514,
+      "step": 3658
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.973787466422676e-05,
+      "loss": 1.2169,
+      "step": 3659
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.972861553232652e-05,
+      "loss": 0.9646,
+      "step": 3660
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9719355727685082e-05,
+      "loss": 1.1046,
+      "step": 3661
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9710095251619856e-05,
+      "loss": 0.9042,
+      "step": 3662
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9700834105448316e-05,
+      "loss": 0.9541,
+      "step": 3663
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9691572290488058e-05,
+      "loss": 1.0352,
+      "step": 3664
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9682309808056753e-05,
+      "loss": 1.1155,
+      "step": 3665
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.967304665947218e-05,
+      "loss": 1.0153,
+      "step": 3666
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.966378284605221e-05,
+      "loss": 1.1298,
+      "step": 3667
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9654518369114815e-05,
+      "loss": 1.2395,
+      "step": 3668
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9645253229978048e-05,
+      "loss": 1.0187,
+      "step": 3669
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.963598742996006e-05,
+      "loss": 1.1036,
+      "step": 3670
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.96267209703791e-05,
+      "loss": 1.2825,
+      "step": 3671
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9617453852553512e-05,
+      "loss": 1.3732,
+      "step": 3672
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9608186077801725e-05,
+      "loss": 1.017,
+      "step": 3673
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.959891764744227e-05,
+      "loss": 1.1318,
+      "step": 3674
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.958964856279377e-05,
+      "loss": 1.1565,
+      "step": 3675
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.958037882517495e-05,
+      "loss": 1.0524,
+      "step": 3676
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.95711084359046e-05,
+      "loss": 1.1732,
+      "step": 3677
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9561837396301624e-05,
+      "loss": 0.8839,
+      "step": 3678
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9552565707685026e-05,
+      "loss": 1.4551,
+      "step": 3679
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.954329337137388e-05,
+      "loss": 1.1476,
+      "step": 3680
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9534020388687367e-05,
+      "loss": 1.0449,
+      "step": 3681
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9524746760944755e-05,
+      "loss": 1.0624,
+      "step": 3682
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.951547248946541e-05,
+      "loss": 0.9659,
+      "step": 3683
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9506197575568785e-05,
+      "loss": 1.1686,
+      "step": 3684
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.949692202057441e-05,
+      "loss": 0.9675,
+      "step": 3685
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.948764582580194e-05,
+      "loss": 1.0217,
+      "step": 3686
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9478368992571088e-05,
+      "loss": 1.0713,
+      "step": 3687
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9469091522201674e-05,
+      "loss": 1.0709,
+      "step": 3688
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.945981341601361e-05,
+      "loss": 1.127,
+      "step": 3689
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9450534675326892e-05,
+      "loss": 1.1317,
+      "step": 3690
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.94412553014616e-05,
+      "loss": 1.0643,
+      "step": 3691
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9431975295737925e-05,
+      "loss": 1.1584,
+      "step": 3692
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9422694659476135e-05,
+      "loss": 1.2667,
+      "step": 3693
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.941341339399658e-05,
+      "loss": 1.275,
+      "step": 3694
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9404131500619708e-05,
+      "loss": 1.4111,
+      "step": 3695
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.939484898066607e-05,
+      "loss": 1.0348,
+      "step": 3696
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9385565835456273e-05,
+      "loss": 1.0746,
+      "step": 3697
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9376282066311044e-05,
+      "loss": 1.0957,
+      "step": 3698
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9366997674551178e-05,
+      "loss": 1.2911,
+      "step": 3699
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.935771266149758e-05,
+      "loss": 1.0464,
+      "step": 3700
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9348427028471226e-05,
+      "loss": 1.1008,
+      "step": 3701
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9339140776793185e-05,
+      "loss": 1.0623,
+      "step": 3702
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9329853907784603e-05,
+      "loss": 1.1461,
+      "step": 3703
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.932056642276675e-05,
+      "loss": 1.0432,
+      "step": 3704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9311278323060932e-05,
+      "loss": 1.1871,
+      "step": 3705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9301989609988573e-05,
+      "loss": 0.8716,
+      "step": 3706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.92927002848712e-05,
+      "loss": 0.9426,
+      "step": 3707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9283410349030394e-05,
+      "loss": 0.8419,
+      "step": 3708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.927411980378783e-05,
+      "loss": 1.0439,
+      "step": 3709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9264828650465285e-05,
+      "loss": 1.3889,
+      "step": 3710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9255536890384622e-05,
+      "loss": 0.9653,
+      "step": 3711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9246244524867755e-05,
+      "loss": 1.0931,
+      "step": 3712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9236951555236735e-05,
+      "loss": 0.9647,
+      "step": 3713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.922765798281367e-05,
+      "loss": 1.2964,
+      "step": 3714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9218363808920757e-05,
+      "loss": 1.129,
+      "step": 3715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.920906903488027e-05,
+      "loss": 1.1339,
+      "step": 3716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9199773662014596e-05,
+      "loss": 1.1559,
+      "step": 3717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9190477691646178e-05,
+      "loss": 1.1858,
+      "step": 3718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9181181125097556e-05,
+      "loss": 0.9343,
+      "step": 3719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.917188396369136e-05,
+      "loss": 1.3469,
+      "step": 3720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9162586208750297e-05,
+      "loss": 0.9604,
+      "step": 3721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9153287861597162e-05,
+      "loss": 1.0172,
+      "step": 3722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9143988923554833e-05,
+      "loss": 1.2606,
+      "step": 3723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9134689395946263e-05,
+      "loss": 0.9508,
+      "step": 3724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9125389280094518e-05,
+      "loss": 1.0676,
+      "step": 3725
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9116088577322696e-05,
+      "loss": 1.3413,
+      "step": 3726
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.910678728895404e-05,
+      "loss": 1.0845,
+      "step": 3727
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9097485416311838e-05,
+      "loss": 1.5452,
+      "step": 3728
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9088182960719468e-05,
+      "loss": 1.0367,
+      "step": 3729
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.907887992350039e-05,
+      "loss": 1.1237,
+      "step": 3730
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9069576305978148e-05,
+      "loss": 1.1295,
+      "step": 3731
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9060272109476373e-05,
+      "loss": 1.1804,
+      "step": 3732
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9050967335318775e-05,
+      "loss": 0.8458,
+      "step": 3733
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9041661984829145e-05,
+      "loss": 0.9447,
+      "step": 3734
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9032356059331358e-05,
+      "loss": 1.0134,
+      "step": 3735
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9023049560149373e-05,
+      "loss": 1.3171,
+      "step": 3736
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.901374248860722e-05,
+      "loss": 1.0348,
+      "step": 3737
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9004434846029026e-05,
+      "loss": 1.0214,
+      "step": 3738
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8995126633738983e-05,
+      "loss": 1.0769,
+      "step": 3739
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8985817853061385e-05,
+      "loss": 0.7445,
+      "step": 3740
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8976508505320583e-05,
+      "loss": 1.0802,
+      "step": 3741
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8967198591841027e-05,
+      "loss": 1.27,
+      "step": 3742
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.895788811394724e-05,
+      "loss": 1.0328,
+      "step": 3743
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8948577072963824e-05,
+      "loss": 0.828,
+      "step": 3744
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8939265470215453e-05,
+      "loss": 1.1525,
+      "step": 3745
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.892995330702691e-05,
+      "loss": 0.9025,
+      "step": 3746
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.892064058472302e-05,
+      "loss": 1.3641,
+      "step": 3747
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.891132730462872e-05,
+      "loss": 1.2723,
+      "step": 3748
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8902013468069005e-05,
+      "loss": 1.069,
+      "step": 3749
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8892699076368968e-05,
+      "loss": 0.9396,
+      "step": 3750
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.888338413085375e-05,
+      "loss": 0.9824,
+      "step": 3751
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.88740686328486e-05,
+      "loss": 1.0226,
+      "step": 3752
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8864752583678844e-05,
+      "loss": 1.1819,
+      "step": 3753
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8855435984669866e-05,
+      "loss": 1.0873,
+      "step": 3754
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8846118837147145e-05,
+      "loss": 1.0414,
+      "step": 3755
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8836801142436238e-05,
+      "loss": 1.1846,
+      "step": 3756
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8827482901862774e-05,
+      "loss": 1.0388,
+      "step": 3757
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8818164116752456e-05,
+      "loss": 1.1671,
+      "step": 3758
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8808844788431072e-05,
+      "loss": 0.969,
+      "step": 3759
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8799524918224484e-05,
+      "loss": 1.2289,
+      "step": 3760
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.879020450745864e-05,
+      "loss": 1.1354,
+      "step": 3761
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.878088355745955e-05,
+      "loss": 1.3086,
+      "step": 3762
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8771562069553304e-05,
+      "loss": 0.713,
+      "step": 3763
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.876224004506608e-05,
+      "loss": 1.2676,
+      "step": 3764
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8752917485324122e-05,
+      "loss": 1.1672,
+      "step": 3765
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.874359439165375e-05,
+      "loss": 1.2341,
+      "step": 3766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.873427076538136e-05,
+      "loss": 1.0585,
+      "step": 3767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8724946607833437e-05,
+      "loss": 1.3713,
+      "step": 3768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.871562192033652e-05,
+      "loss": 1.0546,
+      "step": 3769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.870629670421724e-05,
+      "loss": 1.1571,
+      "step": 3770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8696970960802304e-05,
+      "loss": 1.1024,
+      "step": 3771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8687644691418473e-05,
+      "loss": 1.0884,
+      "step": 3772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8678317897392603e-05,
+      "loss": 0.9806,
+      "step": 3773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8668990580051624e-05,
+      "loss": 0.9564,
+      "step": 3774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.865966274072254e-05,
+      "loss": 1.2506,
+      "step": 3775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.86503343807324e-05,
+      "loss": 1.1261,
+      "step": 3776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.864100550140838e-05,
+      "loss": 1.0128,
+      "step": 3777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.86316761040777e-05,
+      "loss": 1.01,
+      "step": 3778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8622346190067635e-05,
+      "loss": 0.981,
+      "step": 3779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8613015760705565e-05,
+      "loss": 0.9369,
+      "step": 3780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.860368481731893e-05,
+      "loss": 1.1774,
+      "step": 3781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.859435336123525e-05,
+      "loss": 1.0812,
+      "step": 3782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8585021393782113e-05,
+      "loss": 1.1555,
+      "step": 3783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.857568891628718e-05,
+      "loss": 0.9451,
+      "step": 3784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.856635593007818e-05,
+      "loss": 1.0681,
+      "step": 3785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8557022436482927e-05,
+      "loss": 0.8777,
+      "step": 3786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8547688436829285e-05,
+      "loss": 1.1534,
+      "step": 3787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.853835393244522e-05,
+      "loss": 1.0812,
+      "step": 3788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.8529018924658746e-05,
+      "loss": 1.2894,
+      "step": 3789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.851968341479796e-05,
+      "loss": 0.7773,
+      "step": 3790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8510347404191023e-05,
+      "loss": 1.1538,
+      "step": 3791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.850101089416618e-05,
+      "loss": 1.0819,
+      "step": 3792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.849167388605173e-05,
+      "loss": 0.7917,
+      "step": 3793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.848233638117604e-05,
+      "loss": 1.2427,
+      "step": 3794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8472998380867587e-05,
+      "loss": 1.3992,
+      "step": 3795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8463659886454868e-05,
+      "loss": 1.1683,
+      "step": 3796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8454320899266485e-05,
+      "loss": 1.3477,
+      "step": 3797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.844498142063109e-05,
+      "loss": 1.2625,
+      "step": 3798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8435641451877422e-05,
+      "loss": 0.8071,
+      "step": 3799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8426300994334266e-05,
+      "loss": 1.067,
+      "step": 3800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8416960049330494e-05,
+      "loss": 0.8512,
+      "step": 3801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8407618618195065e-05,
+      "loss": 1.0256,
+      "step": 3802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.839827670225696e-05,
+      "loss": 0.9146,
+      "step": 3803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.838893430284527e-05,
+      "loss": 0.99,
+      "step": 3804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8379591421289137e-05,
+      "loss": 1.2368,
+      "step": 3805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8370248058917775e-05,
+      "loss": 0.8915,
+      "step": 3806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.836090421706047e-05,
+      "loss": 1.0336,
+      "step": 3807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8351559897046552e-05,
+      "loss": 1.2026,
+      "step": 3808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.834221510020547e-05,
+      "loss": 0.905,
+      "step": 3809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.83328698278667e-05,
+      "loss": 0.958,
+      "step": 3810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.832352408135978e-05,
+      "loss": 1.194,
+      "step": 3811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8314177862014352e-05,
+      "loss": 1.0918,
+      "step": 3812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8304831171160095e-05,
+      "loss": 1.1956,
+      "step": 3813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8295484010126767e-05,
+      "loss": 0.8998,
+      "step": 3814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8286136380244178e-05,
+      "loss": 0.928,
+      "step": 3815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.827678828284225e-05,
+      "loss": 1.1559,
+      "step": 3816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.82674397192509e-05,
+      "loss": 0.9169,
+      "step": 3817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8258090690800176e-05,
+      "loss": 1.0641,
+      "step": 3818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8248741198820163e-05,
+      "loss": 1.1981,
+      "step": 3819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8239391244641006e-05,
+      "loss": 1.1178,
+      "step": 3820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8230040829592924e-05,
+      "loss": 0.8907,
+      "step": 3821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8220689955006215e-05,
+      "loss": 0.9539,
+      "step": 3822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8211338622211226e-05,
+      "loss": 1.0934,
+      "step": 3823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8201986832538374e-05,
+      "loss": 1.2419,
+      "step": 3824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.819263458731813e-05,
+      "loss": 1.2786,
+      "step": 3825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8183281887881046e-05,
+      "loss": 1.1763,
+      "step": 3826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.817392873555774e-05,
+      "loss": 1.1919,
+      "step": 3827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8164575131678878e-05,
+      "loss": 1.1299,
+      "step": 3828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.81552210775752e-05,
+      "loss": 1.0148,
+      "step": 3829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8145866574577517e-05,
+      "loss": 0.9946,
+      "step": 3830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.813651162401669e-05,
+      "loss": 0.8947,
+      "step": 3831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.812715622722365e-05,
+      "loss": 1.2148,
+      "step": 3832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.811780038552939e-05,
+      "loss": 1.2979,
+      "step": 3833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8108444100264974e-05,
+      "loss": 1.1117,
+      "step": 3834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.809908737276152e-05,
+      "loss": 1.2928,
+      "step": 3835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.80897302043502e-05,
+      "loss": 1.0453,
+      "step": 3836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.808037259636228e-05,
+      "loss": 1.0744,
+      "step": 3837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.807101455012906e-05,
+      "loss": 0.9606,
+      "step": 3838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.806165606698191e-05,
+      "loss": 0.9615,
+      "step": 3839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8052297148252264e-05,
+      "loss": 0.7034,
+      "step": 3840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8042937795271618e-05,
+      "loss": 1.0409,
+      "step": 3841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.803357800937153e-05,
+      "loss": 0.8722,
+      "step": 3842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.8024217791883616e-05,
+      "loss": 1.0044,
+      "step": 3843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.801485714413956e-05,
+      "loss": 1.2213,
+      "step": 3844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.800549606747111e-05,
+      "loss": 1.2263,
+      "step": 3845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7996134563210053e-05,
+      "loss": 1.0599,
+      "step": 3846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7986772632688256e-05,
+      "loss": 1.2892,
+      "step": 3847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7977410277237653e-05,
+      "loss": 0.92,
+      "step": 3848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.796804749819022e-05,
+      "loss": 0.9973,
+      "step": 3849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7958684296878e-05,
+      "loss": 1.1067,
+      "step": 3850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7949320674633108e-05,
+      "loss": 1.0317,
+      "step": 3851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.79399566327877e-05,
+      "loss": 0.795,
+      "step": 3852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7930592172674003e-05,
+      "loss": 0.6664,
+      "step": 3853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7921227295624297e-05,
+      "loss": 0.8231,
+      "step": 3854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7911862002970933e-05,
+      "loss": 1.1781,
+      "step": 3855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7902496296046304e-05,
+      "loss": 1.2996,
+      "step": 3856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.789313017618288e-05,
+      "loss": 1.2264,
+      "step": 3857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7883763644713173e-05,
+      "loss": 0.9937,
+      "step": 3858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7874396702969774e-05,
+      "loss": 1.223,
+      "step": 3859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7865029352285303e-05,
+      "loss": 1.2321,
+      "step": 3860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7855661593992468e-05,
+      "loss": 1.0992,
+      "step": 3861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7846293429424016e-05,
+      "loss": 1.2772,
+      "step": 3862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.783692485991276e-05,
+      "loss": 0.8707,
+      "step": 3863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7827555886791567e-05,
+      "loss": 0.9318,
+      "step": 3864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7818186511393363e-05,
+      "loss": 1.1738,
+      "step": 3865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.780881673505114e-05,
+      "loss": 1.0487,
+      "step": 3866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.779944655909792e-05,
+      "loss": 0.8932,
+      "step": 3867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7790075984866816e-05,
+      "loss": 1.0159,
+      "step": 3868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7780705013690973e-05,
+      "loss": 0.8848,
+      "step": 3869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7771333646903602e-05,
+      "loss": 0.9285,
+      "step": 3870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7761961885837972e-05,
+      "loss": 1.3126,
+      "step": 3871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7752589731827412e-05,
+      "loss": 1.0375,
+      "step": 3872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.7743217186205295e-05,
+      "loss": 1.1877,
+      "step": 3873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7733844250305045e-05,
+      "loss": 1.2168,
+      "step": 3874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.772447092546016e-05,
+      "loss": 1.6537,
+      "step": 3875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7715097213004192e-05,
+      "loss": 1.0442,
+      "step": 3876
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7705723114270732e-05,
+      "loss": 0.8749,
+      "step": 3877
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7696348630593433e-05,
+      "loss": 1.4861,
+      "step": 3878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7686973763306018e-05,
+      "loss": 0.9498,
+      "step": 3879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7677598513742242e-05,
+      "loss": 1.0022,
+      "step": 3880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7668222883235922e-05,
+      "loss": 0.9418,
+      "step": 3881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.765884687312093e-05,
+      "loss": 1.3464,
+      "step": 3882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7649470484731204e-05,
+      "loss": 1.084,
+      "step": 3883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7640093719400712e-05,
+      "loss": 1.1193,
+      "step": 3884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.76307165784635e-05,
+      "loss": 1.1725,
+      "step": 3885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.762133906325365e-05,
+      "loss": 1.2,
+      "step": 3886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7611961175105304e-05,
+      "loss": 0.9171,
+      "step": 3887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7602582915352653e-05,
+      "loss": 0.95,
+      "step": 3888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7593204285329943e-05,
+      "loss": 1.1305,
+      "step": 3889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7583825286371485e-05,
+      "loss": 1.0225,
+      "step": 3890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.757444591981162e-05,
+      "loss": 0.8705,
+      "step": 3891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7565066186984758e-05,
+      "loss": 1.0497,
+      "step": 3892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7555686089225352e-05,
+      "loss": 1.0511,
+      "step": 3893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7546305627867924e-05,
+      "loss": 0.8572,
+      "step": 3894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7536924804247012e-05,
+      "loss": 0.9247,
+      "step": 3895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7527543619697237e-05,
+      "loss": 0.9502,
+      "step": 3896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7518162075553272e-05,
+      "loss": 0.991,
+      "step": 3897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7508780173149823e-05,
+      "loss": 0.9618,
+      "step": 3898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7499397913821656e-05,
+      "loss": 1.0913,
+      "step": 3899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7490015298903582e-05,
+      "loss": 1.2285,
+      "step": 3900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7480632329730482e-05,
+      "loss": 1.1393,
+      "step": 3901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7471249007637255e-05,
+      "loss": 1.3114,
+      "step": 3902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.746186533395888e-05,
+      "loss": 1.099,
+      "step": 3903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7452481310030376e-05,
+      "loss": 1.2295,
+      "step": 3904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.74430969371868e-05,
+      "loss": 1.0321,
+      "step": 3905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.743371221676328e-05,
+      "loss": 1.1438,
+      "step": 3906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7424327150094975e-05,
+      "loss": 1.0077,
+      "step": 3907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.741494173851711e-05,
+      "loss": 1.0483,
+      "step": 3908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7405555983364934e-05,
+      "loss": 1.0061,
+      "step": 3909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7396169885973765e-05,
+      "loss": 1.0729,
+      "step": 3910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7386783447678976e-05,
+      "loss": 1.3285,
+      "step": 3911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7377396669815967e-05,
+      "loss": 1.1617,
+      "step": 3912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7368009553720204e-05,
+      "loss": 1.1415,
+      "step": 3913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.735862210072719e-05,
+      "loss": 1.1199,
+      "step": 3914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7349234312172484e-05,
+      "loss": 1.0807,
+      "step": 3915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7339846189391683e-05,
+      "loss": 1.0519,
+      "step": 3916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7330457733720428e-05,
+      "loss": 1.0175,
+      "step": 3917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7321068946494437e-05,
+      "loss": 0.845,
+      "step": 3918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.731167982904946e-05,
+      "loss": 0.9492,
+      "step": 3919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7302290382721257e-05,
+      "loss": 1.1179,
+      "step": 3920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7292900608845694e-05,
+      "loss": 1.6335,
+      "step": 3921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.728351050875865e-05,
+      "loss": 0.9374,
+      "step": 3922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.727412008379604e-05,
+      "loss": 1.1096,
+      "step": 3923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7264729335293865e-05,
+      "loss": 0.8525,
+      "step": 3924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7255338264588136e-05,
+      "loss": 0.9684,
+      "step": 3925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.724594687301494e-05,
+      "loss": 1.1772,
+      "step": 3926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7236555161910364e-05,
+      "loss": 1.0197,
+      "step": 3927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.722716313261059e-05,
+      "loss": 0.9423,
+      "step": 3928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7217770786451813e-05,
+      "loss": 1.1157,
+      "step": 3929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.720837812477029e-05,
+      "loss": 1.1102,
+      "step": 3930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7198985148902318e-05,
+      "loss": 0.9943,
+      "step": 3931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.718959186018423e-05,
+      "loss": 1.3989,
+      "step": 3932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.718019825995243e-05,
+      "loss": 1.1708,
+      "step": 3933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7170804349543326e-05,
+      "loss": 0.9861,
+      "step": 3934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.71614101302934e-05,
+      "loss": 1.1775,
+      "step": 3935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7152015603539172e-05,
+      "loss": 0.9246,
+      "step": 3936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.71426207706172e-05,
+      "loss": 1.1975,
+      "step": 3937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.713322563286409e-05,
+      "loss": 0.9639,
+      "step": 3938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.712383019161649e-05,
+      "loss": 1.1298,
+      "step": 3939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.71144344482111e-05,
+      "loss": 0.8929,
+      "step": 3940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.710503840398464e-05,
+      "loss": 0.9731,
+      "step": 3941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7095642060273896e-05,
+      "loss": 1.2607,
+      "step": 3942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7086245418415683e-05,
+      "loss": 1.2549,
+      "step": 3943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.707684847974687e-05,
+      "loss": 0.8936,
+      "step": 3944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7067451245604353e-05,
+      "loss": 1.1055,
+      "step": 3945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7058053717325087e-05,
+      "loss": 1.0789,
+      "step": 3946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.704865589624606e-05,
+      "loss": 1.0538,
+      "step": 3947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7039257783704298e-05,
+      "loss": 1.0269,
+      "step": 3948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.702985938103687e-05,
+      "loss": 1.0555,
+      "step": 3949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7020460689580897e-05,
+      "loss": 0.9041,
+      "step": 3950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7011061710673524e-05,
+      "loss": 0.9574,
+      "step": 3951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.7001662445651954e-05,
+      "loss": 1.0245,
+      "step": 3952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.699226289585342e-05,
+      "loss": 1.144,
+      "step": 3953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.69828630626152e-05,
+      "loss": 1.1805,
+      "step": 3954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6973462947274602e-05,
+      "loss": 1.0623,
+      "step": 3955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.6964062551168988e-05,
+      "loss": 1.2415,
+      "step": 3956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6954661875635757e-05,
+      "loss": 0.9182,
+      "step": 3957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.694526092201234e-05,
+      "loss": 1.2062,
+      "step": 3958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.693585969163621e-05,
+      "loss": 1.2448,
+      "step": 3959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6926458185844892e-05,
+      "loss": 1.1643,
+      "step": 3960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.691705640597595e-05,
+      "loss": 1.1539,
+      "step": 3961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6907654353366947e-05,
+      "loss": 1.303,
+      "step": 3962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6898252029355538e-05,
+      "loss": 0.844,
+      "step": 3963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.688884943527938e-05,
+      "loss": 1.1427,
+      "step": 3964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.687944657247619e-05,
+      "loss": 1.0026,
+      "step": 3965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.687004344228372e-05,
+      "loss": 1.1666,
+      "step": 3966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.686064004603975e-05,
+      "loss": 0.9929,
+      "step": 3967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6851236385082103e-05,
+      "loss": 0.8541,
+      "step": 3968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.684183246074864e-05,
+      "loss": 1.0934,
+      "step": 3969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.683242827437726e-05,
+      "loss": 1.2393,
+      "step": 3970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6823023827305892e-05,
+      "loss": 0.9418,
+      "step": 3971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6813619120872525e-05,
+      "loss": 1.2028,
+      "step": 3972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6804214156415152e-05,
+      "loss": 0.9597,
+      "step": 3973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6794808935271827e-05,
+      "loss": 1.2598,
+      "step": 3974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6785403458780638e-05,
+      "loss": 1.2347,
+      "step": 3975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6775997728279695e-05,
+      "loss": 0.9091,
+      "step": 3976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6766591745107156e-05,
+      "loss": 1.217,
+      "step": 3977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6757185510601213e-05,
+      "loss": 0.9851,
+      "step": 3978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6747779026100095e-05,
+      "loss": 1.0527,
+      "step": 3979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.673837229294206e-05,
+      "loss": 0.8538,
+      "step": 3980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6728965312465408e-05,
+      "loss": 0.9066,
+      "step": 3981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6719558086008484e-05,
+      "loss": 0.9312,
+      "step": 3982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6710150614909633e-05,
+      "loss": 1.1548,
+      "step": 3983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.670074290050728e-05,
+      "loss": 0.9227,
+      "step": 3984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6691334944139846e-05,
+      "loss": 1.1819,
+      "step": 3985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6681926747145815e-05,
+      "loss": 1.1927,
+      "step": 3986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6672518310863682e-05,
+      "loss": 1.4073,
+      "step": 3987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6663109636632e-05,
+      "loss": 1.3123,
+      "step": 3988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6653700725789338e-05,
+      "loss": 1.3119,
+      "step": 3989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6644291579674303e-05,
+      "loss": 1.5842,
+      "step": 3990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6634882199625525e-05,
+      "loss": 1.0095,
+      "step": 3991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6625472586981704e-05,
+      "loss": 1.3013,
+      "step": 3992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6616062743081527e-05,
+      "loss": 1.0752,
+      "step": 3993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6606652669263736e-05,
+      "loss": 1.2207,
+      "step": 3994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.659724236686712e-05,
+      "loss": 1.2839,
+      "step": 3995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.658783183723047e-05,
+      "loss": 0.8051,
+      "step": 3996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6578421081692628e-05,
+      "loss": 1.1853,
+      "step": 3997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.656901010159246e-05,
+      "loss": 0.8842,
+      "step": 3998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6559598898268883e-05,
+      "loss": 1.3477,
+      "step": 3999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6550187473060816e-05,
+      "loss": 0.9375,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "eval_loss": 1.1513671875,
+      "eval_runtime": 3.6057,
+      "eval_samples_per_second": 27.734,
+      "eval_steps_per_second": 4.715,
+      "step": 4000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.654077582730723e-05,
+      "loss": 1.0817,
+      "step": 4001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6531363962347115e-05,
+      "loss": 1.3069,
+      "step": 4002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6521951879519517e-05,
+      "loss": 1.2588,
+      "step": 4003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.651253958016348e-05,
+      "loss": 1.0619,
+      "step": 4004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6503127065618088e-05,
+      "loss": 1.3103,
+      "step": 4005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6493714337222487e-05,
+      "loss": 0.8225,
+      "step": 4006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6484301396315802e-05,
+      "loss": 1.3535,
+      "step": 4007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6474888244237227e-05,
+      "loss": 1.0233,
+      "step": 4008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.646547488232597e-05,
+      "loss": 0.9478,
+      "step": 4009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6456061311921278e-05,
+      "loss": 0.8699,
+      "step": 4010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.644664753436241e-05,
+      "loss": 1.2329,
+      "step": 4011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6437233550988673e-05,
+      "loss": 1.3704,
+      "step": 4012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.64278193631394e-05,
+      "loss": 1.0325,
+      "step": 4013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.641840497215395e-05,
+      "loss": 1.0331,
+      "step": 4014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6408990379371695e-05,
+      "loss": 0.813,
+      "step": 4015
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.639957558613207e-05,
+      "loss": 0.8783,
+      "step": 4016
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6390160593774515e-05,
+      "loss": 1.2146,
+      "step": 4017
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6380745403638497e-05,
+      "loss": 0.8674,
+      "step": 4018
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6371330017063517e-05,
+      "loss": 1.1202,
+      "step": 4019
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6361914435389108e-05,
+      "loss": 0.8285,
+      "step": 4020
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6352498659954838e-05,
+      "loss": 1.1144,
+      "step": 4021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.634308269210027e-05,
+      "loss": 1.2841,
+      "step": 4022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.633366653316503e-05,
+      "loss": 1.6042,
+      "step": 4023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.632425018448875e-05,
+      "loss": 1.2251,
+      "step": 4024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6314833647411108e-05,
+      "loss": 1.002,
+      "step": 4025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6305416923271775e-05,
+      "loss": 1.3245,
+      "step": 4026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6296000013410487e-05,
+      "loss": 1.2657,
+      "step": 4027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6286582919166996e-05,
+      "loss": 0.9725,
+      "step": 4028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6277165641881057e-05,
+      "loss": 1.2258,
+      "step": 4029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6267748182892476e-05,
+      "loss": 1.3403,
+      "step": 4030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6258330543541077e-05,
+      "loss": 0.85,
+      "step": 4031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6248912725166714e-05,
+      "loss": 0.939,
+      "step": 4032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.623949472910925e-05,
+      "loss": 0.9022,
+      "step": 4033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6230076556708592e-05,
+      "loss": 1.067,
+      "step": 4034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6220658209304684e-05,
+      "loss": 1.4669,
+      "step": 4035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6211239688237442e-05,
+      "loss": 1.0211,
+      "step": 4036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.620182099484686e-05,
+      "loss": 1.0435,
+      "step": 4037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6192402130472936e-05,
+      "loss": 1.2292,
+      "step": 4038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.6182983096455694e-05,
+      "loss": 1.1278,
+      "step": 4039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6173563894135184e-05,
+      "loss": 1.3451,
+      "step": 4040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6164144524851477e-05,
+      "loss": 0.9766,
+      "step": 4041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6154724989944662e-05,
+      "loss": 1.1221,
+      "step": 4042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.614530529075487e-05,
+      "loss": 1.0024,
+      "step": 4043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.613588542862223e-05,
+      "loss": 0.8168,
+      "step": 4044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6126465404886918e-05,
+      "loss": 0.9498,
+      "step": 4045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.611704522088912e-05,
+      "loss": 1.2786,
+      "step": 4046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.610762487796905e-05,
+      "loss": 0.9893,
+      "step": 4047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.609820437746694e-05,
+      "loss": 0.7427,
+      "step": 4048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.608878372072305e-05,
+      "loss": 1.0192,
+      "step": 4049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6079362909077652e-05,
+      "loss": 1.1025,
+      "step": 4050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.606994194387105e-05,
+      "loss": 0.8831,
+      "step": 4051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.606052082644357e-05,
+      "loss": 0.6434,
+      "step": 4052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.605109955813555e-05,
+      "loss": 1.1748,
+      "step": 4053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6041678140287364e-05,
+      "loss": 1.1726,
+      "step": 4054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6032256574239394e-05,
+      "loss": 1.2231,
+      "step": 4055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6022834861332053e-05,
+      "loss": 0.8105,
+      "step": 4056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6013413002905763e-05,
+      "loss": 0.9291,
+      "step": 4057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.6003991000300977e-05,
+      "loss": 1.2656,
+      "step": 4058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.599456885485817e-05,
+      "loss": 1.0023,
+      "step": 4059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5985146567917823e-05,
+      "loss": 1.0038,
+      "step": 4060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5975724140820456e-05,
+      "loss": 1.0157,
+      "step": 4061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.59663015749066e-05,
+      "loss": 1.2053,
+      "step": 4062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5956878871516806e-05,
+      "loss": 1.1625,
+      "step": 4063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.594745603199163e-05,
+      "loss": 1.0564,
+      "step": 4064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.593803305767168e-05,
+      "loss": 0.9465,
+      "step": 4065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.592860994989755e-05,
+      "loss": 1.0895,
+      "step": 4066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5919186710009873e-05,
+      "loss": 0.9587,
+      "step": 4067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5909763339349303e-05,
+      "loss": 1.0201,
+      "step": 4068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5900339839256494e-05,
+      "loss": 1.1057,
+      "step": 4069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.589091621107214e-05,
+      "loss": 1.2838,
+      "step": 4070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5881492456136936e-05,
+      "loss": 1.0997,
+      "step": 4071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5872068575791596e-05,
+      "loss": 1.0076,
+      "step": 4072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.586264457137687e-05,
+      "loss": 0.9359,
+      "step": 4073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.58532204442335e-05,
+      "loss": 0.83,
+      "step": 4074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5843796195702274e-05,
+      "loss": 1.0176,
+      "step": 4075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.583437182712397e-05,
+      "loss": 1.3296,
+      "step": 4076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5824947339839406e-05,
+      "loss": 1.2894,
+      "step": 4077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.581552273518939e-05,
+      "loss": 0.976,
+      "step": 4078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5806098014514778e-05,
+      "loss": 1.2369,
+      "step": 4079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5796673179156416e-05,
+      "loss": 1.1606,
+      "step": 4080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.578724823045518e-05,
+      "loss": 1.2233,
+      "step": 4081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5777823169751967e-05,
+      "loss": 0.9287,
+      "step": 4082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5768397998387673e-05,
+      "loss": 0.7772,
+      "step": 4083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5758972717703232e-05,
+      "loss": 1.1132,
+      "step": 4084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.574954732903957e-05,
+      "loss": 1.3114,
+      "step": 4085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5740121833737635e-05,
+      "loss": 1.1255,
+      "step": 4086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5730696233138402e-05,
+      "loss": 1.0637,
+      "step": 4087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5721270528582852e-05,
+      "loss": 0.9866,
+      "step": 4088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.571184472141198e-05,
+      "loss": 1.2771,
+      "step": 4089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5702418812966806e-05,
+      "loss": 0.9723,
+      "step": 4090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.569299280458835e-05,
+      "loss": 1.2471,
+      "step": 4091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.568356669761765e-05,
+      "loss": 1.2213,
+      "step": 4092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5674140493395765e-05,
+      "loss": 1.3842,
+      "step": 4093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5664714193263757e-05,
+      "loss": 1.0623,
+      "step": 4094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.565528779856271e-05,
+      "loss": 1.1576,
+      "step": 4095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5645861310633722e-05,
+      "loss": 1.0681,
+      "step": 4096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5636434730817903e-05,
+      "loss": 0.9899,
+      "step": 4097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5627008060456376e-05,
+      "loss": 1.2168,
+      "step": 4098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.561758130089027e-05,
+      "loss": 1.3126,
+      "step": 4099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5608154453460726e-05,
+      "loss": 1.1031,
+      "step": 4100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.559872751950892e-05,
+      "loss": 0.8368,
+      "step": 4101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5589300500376012e-05,
+      "loss": 1.1547,
+      "step": 4102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5579873397403188e-05,
+      "loss": 1.187,
+      "step": 4103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.557044621193165e-05,
+      "loss": 1.2656,
+      "step": 4104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.55610189453026e-05,
+      "loss": 1.1818,
+      "step": 4105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5551591598857254e-05,
+      "loss": 1.3118,
+      "step": 4106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.554216417393685e-05,
+      "loss": 0.8501,
+      "step": 4107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.553273667188263e-05,
+      "loss": 0.87,
+      "step": 4108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.552330909403584e-05,
+      "loss": 0.8868,
+      "step": 4109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5513881441737753e-05,
+      "loss": 0.9999,
+      "step": 4110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5504453716329635e-05,
+      "loss": 1.2026,
+      "step": 4111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5495025919152782e-05,
+      "loss": 1.1616,
+      "step": 4112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.548559805154847e-05,
+      "loss": 1.1624,
+      "step": 4113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.547617011485801e-05,
+      "loss": 0.7328,
+      "step": 4114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.546674211042273e-05,
+      "loss": 1.264,
+      "step": 4115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5457314039583946e-05,
+      "loss": 1.1868,
+      "step": 4116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5447885903682988e-05,
+      "loss": 0.8885,
+      "step": 4117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5438457704061202e-05,
+      "loss": 1.1123,
+      "step": 4118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.542902944205995e-05,
+      "loss": 1.056,
+      "step": 4119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.541960111902057e-05,
+      "loss": 1.0126,
+      "step": 4120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5410172736284443e-05,
+      "loss": 1.1879,
+      "step": 4121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.5400744295192953e-05,
+      "loss": 1.0042,
+      "step": 4122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.539131579708749e-05,
+      "loss": 1.1431,
+      "step": 4123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.538188724330943e-05,
+      "loss": 1.0731,
+      "step": 4124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5372458635200192e-05,
+      "loss": 1.1338,
+      "step": 4125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5363029974101178e-05,
+      "loss": 1.129,
+      "step": 4126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5353601261353815e-05,
+      "loss": 1.3665,
+      "step": 4127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5344172498299505e-05,
+      "loss": 1.0684,
+      "step": 4128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.533474368627971e-05,
+      "loss": 0.8275,
+      "step": 4129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5325314826635853e-05,
+      "loss": 1.3013,
+      "step": 4130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5315885920709386e-05,
+      "loss": 1.2766,
+      "step": 4131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5306456969841752e-05,
+      "loss": 1.142,
+      "step": 4132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.529702797537442e-05,
+      "loss": 1.1741,
+      "step": 4133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.528759893864885e-05,
+      "loss": 1.0737,
+      "step": 4134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.527816986100651e-05,
+      "loss": 1.1182,
+      "step": 4135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5268740743788888e-05,
+      "loss": 1.0024,
+      "step": 4136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5259311588337465e-05,
+      "loss": 0.8925,
+      "step": 4137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.524988239599371e-05,
+      "loss": 0.9546,
+      "step": 4138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.524045316809914e-05,
+      "loss": 1.2053,
+      "step": 4139
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5231023905995245e-05,
+      "loss": 1.2483,
+      "step": 4140
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.522159461102352e-05,
+      "loss": 0.8617,
+      "step": 4141
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5212165284525485e-05,
+      "loss": 1.2485,
+      "step": 4142
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5202735927842642e-05,
+      "loss": 1.1235,
+      "step": 4143
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5193306542316525e-05,
+      "loss": 1.0698,
+      "step": 4144
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5183877129288634e-05,
+      "loss": 0.9886,
+      "step": 4145
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.51744476901005e-05,
+      "loss": 1.1371,
+      "step": 4146
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5165018226093656e-05,
+      "loss": 1.4045,
+      "step": 4147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5155588738609637e-05,
+      "loss": 1.2054,
+      "step": 4148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5146159228989968e-05,
+      "loss": 1.1609,
+      "step": 4149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5136729698576195e-05,
+      "loss": 1.2103,
+      "step": 4150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.512730014870986e-05,
+      "loss": 1.1375,
+      "step": 4151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.51178705807325e-05,
+      "loss": 1.2052,
+      "step": 4152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5108440995985667e-05,
+      "loss": 0.9625,
+      "step": 4153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5099011395810905e-05,
+      "loss": 1.2383,
+      "step": 4154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5089581781549775e-05,
+      "loss": 0.9574,
+      "step": 4155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5080152154543824e-05,
+      "loss": 0.9369,
+      "step": 4156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.507072251613461e-05,
+      "loss": 1.2847,
+      "step": 4157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.506129286766369e-05,
+      "loss": 1.0421,
+      "step": 4158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5051863210472616e-05,
+      "loss": 1.0574,
+      "step": 4159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.504243354590295e-05,
+      "loss": 0.8939,
+      "step": 4160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5033003875296252e-05,
+      "loss": 1.1747,
+      "step": 4161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.502357419999409e-05,
+      "loss": 1.2882,
+      "step": 4162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5014144521338014e-05,
+      "loss": 0.9987,
+      "step": 4163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.5004714840669595e-05,
+      "loss": 0.9786,
+      "step": 4164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4995285159330407e-05,
+      "loss": 1.2795,
+      "step": 4165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.498585547866199e-05,
+      "loss": 0.9351,
+      "step": 4166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.497642580000592e-05,
+      "loss": 1.0524,
+      "step": 4167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4966996124703754e-05,
+      "loss": 0.9728,
+      "step": 4168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4957566454097063e-05,
+      "loss": 1.196,
+      "step": 4169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.49481367895274e-05,
+      "loss": 1.6565,
+      "step": 4170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4938707132336323e-05,
+      "loss": 0.9751,
+      "step": 4171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.492927748386539e-05,
+      "loss": 0.9999,
+      "step": 4172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4919847845456175e-05,
+      "loss": 1.2676,
+      "step": 4173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4910418218450224e-05,
+      "loss": 1.2761,
+      "step": 4174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4900988604189097e-05,
+      "loss": 1.2916,
+      "step": 4175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.489155900401434e-05,
+      "loss": 1.2173,
+      "step": 4176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4882129419267505e-05,
+      "loss": 1.0818,
+      "step": 4177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4872699851290146e-05,
+      "loss": 0.9467,
+      "step": 4178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.486327030142381e-05,
+      "loss": 0.7906,
+      "step": 4179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4853840771010038e-05,
+      "loss": 1.2034,
+      "step": 4180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.484441126139037e-05,
+      "loss": 0.9089,
+      "step": 4181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4834981773906347e-05,
+      "loss": 1.0851,
+      "step": 4182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4825552309899506e-05,
+      "loss": 0.829,
+      "step": 4183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.481612287071138e-05,
+      "loss": 1.1041,
+      "step": 4184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4806693457683484e-05,
+      "loss": 1.1193,
+      "step": 4185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4797264072157357e-05,
+      "loss": 1.2767,
+      "step": 4186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4787834715474517e-05,
+      "loss": 1.2407,
+      "step": 4187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.477840538897648e-05,
+      "loss": 0.8976,
+      "step": 4188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4768976094004764e-05,
+      "loss": 1.1489,
+      "step": 4189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4759546831900864e-05,
+      "loss": 0.8269,
+      "step": 4190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.475011760400629e-05,
+      "loss": 1.5598,
+      "step": 4191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4740688411662545e-05,
+      "loss": 0.9564,
+      "step": 4192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4731259256211118e-05,
+      "loss": 0.9192,
+      "step": 4193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4721830138993492e-05,
+      "loss": 1.1902,
+      "step": 4194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4712401061351156e-05,
+      "loss": 1.0421,
+      "step": 4195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4702972024625586e-05,
+      "loss": 1.2744,
+      "step": 4196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.469354303015826e-05,
+      "loss": 1.2428,
+      "step": 4197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4684114079290627e-05,
+      "loss": 1.0599,
+      "step": 4198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.467468517336415e-05,
+      "loss": 1.1058,
+      "step": 4199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.466525631372029e-05,
+      "loss": 1.1042,
+      "step": 4200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4655827501700494e-05,
+      "loss": 1.2207,
+      "step": 4201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4646398738646194e-05,
+      "loss": 0.8665,
+      "step": 4202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4636970025898828e-05,
+      "loss": 1.1643,
+      "step": 4203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.4627541364799814e-05,
+      "loss": 1.0948,
+      "step": 4204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.461811275669057e-05,
+      "loss": 1.1146,
+      "step": 4205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.460868420291252e-05,
+      "loss": 1.2668,
+      "step": 4206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4599255704807053e-05,
+      "loss": 1.2639,
+      "step": 4207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4589827263715563e-05,
+      "loss": 1.2065,
+      "step": 4208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.458039888097944e-05,
+      "loss": 0.8767,
+      "step": 4209
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4570970557940064e-05,
+      "loss": 0.9127,
+      "step": 4210
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4561542295938807e-05,
+      "loss": 0.9298,
+      "step": 4211
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.455211409631702e-05,
+      "loss": 1.1925,
+      "step": 4212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4542685960416053e-05,
+      "loss": 0.7905,
+      "step": 4213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.453325788957727e-05,
+      "loss": 1.2271,
+      "step": 4214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.452382988514199e-05,
+      "loss": 1.2137,
+      "step": 4215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4514401948451536e-05,
+      "loss": 1.3168,
+      "step": 4216
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4504974080847227e-05,
+      "loss": 0.9625,
+      "step": 4217
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.449554628367037e-05,
+      "loss": 0.927,
+      "step": 4218
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4486118558262253e-05,
+      "loss": 1.1338,
+      "step": 4219
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.447669090596416e-05,
+      "loss": 0.9807,
+      "step": 4220
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4467263328117373e-05,
+      "loss": 0.993,
+      "step": 4221
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4457835826063156e-05,
+      "loss": 1.0404,
+      "step": 4222
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4448408401142755e-05,
+      "loss": 1.1157,
+      "step": 4223
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.443898105469741e-05,
+      "loss": 1.0076,
+      "step": 4224
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4429553788068367e-05,
+      "loss": 0.7812,
+      "step": 4225
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4420126602596814e-05,
+      "loss": 1.2299,
+      "step": 4226
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.441069949962399e-05,
+      "loss": 0.8953,
+      "step": 4227
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4401272480491087e-05,
+      "loss": 1.1897,
+      "step": 4228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4391845546539277e-05,
+      "loss": 1.5974,
+      "step": 4229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4382418699109737e-05,
+      "loss": 1.0942,
+      "step": 4230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4372991939543627e-05,
+      "loss": 1.0612,
+      "step": 4231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.43635652691821e-05,
+      "loss": 1.1764,
+      "step": 4232
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.435413868936628e-05,
+      "loss": 1.4445,
+      "step": 4233
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4344712201437293e-05,
+      "loss": 1.5046,
+      "step": 4234
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.433528580673625e-05,
+      "loss": 0.926,
+      "step": 4235
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4325859506604248e-05,
+      "loss": 1.1187,
+      "step": 4236
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4316433302382356e-05,
+      "loss": 1.0785,
+      "step": 4237
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.430700719541166e-05,
+      "loss": 1.2302,
+      "step": 4238
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.42975811870332e-05,
+      "loss": 0.8024,
+      "step": 4239
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4288155278588014e-05,
+      "loss": 1.0017,
+      "step": 4240
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4278729471417147e-05,
+      "loss": 1.2427,
+      "step": 4241
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4269303766861604e-05,
+      "loss": 1.2535,
+      "step": 4242
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.425987816626237e-05,
+      "loss": 1.005,
+      "step": 4243
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.425045267096044e-05,
+      "loss": 0.8784,
+      "step": 4244
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.424102728229677e-05,
+      "loss": 1.2996,
+      "step": 4245
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.423160200161233e-05,
+      "loss": 1.2649,
+      "step": 4246
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.422217683024804e-05,
+      "loss": 0.8777,
+      "step": 4247
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4212751769544824e-05,
+      "loss": 1.1971,
+      "step": 4248
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.420332682084359e-05,
+      "loss": 1.3027,
+      "step": 4249
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4193901985485234e-05,
+      "loss": 1.2697,
+      "step": 4250
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4184477264810616e-05,
+      "loss": 1.014,
+      "step": 4251
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4175052660160603e-05,
+      "loss": 1.0944,
+      "step": 4252
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4165628172876027e-05,
+      "loss": 1.3755,
+      "step": 4253
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.415620380429773e-05,
+      "loss": 0.9968,
+      "step": 4254
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4146779555766496e-05,
+      "loss": 0.9729,
+      "step": 4255
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4137355428623138e-05,
+      "loss": 1.4142,
+      "step": 4256
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.412793142420841e-05,
+      "loss": 0.9351,
+      "step": 4257
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4118507543863073e-05,
+      "loss": 1.1935,
+      "step": 4258
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4109083788927862e-05,
+      "loss": 1.3853,
+      "step": 4259
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4099660160743512e-05,
+      "loss": 1.0071,
+      "step": 4260
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4090236660650703e-05,
+      "loss": 1.5024,
+      "step": 4261
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4080813289990133e-05,
+      "loss": 0.9539,
+      "step": 4262
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4071390050102457e-05,
+      "loss": 1.1135,
+      "step": 4263
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4061966942328336e-05,
+      "loss": 1.123,
+      "step": 4264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.405254396800838e-05,
+      "loss": 0.9452,
+      "step": 4265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4043121128483207e-05,
+      "loss": 1.52,
+      "step": 4266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.40336984250934e-05,
+      "loss": 0.9491,
+      "step": 4267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.402427585917954e-05,
+      "loss": 0.9915,
+      "step": 4268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4014853432082173e-05,
+      "loss": 0.9741,
+      "step": 4269
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.4005431145141837e-05,
+      "loss": 1.4944,
+      "step": 4270
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3996008999699025e-05,
+      "loss": 0.9323,
+      "step": 4271
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3986586997094243e-05,
+      "loss": 1.1444,
+      "step": 4272
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3977165138667953e-05,
+      "loss": 1.0947,
+      "step": 4273
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3967743425760615e-05,
+      "loss": 1.3136,
+      "step": 4274
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3958321859712642e-05,
+      "loss": 1.2072,
+      "step": 4275
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3948900441864454e-05,
+      "loss": 0.8481,
+      "step": 4276
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3939479173556437e-05,
+      "loss": 1.228,
+      "step": 4277
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3930058056128962e-05,
+      "loss": 1.0629,
+      "step": 4278
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.392063709092236e-05,
+      "loss": 0.9921,
+      "step": 4279
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.391121627927696e-05,
+      "loss": 1.0938,
+      "step": 4280
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.390179562253306e-05,
+      "loss": 1.0989,
+      "step": 4281
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.389237512203095e-05,
+      "loss": 1.0536,
+      "step": 4282
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.388295477911088e-05,
+      "loss": 1.1265,
+      "step": 4283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3873534595113088e-05,
+      "loss": 1.5273,
+      "step": 4284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3864114571377773e-05,
+      "loss": 1.2222,
+      "step": 4285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3854694709245136e-05,
+      "loss": 0.8062,
+      "step": 4286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.384527501005534e-05,
+      "loss": 1.1729,
+      "step": 4287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3835855475148533e-05,
+      "loss": 1.0363,
+      "step": 4288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.3826436105864822e-05,
+      "loss": 1.01,
+      "step": 4289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.381701690354431e-05,
+      "loss": 0.754,
+      "step": 4290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3807597869527067e-05,
+      "loss": 0.8248,
+      "step": 4291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.379817900515315e-05,
+      "loss": 1.1764,
+      "step": 4292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.378876031176257e-05,
+      "loss": 1.2185,
+      "step": 4293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3779341790695326e-05,
+      "loss": 0.929,
+      "step": 4294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3769923443291404e-05,
+      "loss": 0.9995,
+      "step": 4295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.376050527089075e-05,
+      "loss": 0.9899,
+      "step": 4296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.375108727483329e-05,
+      "loss": 1.0863,
+      "step": 4297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3741669456458925e-05,
+      "loss": 1.1828,
+      "step": 4298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.373225181710753e-05,
+      "loss": 1.0389,
+      "step": 4299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.372283435811895e-05,
+      "loss": 1.1377,
+      "step": 4300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.371341708083301e-05,
+      "loss": 1.1489,
+      "step": 4301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.370399998658952e-05,
+      "loss": 0.9573,
+      "step": 4302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.369458307672823e-05,
+      "loss": 0.9951,
+      "step": 4303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3685166352588904e-05,
+      "loss": 1.0276,
+      "step": 4304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3675749815511255e-05,
+      "loss": 0.899,
+      "step": 4305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3666333466834983e-05,
+      "loss": 1.0879,
+      "step": 4306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.365691730789974e-05,
+      "loss": 1.0624,
+      "step": 4307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3647501340045164e-05,
+      "loss": 0.9105,
+      "step": 4308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.363808556461089e-05,
+      "loss": 1.1549,
+      "step": 4309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3628669982936485e-05,
+      "loss": 0.8657,
+      "step": 4310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.361925459636151e-05,
+      "loss": 0.9341,
+      "step": 4311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.360983940622549e-05,
+      "loss": 1.0808,
+      "step": 4312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3600424413867936e-05,
+      "loss": 1.1106,
+      "step": 4313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.359100962062831e-05,
+      "loss": 1.1342,
+      "step": 4314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.358159502784606e-05,
+      "loss": 1.2644,
+      "step": 4315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3572180636860603e-05,
+      "loss": 1.1326,
+      "step": 4316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3562766449011333e-05,
+      "loss": 1.0563,
+      "step": 4317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.35533524656376e-05,
+      "loss": 0.8373,
+      "step": 4318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.354393868807873e-05,
+      "loss": 0.9789,
+      "step": 4319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.353452511767404e-05,
+      "loss": 1.2393,
+      "step": 4320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3525111755762782e-05,
+      "loss": 1.119,
+      "step": 4321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3515698603684197e-05,
+      "loss": 1.293,
+      "step": 4322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.350628566277752e-05,
+      "loss": 1.2043,
+      "step": 4323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.349687293438191e-05,
+      "loss": 1.1824,
+      "step": 4324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3487460419836528e-05,
+      "loss": 0.8984,
+      "step": 4325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3478048120480485e-05,
+      "loss": 1.0865,
+      "step": 4326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3468636037652887e-05,
+      "loss": 1.3188,
+      "step": 4327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.345922417269278e-05,
+      "loss": 1.2308,
+      "step": 4328
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.344981252693919e-05,
+      "loss": 0.9276,
+      "step": 4329
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3440401101731123e-05,
+      "loss": 0.973,
+      "step": 4330
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3430989898407545e-05,
+      "loss": 1.0186,
+      "step": 4331
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.342157891830738e-05,
+      "loss": 1.241,
+      "step": 4332
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.341216816276954e-05,
+      "loss": 0.9397,
+      "step": 4333
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3402757633132887e-05,
+      "loss": 1.1635,
+      "step": 4334
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.339334733073626e-05,
+      "loss": 0.8735,
+      "step": 4335
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3383937256918475e-05,
+      "loss": 0.7655,
+      "step": 4336
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.33745274130183e-05,
+      "loss": 1.1309,
+      "step": 4337
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3365117800374477e-05,
+      "loss": 0.8737,
+      "step": 4338
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3355708420325707e-05,
+      "loss": 0.8618,
+      "step": 4339
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3346299274210668e-05,
+      "loss": 1.1519,
+      "step": 4340
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.333689036336801e-05,
+      "loss": 1.2206,
+      "step": 4341
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.332748168913632e-05,
+      "loss": 1.14,
+      "step": 4342
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.331807325285419e-05,
+      "loss": 1.0753,
+      "step": 4343
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.330866505586016e-05,
+      "loss": 1.0422,
+      "step": 4344
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3299257099492732e-05,
+      "loss": 1.0574,
+      "step": 4345
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3289849385090373e-05,
+      "loss": 1.1478,
+      "step": 4346
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3280441913991526e-05,
+      "loss": 1.0428,
+      "step": 4347
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3271034687534594e-05,
+      "loss": 0.9268,
+      "step": 4348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.326162770705794e-05,
+      "loss": 1.3085,
+      "step": 4349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3252220973899908e-05,
+      "loss": 0.9883,
+      "step": 4350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3242814489398793e-05,
+      "loss": 1.0173,
+      "step": 4351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.323340825489285e-05,
+      "loss": 1.057,
+      "step": 4352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.322400227172031e-05,
+      "loss": 0.9612,
+      "step": 4353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3214596541219368e-05,
+      "loss": 0.9292,
+      "step": 4354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.320519106472818e-05,
+      "loss": 1.1415,
+      "step": 4355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3195785843584854e-05,
+      "loss": 1.1296,
+      "step": 4356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3186380879127484e-05,
+      "loss": 0.9545,
+      "step": 4357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3176976172694114e-05,
+      "loss": 1.1526,
+      "step": 4358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3167571725622757e-05,
+      "loss": 1.2202,
+      "step": 4359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.315816753925137e-05,
+      "loss": 1.0775,
+      "step": 4360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3148763614917906e-05,
+      "loss": 0.965,
+      "step": 4361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.313935995396025e-05,
+      "loss": 1.1194,
+      "step": 4362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.312995655771628e-05,
+      "loss": 0.9736,
+      "step": 4363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.312055342752381e-05,
+      "loss": 1.0005,
+      "step": 4364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3111150564720627e-05,
+      "loss": 1.0759,
+      "step": 4365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.310174797064447e-05,
+      "loss": 1.3882,
+      "step": 4366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.309234564663306e-05,
+      "loss": 0.9017,
+      "step": 4367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.308294359402406e-05,
+      "loss": 1.2994,
+      "step": 4368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.307354181415511e-05,
+      "loss": 0.8955,
+      "step": 4369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3064140308363792e-05,
+      "loss": 1.0133,
+      "step": 4370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.305473907798767e-05,
+      "loss": 1.0344,
+      "step": 4371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.3045338124364252e-05,
+      "loss": 1.0677,
+      "step": 4372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3035937448831024e-05,
+      "loss": 1.1628,
+      "step": 4373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3026537052725407e-05,
+      "loss": 0.8965,
+      "step": 4374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.3017136937384812e-05,
+      "loss": 0.9427,
+      "step": 4375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.300773710414658e-05,
+      "loss": 0.9442,
+      "step": 4376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2998337554348042e-05,
+      "loss": 1.1562,
+      "step": 4377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.298893828932647e-05,
+      "loss": 1.0861,
+      "step": 4378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2979539310419106e-05,
+      "loss": 1.0321,
+      "step": 4379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.297014061896313e-05,
+      "loss": 1.2247,
+      "step": 4380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2960742216295704e-05,
+      "loss": 0.9954,
+      "step": 4381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2951344103753942e-05,
+      "loss": 0.899,
+      "step": 4382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2941946282674915e-05,
+      "loss": 1.1914,
+      "step": 4383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2932548754395653e-05,
+      "loss": 1.0753,
+      "step": 4384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.292315152025314e-05,
+      "loss": 1.0789,
+      "step": 4385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2913754581584322e-05,
+      "loss": 1.1287,
+      "step": 4386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2904357939726117e-05,
+      "loss": 1.0394,
+      "step": 4387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.289496159601537e-05,
+      "loss": 1.3079,
+      "step": 4388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.288556555178891e-05,
+      "loss": 1.2068,
+      "step": 4389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.287616980838351e-05,
+      "loss": 0.97,
+      "step": 4390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2866774367135912e-05,
+      "loss": 1.077,
+      "step": 4391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2857379229382802e-05,
+      "loss": 1.1102,
+      "step": 4392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2847984396460837e-05,
+      "loss": 1.0695,
+      "step": 4393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2838589869706605e-05,
+      "loss": 1.0978,
+      "step": 4394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.282919565045668e-05,
+      "loss": 1.0566,
+      "step": 4395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2819801740047574e-05,
+      "loss": 0.9757,
+      "step": 4396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2810408139815774e-05,
+      "loss": 0.9597,
+      "step": 4397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2801014851097688e-05,
+      "loss": 1.1503,
+      "step": 4398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2791621875229716e-05,
+      "loss": 1.1271,
+      "step": 4399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2782229213548193e-05,
+      "loss": 1.2981,
+      "step": 4400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2772836867389424e-05,
+      "loss": 1.0648,
+      "step": 4401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2763444838089645e-05,
+      "loss": 1.3167,
+      "step": 4402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2754053126985063e-05,
+      "loss": 1.123,
+      "step": 4403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.274466173541186e-05,
+      "loss": 1.0371,
+      "step": 4404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2735270664706134e-05,
+      "loss": 1.0658,
+      "step": 4405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2725879916203956e-05,
+      "loss": 1.0726,
+      "step": 4406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2716489491241354e-05,
+      "loss": 0.8837,
+      "step": 4407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.270709939115431e-05,
+      "loss": 0.9315,
+      "step": 4408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2697709617278746e-05,
+      "loss": 1.1299,
+      "step": 4409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.268832017095055e-05,
+      "loss": 1.0243,
+      "step": 4410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2678931053505566e-05,
+      "loss": 1.22,
+      "step": 4411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2669542266279574e-05,
+      "loss": 1.2964,
+      "step": 4412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.266015381060833e-05,
+      "loss": 0.9294,
+      "step": 4413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2650765687827526e-05,
+      "loss": 0.9106,
+      "step": 4414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2641377899272824e-05,
+      "loss": 1.2039,
+      "step": 4415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.263199044627981e-05,
+      "loss": 0.8606,
+      "step": 4416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.262260333018403e-05,
+      "loss": 1.0071,
+      "step": 4417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.261321655232103e-05,
+      "loss": 1.057,
+      "step": 4418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2603830114026238e-05,
+      "loss": 0.8917,
+      "step": 4419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2594444016635072e-05,
+      "loss": 1.1676,
+      "step": 4420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2585058261482896e-05,
+      "loss": 0.902,
+      "step": 4421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.257567284990503e-05,
+      "loss": 1.1292,
+      "step": 4422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2566287783236725e-05,
+      "loss": 0.6831,
+      "step": 4423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.25569030628132e-05,
+      "loss": 1.1583,
+      "step": 4424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.254751868996963e-05,
+      "loss": 1.3113,
+      "step": 4425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.253813466604113e-05,
+      "loss": 1.0605,
+      "step": 4426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.252875099236275e-05,
+      "loss": 1.0363,
+      "step": 4427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2519367670269527e-05,
+      "loss": 1.3359,
+      "step": 4428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2509984701096424e-05,
+      "loss": 0.8716,
+      "step": 4429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2500602086178347e-05,
+      "loss": 1.282,
+      "step": 4430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2491219826850176e-05,
+      "loss": 1.142,
+      "step": 4431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.248183792444673e-05,
+      "loss": 1.2954,
+      "step": 4432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2472456380302762e-05,
+      "loss": 1.0814,
+      "step": 4433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.246307519575299e-05,
+      "loss": 0.8905,
+      "step": 4434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2453694372132082e-05,
+      "loss": 1.0427,
+      "step": 4435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.244431391077465e-05,
+      "loss": 0.9725,
+      "step": 4436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2434933813015245e-05,
+      "loss": 1.2279,
+      "step": 4437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2425554080188384e-05,
+      "loss": 1.1381,
+      "step": 4438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.241617471362852e-05,
+      "loss": 0.88,
+      "step": 4439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2406795714670063e-05,
+      "loss": 1.345,
+      "step": 4440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.239741708464736e-05,
+      "loss": 1.1371,
+      "step": 4441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2388038824894708e-05,
+      "loss": 0.9735,
+      "step": 4442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.237866093674636e-05,
+      "loss": 1.2211,
+      "step": 4443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2369283421536503e-05,
+      "loss": 1.0537,
+      "step": 4444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2359906280599287e-05,
+      "loss": 0.8673,
+      "step": 4445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2350529515268802e-05,
+      "loss": 1.1062,
+      "step": 4446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.234115312687907e-05,
+      "loss": 1.2935,
+      "step": 4447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2331777116764087e-05,
+      "loss": 1.0338,
+      "step": 4448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2322401486257764e-05,
+      "loss": 1.0815,
+      "step": 4449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.231302623669399e-05,
+      "loss": 0.8746,
+      "step": 4450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.230365136940657e-05,
+      "loss": 1.301,
+      "step": 4451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2294276885729277e-05,
+      "loss": 0.8799,
+      "step": 4452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2284902786995814e-05,
+      "loss": 1.2078,
+      "step": 4453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2275529074539848e-05,
+      "loss": 0.7942,
+      "step": 4454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2266155749694967e-05,
+      "loss": 1.1947,
+      "step": 4455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.2256782813794718e-05,
+      "loss": 1.0564,
+      "step": 4456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2247410268172593e-05,
+      "loss": 1.1538,
+      "step": 4457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2238038114162023e-05,
+      "loss": 0.7292,
+      "step": 4458
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2228666353096397e-05,
+      "loss": 0.7489,
+      "step": 4459
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2219294986309033e-05,
+      "loss": 1.2328,
+      "step": 4460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.220992401513319e-05,
+      "loss": 1.265,
+      "step": 4461
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2200553440902083e-05,
+      "loss": 1.0458,
+      "step": 4462
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2191183264948865e-05,
+      "loss": 1.0621,
+      "step": 4463
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2181813488606643e-05,
+      "loss": 1.2129,
+      "step": 4464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2172444113208442e-05,
+      "loss": 0.9777,
+      "step": 4465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2163075140087248e-05,
+      "loss": 1.0208,
+      "step": 4466
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2153706570575987e-05,
+      "loss": 1.1398,
+      "step": 4467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.214433840600754e-05,
+      "loss": 0.9788,
+      "step": 4468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2134970647714706e-05,
+      "loss": 1.1947,
+      "step": 4469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2125603297030238e-05,
+      "loss": 0.9043,
+      "step": 4470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2116236355286826e-05,
+      "loss": 0.9973,
+      "step": 4471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.210686982381712e-05,
+      "loss": 0.8636,
+      "step": 4472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2097503703953695e-05,
+      "loss": 1.3833,
+      "step": 4473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2088137997029073e-05,
+      "loss": 1.1495,
+      "step": 4474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.207877270437571e-05,
+      "loss": 1.3628,
+      "step": 4475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2069407827326002e-05,
+      "loss": 1.1707,
+      "step": 4476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2060043367212306e-05,
+      "loss": 0.7323,
+      "step": 4477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.20506793253669e-05,
+      "loss": 1.0433,
+      "step": 4478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2041315703122005e-05,
+      "loss": 1.0745,
+      "step": 4479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.203195250180979e-05,
+      "loss": 1.1704,
+      "step": 4480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2022589722762353e-05,
+      "loss": 1.0317,
+      "step": 4481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.2013227367311753e-05,
+      "loss": 1.1066,
+      "step": 4482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.200386543678996e-05,
+      "loss": 1.0413,
+      "step": 4483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1994503932528905e-05,
+      "loss": 1.0695,
+      "step": 4484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.198514285586044e-05,
+      "loss": 0.963,
+      "step": 4485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1975782208116383e-05,
+      "loss": 0.9156,
+      "step": 4486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1966421990628472e-05,
+      "loss": 0.8339,
+      "step": 4487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1957062204728388e-05,
+      "loss": 1.1335,
+      "step": 4488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1947702851747738e-05,
+      "loss": 1.1921,
+      "step": 4489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1938343933018095e-05,
+      "loss": 1.1498,
+      "step": 4490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1928985449870943e-05,
+      "loss": 1.2083,
+      "step": 4491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1919627403637728e-05,
+      "loss": 0.8204,
+      "step": 4492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1910269795649802e-05,
+      "loss": 1.0548,
+      "step": 4493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.190091262723849e-05,
+      "loss": 1.1232,
+      "step": 4494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.189155589973503e-05,
+      "loss": 1.1727,
+      "step": 4495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1882199614470618e-05,
+      "loss": 1.1263,
+      "step": 4496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1872843772776357e-05,
+      "loss": 1.1324,
+      "step": 4497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1863488375983308e-05,
+      "loss": 1.5618,
+      "step": 4498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1854133425422486e-05,
+      "loss": 1.0471,
+      "step": 4499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1844778922424796e-05,
+      "loss": 0.9675,
+      "step": 4500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.183542486832112e-05,
+      "loss": 0.8038,
+      "step": 4501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.182607126444226e-05,
+      "loss": 0.9714,
+      "step": 4502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1816718112118956e-05,
+      "loss": 1.176,
+      "step": 4503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1807365412681877e-05,
+      "loss": 1.2814,
+      "step": 4504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1798013167461635e-05,
+      "loss": 1.1995,
+      "step": 4505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1788661377788784e-05,
+      "loss": 0.9828,
+      "step": 4506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1779310044993788e-05,
+      "loss": 1.0945,
+      "step": 4507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1769959170407082e-05,
+      "loss": 0.9121,
+      "step": 4508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1760608755359003e-05,
+      "loss": 0.9197,
+      "step": 4509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1751258801179853e-05,
+      "loss": 1.2498,
+      "step": 4510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1741909309199834e-05,
+      "loss": 0.9463,
+      "step": 4511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1732560280749097e-05,
+      "loss": 0.8949,
+      "step": 4512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.172321171715776e-05,
+      "loss": 1.1105,
+      "step": 4513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.171386361975582e-05,
+      "loss": 1.0914,
+      "step": 4514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.170451598987324e-05,
+      "loss": 1.3027,
+      "step": 4515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.169516882883991e-05,
+      "loss": 0.9709,
+      "step": 4516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1685822137985657e-05,
+      "loss": 0.9685,
+      "step": 4517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1676475918640225e-05,
+      "loss": 0.9353,
+      "step": 4518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.166713017213331e-05,
+      "loss": 0.9033,
+      "step": 4519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1657784899794532e-05,
+      "loss": 0.9775,
+      "step": 4520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.164844010295345e-05,
+      "loss": 1.1429,
+      "step": 4521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1639095782939544e-05,
+      "loss": 1.26,
+      "step": 4522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.162975194108223e-05,
+      "loss": 1.2466,
+      "step": 4523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1620408578710873e-05,
+      "loss": 1.0536,
+      "step": 4524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.161106569715474e-05,
+      "loss": 0.9515,
+      "step": 4525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.160172329774304e-05,
+      "loss": 0.9501,
+      "step": 4526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1592381381804944e-05,
+      "loss": 1.1273,
+      "step": 4527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1583039950669505e-05,
+      "loss": 1.2029,
+      "step": 4528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.157369900566574e-05,
+      "loss": 1.0002,
+      "step": 4529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1564358548122587e-05,
+      "loss": 0.7924,
+      "step": 4530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.155501857936892e-05,
+      "loss": 1.1555,
+      "step": 4531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1545679100733524e-05,
+      "loss": 0.9413,
+      "step": 4532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1536340113545138e-05,
+      "loss": 1.0177,
+      "step": 4533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.152700161913242e-05,
+      "loss": 0.8101,
+      "step": 4534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1517663618823966e-05,
+      "loss": 1.244,
+      "step": 4535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1508326113948286e-05,
+      "loss": 0.8328,
+      "step": 4536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.149898910583383e-05,
+      "loss": 1.0471,
+      "step": 4537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.1489652595808983e-05,
+      "loss": 0.7494,
+      "step": 4538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.148031658520204e-05,
+      "loss": 0.9478,
+      "step": 4539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1470981075341253e-05,
+      "loss": 1.0045,
+      "step": 4540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1461646067554785e-05,
+      "loss": 0.9775,
+      "step": 4541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1452311563170717e-05,
+      "loss": 1.1836,
+      "step": 4542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.144297756351708e-05,
+      "loss": 1.0436,
+      "step": 4543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1433644069921822e-05,
+      "loss": 0.9739,
+      "step": 4544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.142431108371283e-05,
+      "loss": 1.11,
+      "step": 4545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1414978606217893e-05,
+      "loss": 1.1669,
+      "step": 4546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1405646638764754e-05,
+      "loss": 0.9243,
+      "step": 4547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1396315182681074e-05,
+      "loss": 1.0952,
+      "step": 4548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.138698423929445e-05,
+      "loss": 1.0901,
+      "step": 4549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.137765380993238e-05,
+      "loss": 0.9191,
+      "step": 4550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1368323895922316e-05,
+      "loss": 0.9247,
+      "step": 4551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1358994498591623e-05,
+      "loss": 1.1316,
+      "step": 4552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1349665619267594e-05,
+      "loss": 1.0894,
+      "step": 4553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1340337259277464e-05,
+      "loss": 0.9077,
+      "step": 4554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.133100941994838e-05,
+      "loss": 1.1875,
+      "step": 4555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.13216821026074e-05,
+      "loss": 1.1765,
+      "step": 4556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1312355308581533e-05,
+      "loss": 1.095,
+      "step": 4557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1303029039197702e-05,
+      "loss": 1.2386,
+      "step": 4558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1293703295782767e-05,
+      "loss": 1.1466,
+      "step": 4559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1284378079663485e-05,
+      "loss": 1.178,
+      "step": 4560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1275053392166572e-05,
+      "loss": 1.2092,
+      "step": 4561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1265729234618644e-05,
+      "loss": 1.0604,
+      "step": 4562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1256405608346265e-05,
+      "loss": 1.2383,
+      "step": 4563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.124708251467589e-05,
+      "loss": 1.2394,
+      "step": 4564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1237759954933926e-05,
+      "loss": 0.9951,
+      "step": 4565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1228437930446698e-05,
+      "loss": 1.3236,
+      "step": 4566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1219116442540453e-05,
+      "loss": 1.2313,
+      "step": 4567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.120979549254136e-05,
+      "loss": 1.292,
+      "step": 4568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1200475081775518e-05,
+      "loss": 0.9216,
+      "step": 4569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.119115521156893e-05,
+      "loss": 0.9789,
+      "step": 4570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1181835883247546e-05,
+      "loss": 1.0499,
+      "step": 4571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1172517098137228e-05,
+      "loss": 1.2268,
+      "step": 4572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1163198857563768e-05,
+      "loss": 0.9183,
+      "step": 4573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.115388116285286e-05,
+      "loss": 0.9452,
+      "step": 4574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.114456401533014e-05,
+      "loss": 0.8796,
+      "step": 4575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1135247416321165e-05,
+      "loss": 1.1482,
+      "step": 4576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1125931367151407e-05,
+      "loss": 1.1782,
+      "step": 4577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.111661586914626e-05,
+      "loss": 1.0217,
+      "step": 4578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1107300923631045e-05,
+      "loss": 1.0446,
+      "step": 4579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1097986531930997e-05,
+      "loss": 1.1732,
+      "step": 4580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.108867269537128e-05,
+      "loss": 0.9133,
+      "step": 4581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.107935941527698e-05,
+      "loss": 1.4253,
+      "step": 4582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.10700466929731e-05,
+      "loss": 1.0878,
+      "step": 4583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1060734529784552e-05,
+      "loss": 0.9193,
+      "step": 4584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1051422927036185e-05,
+      "loss": 1.0007,
+      "step": 4585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.1042111886052767e-05,
+      "loss": 1.2964,
+      "step": 4586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.103280140815898e-05,
+      "loss": 1.0694,
+      "step": 4587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.102349149467942e-05,
+      "loss": 0.7375,
+      "step": 4588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.101418214693862e-05,
+      "loss": 1.0314,
+      "step": 4589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.100487336626102e-05,
+      "loss": 1.197,
+      "step": 4590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0995565153970986e-05,
+      "loss": 1.0387,
+      "step": 4591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.098625751139279e-05,
+      "loss": 1.1495,
+      "step": 4592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.097695043985064e-05,
+      "loss": 1.2437,
+      "step": 4593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0967643940668645e-05,
+      "loss": 1.1362,
+      "step": 4594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0958338015170858e-05,
+      "loss": 1.1036,
+      "step": 4595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0949032664681228e-05,
+      "loss": 1.0663,
+      "step": 4596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0939727890523633e-05,
+      "loss": 0.9512,
+      "step": 4597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0930423694021858e-05,
+      "loss": 1.2489,
+      "step": 4598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0921120076499618e-05,
+      "loss": 0.896,
+      "step": 4599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0911817039280535e-05,
+      "loss": 1.0348,
+      "step": 4600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0902514583688168e-05,
+      "loss": 0.8151,
+      "step": 4601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0893212711045963e-05,
+      "loss": 1.0493,
+      "step": 4602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0883911422677306e-05,
+      "loss": 0.9624,
+      "step": 4603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0874610719905495e-05,
+      "loss": 1.1251,
+      "step": 4604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.086531060405375e-05,
+      "loss": 1.1812,
+      "step": 4605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0856011076445183e-05,
+      "loss": 1.2269,
+      "step": 4606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0846712138402837e-05,
+      "loss": 0.986,
+      "step": 4607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.083741379124971e-05,
+      "loss": 1.0292,
+      "step": 4608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0828116036308642e-05,
+      "loss": 1.1486,
+      "step": 4609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0818818874902446e-05,
+      "loss": 1.1597,
+      "step": 4610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0809522308353828e-05,
+      "loss": 1.0614,
+      "step": 4611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0800226337985413e-05,
+      "loss": 1.0349,
+      "step": 4612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0790930965119735e-05,
+      "loss": 1.1243,
+      "step": 4613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.078163619107925e-05,
+      "loss": 1.0332,
+      "step": 4614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.077234201718633e-05,
+      "loss": 1.1428,
+      "step": 4615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.076304844476327e-05,
+      "loss": 1.0426,
+      "step": 4616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.075375547513225e-05,
+      "loss": 1.1976,
+      "step": 4617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.074446310961539e-05,
+      "loss": 1.2067,
+      "step": 4618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0735171349534717e-05,
+      "loss": 1.2264,
+      "step": 4619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.072588019621218e-05,
+      "loss": 1.088,
+      "step": 4620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0716589650969608e-05,
+      "loss": 1.3242,
+      "step": 4621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.0707299715128804e-05,
+      "loss": 1.1624,
+      "step": 4622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0698010390011423e-05,
+      "loss": 1.1226,
+      "step": 4623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0688721676939073e-05,
+      "loss": 0.7993,
+      "step": 4624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.067943357723326e-05,
+      "loss": 1.4851,
+      "step": 4625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.06701460922154e-05,
+      "loss": 1.5027,
+      "step": 4626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0660859223206824e-05,
+      "loss": 0.7646,
+      "step": 4627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0651572971528776e-05,
+      "loss": 1.2681,
+      "step": 4628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.064228733850242e-05,
+      "loss": 1.0024,
+      "step": 4629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0633002325448824e-05,
+      "loss": 1.1025,
+      "step": 4630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0623717933688965e-05,
+      "loss": 1.0739,
+      "step": 4631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0614434164543736e-05,
+      "loss": 1.0973,
+      "step": 4632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.060515101933394e-05,
+      "loss": 1.1023,
+      "step": 4633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0595868499380298e-05,
+      "loss": 0.8729,
+      "step": 4634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0586586606003422e-05,
+      "loss": 0.8535,
+      "step": 4635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.057730534052387e-05,
+      "loss": 1.1021,
+      "step": 4636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0568024704262077e-05,
+      "loss": 0.9785,
+      "step": 4637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0558744698538402e-05,
+      "loss": 0.8149,
+      "step": 4638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0549465324673114e-05,
+      "loss": 0.9301,
+      "step": 4639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.05401865839864e-05,
+      "loss": 1.0244,
+      "step": 4640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.053090847779833e-05,
+      "loss": 1.0267,
+      "step": 4641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0521631007428918e-05,
+      "loss": 1.1895,
+      "step": 4642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0512354174198066e-05,
+      "loss": 1.0333,
+      "step": 4643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0503077979425597e-05,
+      "loss": 1.2377,
+      "step": 4644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0493802424431228e-05,
+      "loss": 1.0228,
+      "step": 4645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0484527510534595e-05,
+      "loss": 1.1475,
+      "step": 4646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0475253239055247e-05,
+      "loss": 1.0182,
+      "step": 4647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0465979611312632e-05,
+      "loss": 1.0038,
+      "step": 4648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0456706628626122e-05,
+      "loss": 1.2496,
+      "step": 4649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.044743429231498e-05,
+      "loss": 0.8834,
+      "step": 4650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0438162603698376e-05,
+      "loss": 1.2223,
+      "step": 4651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0428891564095404e-05,
+      "loss": 1.2417,
+      "step": 4652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0419621174825053e-05,
+      "loss": 1.108,
+      "step": 4653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0410351437206233e-05,
+      "loss": 0.8918,
+      "step": 4654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0401082352557736e-05,
+      "loss": 1.3031,
+      "step": 4655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0391813922198284e-05,
+      "loss": 1.0486,
+      "step": 4656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0382546147446497e-05,
+      "loss": 1.0498,
+      "step": 4657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.037327902962091e-05,
+      "loss": 0.9825,
+      "step": 4658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.036401257003995e-05,
+      "loss": 1.2141,
+      "step": 4659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.035474677002196e-05,
+      "loss": 1.0717,
+      "step": 4660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.034548163088519e-05,
+      "loss": 0.8719,
+      "step": 4661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0336217153947788e-05,
+      "loss": 0.8802,
+      "step": 4662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.032695334052782e-05,
+      "loss": 1.2178,
+      "step": 4663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0317690191943252e-05,
+      "loss": 1.0155,
+      "step": 4664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0308427709511948e-05,
+      "loss": 0.9348,
+      "step": 4665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0299165894551687e-05,
+      "loss": 1.395,
+      "step": 4666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0289904748380146e-05,
+      "loss": 1.051,
+      "step": 4667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.028064427231492e-05,
+      "loss": 1.0701,
+      "step": 4668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0271384467673484e-05,
+      "loss": 1.0492,
+      "step": 4669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.026212533577324e-05,
+      "loss": 1.0137,
+      "step": 4670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.025286687793149e-05,
+      "loss": 1.027,
+      "step": 4671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0243609095465436e-05,
+      "loss": 1.1206,
+      "step": 4672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.023435198969218e-05,
+      "loss": 1.0623,
+      "step": 4673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0225095561928735e-05,
+      "loss": 1.0558,
+      "step": 4674
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0215839813492012e-05,
+      "loss": 1.35,
+      "step": 4675
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.020658474569883e-05,
+      "loss": 0.8989,
+      "step": 4676
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0197330359865914e-05,
+      "loss": 1.2115,
+      "step": 4677
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.018807665730989e-05,
+      "loss": 1.1071,
+      "step": 4678
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.017882363934728e-05,
+      "loss": 1.1891,
+      "step": 4679
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0169571307294505e-05,
+      "loss": 0.7744,
+      "step": 4680
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0160319662467903e-05,
+      "loss": 1.0698,
+      "step": 4681
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0151068706183718e-05,
+      "loss": 1.4342,
+      "step": 4682
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0141818439758063e-05,
+      "loss": 0.8307,
+      "step": 4683
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0132568864506997e-05,
+      "loss": 0.8927,
+      "step": 4684
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0123319981746446e-05,
+      "loss": 0.7632,
+      "step": 4685
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.011407179279226e-05,
+      "loss": 0.8348,
+      "step": 4686
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0104824298960167e-05,
+      "loss": 0.8171,
+      "step": 4687
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0095577501565825e-05,
+      "loss": 0.9352,
+      "step": 4688
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0086331401924763e-05,
+      "loss": 1.1146,
+      "step": 4689
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0077086001352437e-05,
+      "loss": 1.1902,
+      "step": 4690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0067841301164194e-05,
+      "loss": 1.0225,
+      "step": 4691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.005859730267528e-05,
+      "loss": 0.63,
+      "step": 4692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.004935400720083e-05,
+      "loss": 1.2166,
+      "step": 4693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0040111416055897e-05,
+      "loss": 0.7346,
+      "step": 4694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.003086953055542e-05,
+      "loss": 1.004,
+      "step": 4695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.002162835201426e-05,
+      "loss": 0.8862,
+      "step": 4696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0012387881747145e-05,
+      "loss": 1.0192,
+      "step": 4697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.0003148121068723e-05,
+      "loss": 0.9941,
+      "step": 4698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9993909071293536e-05,
+      "loss": 1.2034,
+      "step": 4699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9984670733736036e-05,
+      "loss": 0.9317,
+      "step": 4700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9975433109710552e-05,
+      "loss": 1.1641,
+      "step": 4701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9966196200531324e-05,
+      "loss": 0.9916,
+      "step": 4702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9956960007512484e-05,
+      "loss": 1.1497,
+      "step": 4703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9947724531968083e-05,
+      "loss": 1.0205,
+      "step": 4704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9938489775212043e-05,
+      "loss": 0.8961,
+      "step": 4705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.99292557385582e-05,
+      "loss": 1.1057,
+      "step": 4706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9920022423320286e-05,
+      "loss": 0.9967,
+      "step": 4707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9910789830811916e-05,
+      "loss": 1.2727,
+      "step": 4708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.990155796234662e-05,
+      "loss": 0.7892,
+      "step": 4709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9892326819237812e-05,
+      "loss": 1.1412,
+      "step": 4710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.988309640279882e-05,
+      "loss": 1.0497,
+      "step": 4711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9873866714342844e-05,
+      "loss": 1.2675,
+      "step": 4712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9864637755182996e-05,
+      "loss": 0.9365,
+      "step": 4713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.985540952663229e-05,
+      "loss": 1.0417,
+      "step": 4714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9846182030003632e-05,
+      "loss": 1.0989,
+      "step": 4715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9836955266609794e-05,
+      "loss": 0.981,
+      "step": 4716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9827729237763503e-05,
+      "loss": 1.2134,
+      "step": 4717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9818503944777332e-05,
+      "loss": 0.9834,
+      "step": 4718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9809279388963763e-05,
+      "loss": 1.2942,
+      "step": 4719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.980005557163518e-05,
+      "loss": 0.9777,
+      "step": 4720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9790832494103864e-05,
+      "loss": 1.2368,
+      "step": 4721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.978161015768197e-05,
+      "loss": 0.9491,
+      "step": 4722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.977238856368157e-05,
+      "loss": 1.012,
+      "step": 4723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9763167713414622e-05,
+      "loss": 1.0404,
+      "step": 4724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9753947608192983e-05,
+      "loss": 0.9979,
+      "step": 4725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9744728249328387e-05,
+      "loss": 1.0604,
+      "step": 4726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9735509638132483e-05,
+      "loss": 0.8573,
+      "step": 4727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9726291775916806e-05,
+      "loss": 1.0264,
+      "step": 4728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9717074663992783e-05,
+      "loss": 1.0677,
+      "step": 4729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9707858303671713e-05,
+      "loss": 0.7637,
+      "step": 4730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9698642696264853e-05,
+      "loss": 1.3932,
+      "step": 4731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9689427843083272e-05,
+      "loss": 0.9474,
+      "step": 4732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9680213745437987e-05,
+      "loss": 0.8484,
+      "step": 4733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9671000404639882e-05,
+      "loss": 1.1987,
+      "step": 4734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9661787821999756e-05,
+      "loss": 1.1066,
+      "step": 4735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9652575998828262e-05,
+      "loss": 1.2283,
+      "step": 4736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9643364936435984e-05,
+      "loss": 1.1487,
+      "step": 4737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9634154636133375e-05,
+      "loss": 1.0892,
+      "step": 4738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9624945099230796e-05,
+      "loss": 1.1014,
+      "step": 4739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9615736327038477e-05,
+      "loss": 1.1688,
+      "step": 4740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.960652832086656e-05,
+      "loss": 1.0325,
+      "step": 4741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9597321082025065e-05,
+      "loss": 1.0064,
+      "step": 4742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.958811461182391e-05,
+      "loss": 1.0685,
+      "step": 4743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.957890891157291e-05,
+      "loss": 1.101,
+      "step": 4744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9569703982581754e-05,
+      "loss": 1.0398,
+      "step": 4745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9560499826160027e-05,
+      "loss": 1.1216,
+      "step": 4746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.955129644361721e-05,
+      "loss": 0.9169,
+      "step": 4747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.954209383626267e-05,
+      "loss": 1.1687,
+      "step": 4748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9532892005405668e-05,
+      "loss": 0.7224,
+      "step": 4749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.952369095235534e-05,
+      "loss": 1.2413,
+      "step": 4750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9514490678420733e-05,
+      "loss": 1.0569,
+      "step": 4751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.950529118491076e-05,
+      "loss": 1.3538,
+      "step": 4752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9496092473134257e-05,
+      "loss": 1.0059,
+      "step": 4753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.94868945443999e-05,
+      "loss": 0.8438,
+      "step": 4754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9477697400016297e-05,
+      "loss": 0.9928,
+      "step": 4755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.946850104129192e-05,
+      "loss": 1.1599,
+      "step": 4756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9459305469535136e-05,
+      "loss": 0.9717,
+      "step": 4757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.945011068605421e-05,
+      "loss": 0.9432,
+      "step": 4758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9440916692157285e-05,
+      "loss": 1.1244,
+      "step": 4759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9431723489152388e-05,
+      "loss": 1.1857,
+      "step": 4760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9422531078347438e-05,
+      "loss": 1.082,
+      "step": 4761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.941333946105024e-05,
+      "loss": 1.1528,
+      "step": 4762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9404148638568494e-05,
+      "loss": 0.9917,
+      "step": 4763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9394958612209775e-05,
+      "loss": 0.9608,
+      "step": 4764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9385769383281547e-05,
+      "loss": 1.203,
+      "step": 4765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.937658095309117e-05,
+      "loss": 0.9111,
+      "step": 4766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9367393322945888e-05,
+      "loss": 0.9257,
+      "step": 4767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9358206494152815e-05,
+      "loss": 1.1289,
+      "step": 4768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9349020468018967e-05,
+      "loss": 1.1,
+      "step": 4769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9339835245851245e-05,
+      "loss": 1.1124,
+      "step": 4770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9330650828956428e-05,
+      "loss": 0.9072,
+      "step": 4771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9321467218641194e-05,
+      "loss": 0.6929,
+      "step": 4772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.93122844162121e-05,
+      "loss": 1.6658,
+      "step": 4773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9303102422975566e-05,
+      "loss": 0.9398,
+      "step": 4774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9293921240237933e-05,
+      "loss": 1.1017,
+      "step": 4775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9284740869305405e-05,
+      "loss": 0.9264,
+      "step": 4776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9275561311484076e-05,
+      "loss": 0.9514,
+      "step": 4777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9266382568079922e-05,
+      "loss": 0.8643,
+      "step": 4778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9257204640398806e-05,
+      "loss": 1.2646,
+      "step": 4779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.924802752974647e-05,
+      "loss": 1.2584,
+      "step": 4780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9238851237428555e-05,
+      "loss": 1.2125,
+      "step": 4781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.922967576475056e-05,
+      "loss": 0.9559,
+      "step": 4782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9220501113017888e-05,
+      "loss": 1.2827,
+      "step": 4783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9211327283535816e-05,
+      "loss": 0.9758,
+      "step": 4784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9202154277609514e-05,
+      "loss": 0.8427,
+      "step": 4785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.919298209654402e-05,
+      "loss": 1.1794,
+      "step": 4786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9183810741644273e-05,
+      "loss": 0.8201,
+      "step": 4787
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.917464021421507e-05,
+      "loss": 1.0712,
+      "step": 4788
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.916547051556111e-05,
+      "loss": 0.7563,
+      "step": 4789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9156301646986968e-05,
+      "loss": 0.8903,
+      "step": 4790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9147133609797107e-05,
+      "loss": 0.8151,
+      "step": 4791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9137966405295855e-05,
+      "loss": 1.1863,
+      "step": 4792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.912880003478744e-05,
+      "loss": 0.9327,
+      "step": 4793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9119634499575955e-05,
+      "loss": 1.1348,
+      "step": 4794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9110469800965398e-05,
+      "loss": 0.8583,
+      "step": 4795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9101305940259616e-05,
+      "loss": 1.084,
+      "step": 4796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9092142918762358e-05,
+      "loss": 1.0697,
+      "step": 4797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.908298073777725e-05,
+      "loss": 1.1666,
+      "step": 4798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9073819398607805e-05,
+      "loss": 1.0671,
+      "step": 4799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9064658902557403e-05,
+      "loss": 0.8923,
+      "step": 4800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.905549925092931e-05,
+      "loss": 1.0554,
+      "step": 4801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9046340445026672e-05,
+      "loss": 1.1142,
+      "step": 4802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9037182486152516e-05,
+      "loss": 0.8335,
+      "step": 4803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.902802537560974e-05,
+      "loss": 0.8301,
+      "step": 4804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9018869114701136e-05,
+      "loss": 1.1007,
+      "step": 4805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9009713704729367e-05,
+      "loss": 1.4226,
+      "step": 4806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9000559146996968e-05,
+      "loss": 1.0944,
+      "step": 4807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8991405442806364e-05,
+      "loss": 1.0675,
+      "step": 4808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8982252593459863e-05,
+      "loss": 1.0096,
+      "step": 4809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.897310060025963e-05,
+      "loss": 0.9094,
+      "step": 4810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8963949464507712e-05,
+      "loss": 1.0648,
+      "step": 4811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8954799187506075e-05,
+      "loss": 1.0756,
+      "step": 4812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.894564977055651e-05,
+      "loss": 1.0662,
+      "step": 4813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8936501214960706e-05,
+      "loss": 1.0461,
+      "step": 4814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8927353522020235e-05,
+      "loss": 1.1439,
+      "step": 4815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8918206693036548e-05,
+      "loss": 1.0342,
+      "step": 4816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8909060729310952e-05,
+      "loss": 1.0593,
+      "step": 4817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.889991563214465e-05,
+      "loss": 1.1973,
+      "step": 4818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8890771402838722e-05,
+      "loss": 0.8007,
+      "step": 4819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.888162804269412e-05,
+      "loss": 0.9733,
+      "step": 4820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.887248555301166e-05,
+      "loss": 0.9985,
+      "step": 4821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.886334393509206e-05,
+      "loss": 0.8914,
+      "step": 4822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8854203190235893e-05,
+      "loss": 1.166,
+      "step": 4823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8845063319743624e-05,
+      "loss": 1.264,
+      "step": 4824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8835924324915556e-05,
+      "loss": 0.856,
+      "step": 4825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8826786207051934e-05,
+      "loss": 0.8768,
+      "step": 4826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8817648967452818e-05,
+      "loss": 0.7727,
+      "step": 4827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8808512607418173e-05,
+      "loss": 0.9916,
+      "step": 4828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.879937712824782e-05,
+      "loss": 0.8462,
+      "step": 4829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8790242531241487e-05,
+      "loss": 0.8467,
+      "step": 4830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.878110881769873e-05,
+      "loss": 1.1755,
+      "step": 4831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.877197598891902e-05,
+      "loss": 1.0945,
+      "step": 4832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8762844046201683e-05,
+      "loss": 1.0769,
+      "step": 4833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8753712990845924e-05,
+      "loss": 0.8943,
+      "step": 4834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8744582824150817e-05,
+      "loss": 0.9235,
+      "step": 4835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8735453547415312e-05,
+      "loss": 0.9211,
+      "step": 4836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8726325161938235e-05,
+      "loss": 0.9481,
+      "step": 4837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8717197669018293e-05,
+      "loss": 1.0303,
+      "step": 4838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8708071069954027e-05,
+      "loss": 0.9053,
+      "step": 4839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.869894536604392e-05,
+      "loss": 1.0507,
+      "step": 4840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.868982055858626e-05,
+      "loss": 1.3274,
+      "step": 4841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.868069664887925e-05,
+      "loss": 1.1206,
+      "step": 4842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.867157363822094e-05,
+      "loss": 1.0551,
+      "step": 4843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866245152790928e-05,
+      "loss": 1.1271,
+      "step": 4844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653330319242047e-05,
+      "loss": 1.0461,
+      "step": 4845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864421001351694e-05,
+      "loss": 1.0066,
+      "step": 4846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8635090612031496e-05,
+      "loss": 0.9922,
+      "step": 4847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8625972116083146e-05,
+      "loss": 1.2715,
+      "step": 4848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8616854526969167e-05,
+      "loss": 1.2225,
+      "step": 4849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8607737845986722e-05,
+      "loss": 1.0751,
+      "step": 4850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.859862207443285e-05,
+      "loss": 1.0657,
+      "step": 4851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8589507213604445e-05,
+      "loss": 0.7058,
+      "step": 4852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.858039326479829e-05,
+      "loss": 1.0627,
+      "step": 4853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8571280229311034e-05,
+      "loss": 0.734,
+      "step": 4854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8562168108439176e-05,
+      "loss": 0.8276,
+      "step": 4855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8553056903479104e-05,
+      "loss": 0.9316,
+      "step": 4856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.854394661572707e-05,
+      "loss": 1.0692,
+      "step": 4857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8534837246479207e-05,
+      "loss": 0.8521,
+      "step": 4858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8525728797031492e-05,
+      "loss": 1.2822,
+      "step": 4859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8516621268679792e-05,
+      "loss": 0.6705,
+      "step": 4860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.850751466271984e-05,
+      "loss": 0.7603,
+      "step": 4861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8498408980447234e-05,
+      "loss": 1.036,
+      "step": 4862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8489304223157437e-05,
+      "loss": 0.7891,
+      "step": 4863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.848020039214579e-05,
+      "loss": 1.1355,
+      "step": 4864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.84710974887075e-05,
+      "loss": 0.996,
+      "step": 4865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.846199551413762e-05,
+      "loss": 0.9885,
+      "step": 4866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8452894469731114e-05,
+      "loss": 0.9513,
+      "step": 4867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8443794356782784e-05,
+      "loss": 1.0826,
+      "step": 4868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8434695176587296e-05,
+      "loss": 0.9792,
+      "step": 4869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8425596930439202e-05,
+      "loss": 1.0336,
+      "step": 4870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.84164996196329e-05,
+      "loss": 0.9086,
+      "step": 4871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8407403245462685e-05,
+      "loss": 1.1215,
+      "step": 4872
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8398307809222683e-05,
+      "loss": 0.8257,
+      "step": 4873
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8389213312206904e-05,
+      "loss": 1.0278,
+      "step": 4874
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.838011975570923e-05,
+      "loss": 1.0602,
+      "step": 4875
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.837102714102341e-05,
+      "loss": 0.7745,
+      "step": 4876
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.836193546944304e-05,
+      "loss": 1.1698,
+      "step": 4877
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.83528447422616e-05,
+      "loss": 0.7454,
+      "step": 4878
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8343754960772426e-05,
+      "loss": 1.0237,
+      "step": 4879
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.833466612626873e-05,
+      "loss": 1.3406,
+      "step": 4880
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8325578240043577e-05,
+      "loss": 0.9814,
+      "step": 4881
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8316491303389914e-05,
+      "loss": 0.8636,
+      "step": 4882
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8307405317600523e-05,
+      "loss": 1.1147,
+      "step": 4883
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.829832028396808e-05,
+      "loss": 1.027,
+      "step": 4884
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.828923620378511e-05,
+      "loss": 1.0042,
+      "step": 4885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.828015307834402e-05,
+      "loss": 1.0945,
+      "step": 4886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.827107090893705e-05,
+      "loss": 1.0223,
+      "step": 4887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.826198969685633e-05,
+      "loss": 1.0066,
+      "step": 4888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.825290944339385e-05,
+      "loss": 1.3215,
+      "step": 4889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8243830149841462e-05,
+      "loss": 1.0529,
+      "step": 4890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8234751817490867e-05,
+      "loss": 0.769,
+      "step": 4891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.822567444763365e-05,
+      "loss": 0.8167,
+      "step": 4892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.821659804156124e-05,
+      "loss": 1.1125,
+      "step": 4893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8207522600564952e-05,
+      "loss": 1.1416,
+      "step": 4894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.819844812593595e-05,
+      "loss": 0.8896,
+      "step": 4895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.818937461896526e-05,
+      "loss": 1.0337,
+      "step": 4896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.818030208094377e-05,
+      "loss": 1.1929,
+      "step": 4897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.817123051316223e-05,
+      "loss": 0.9832,
+      "step": 4898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.816215991691126e-05,
+      "loss": 0.834,
+      "step": 4899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.815309029348133e-05,
+      "loss": 1.1003,
+      "step": 4900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8144021644162778e-05,
+      "loss": 0.9719,
+      "step": 4901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8134953970245805e-05,
+      "loss": 1.1866,
+      "step": 4902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.812588727302047e-05,
+      "loss": 1.0205,
+      "step": 4903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8116821553776697e-05,
+      "loss": 0.8151,
+      "step": 4904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.810775681380426e-05,
+      "loss": 1.0884,
+      "step": 4905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.809869305439281e-05,
+      "loss": 0.799,
+      "step": 4906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.808963027683184e-05,
+      "loss": 0.988,
+      "step": 4907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8080568482410725e-05,
+      "loss": 1.1257,
+      "step": 4908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8071507672418685e-05,
+      "loss": 0.8278,
+      "step": 4909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.80624478481448e-05,
+      "loss": 0.8424,
+      "step": 4910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8053389010878026e-05,
+      "loss": 0.8264,
+      "step": 4911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.804433116190714e-05,
+      "loss": 1.0928,
+      "step": 4912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8035274302520828e-05,
+      "loss": 0.92,
+      "step": 4913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.80262184340076e-05,
+      "loss": 1.1193,
+      "step": 4914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8017163557655843e-05,
+      "loss": 0.8425,
+      "step": 4915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8008109674753794e-05,
+      "loss": 0.8983,
+      "step": 4916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.799905678658954e-05,
+      "loss": 1.2822,
+      "step": 4917
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.799000489445105e-05,
+      "loss": 0.7948,
+      "step": 4918
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7980953999626144e-05,
+      "loss": 0.9042,
+      "step": 4919
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.797190410340247e-05,
+      "loss": 0.9606,
+      "step": 4920
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7962855207067593e-05,
+      "loss": 1.0886,
+      "step": 4921
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.795380731190887e-05,
+      "loss": 1.2894,
+      "step": 4922
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.794476041921357e-05,
+      "loss": 1.2415,
+      "step": 4923
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.793571453026878e-05,
+      "loss": 1.0699,
+      "step": 4924
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.792666964636148e-05,
+      "loss": 0.7443,
+      "step": 4925
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7917625768778468e-05,
+      "loss": 0.99,
+      "step": 4926
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7908582898806424e-05,
+      "loss": 0.8219,
+      "step": 4927
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7899541037731885e-05,
+      "loss": 0.8921,
+      "step": 4928
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.789050018684124e-05,
+      "loss": 0.8236,
+      "step": 4929
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.788146034742072e-05,
+      "loss": 1.1813,
+      "step": 4930
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.787242152075643e-05,
+      "loss": 0.9635,
+      "step": 4931
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7863383708134336e-05,
+      "loss": 1.4326,
+      "step": 4932
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7854346910840243e-05,
+      "loss": 0.7604,
+      "step": 4933
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.78453111301598e-05,
+      "loss": 0.9784,
+      "step": 4934
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7836276367378567e-05,
+      "loss": 0.938,
+      "step": 4935
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7827242623781892e-05,
+      "loss": 0.8726,
+      "step": 4936
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7818209900655018e-05,
+      "loss": 0.9187,
+      "step": 4937
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7809178199283036e-05,
+      "loss": 0.9788,
+      "step": 4938
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7800147520950888e-05,
+      "loss": 0.8772,
+      "step": 4939
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.779111786694336e-05,
+      "loss": 1.0345,
+      "step": 4940
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7782089238545112e-05,
+      "loss": 1.1907,
+      "step": 4941
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7773061637040643e-05,
+      "loss": 1.0033,
+      "step": 4942
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7764035063714324e-05,
+      "loss": 0.8455,
+      "step": 4943
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.775500951985035e-05,
+      "loss": 1.0193,
+      "step": 4944
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7745985006732802e-05,
+      "loss": 0.757,
+      "step": 4945
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.773696152564559e-05,
+      "loss": 1.002,
+      "step": 4946
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7727939077872496e-05,
+      "loss": 0.9011,
+      "step": 4947
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7718917664697125e-05,
+      "loss": 1.1416,
+      "step": 4948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7709897287402988e-05,
+      "loss": 0.8818,
+      "step": 4949
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7700877947273393e-05,
+      "loss": 1.0781,
+      "step": 4950
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7691859645591528e-05,
+      "loss": 1.1052,
+      "step": 4951
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.768284238364043e-05,
+      "loss": 1.1104,
+      "step": 4952
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7673826162702996e-05,
+      "loss": 0.9127,
+      "step": 4953
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7664810984061948e-05,
+      "loss": 1.1366,
+      "step": 4954
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7655796848999883e-05,
+      "loss": 1.0109,
+      "step": 4955
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7646783758799252e-05,
+      "loss": 1.3823,
+      "step": 4956
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7637771714742345e-05,
+      "loss": 0.9987,
+      "step": 4957
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7628760718111307e-05,
+      "loss": 0.9727,
+      "step": 4958
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7619750770188133e-05,
+      "loss": 0.6977,
+      "step": 4959
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7610741872254674e-05,
+      "loss": 0.7942,
+      "step": 4960
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.760173402559262e-05,
+      "loss": 0.9951,
+      "step": 4961
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7592727231483526e-05,
+      "loss": 0.8323,
+      "step": 4962
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.75837214912088e-05,
+      "loss": 0.9185,
+      "step": 4963
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7574716806049675e-05,
+      "loss": 1.1011,
+      "step": 4964
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.756571317728725e-05,
+      "loss": 0.8604,
+      "step": 4965
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7556710606202488e-05,
+      "loss": 1.2261,
+      "step": 4966
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7547709094076174e-05,
+      "loss": 0.9487,
+      "step": 4967
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7538708642188956e-05,
+      "loss": 0.9196,
+      "step": 4968
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7529709251821334e-05,
+      "loss": 1.0896,
+      "step": 4969
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7520710924253648e-05,
+      "loss": 1.5024,
+      "step": 4970
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7511713660766105e-05,
+      "loss": 0.7434,
+      "step": 4971
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7502717462638734e-05,
+      "loss": 1.0146,
+      "step": 4972
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.749372233115143e-05,
+      "loss": 0.8556,
+      "step": 4973
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7484728267583934e-05,
+      "loss": 0.8446,
+      "step": 4974
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7475735273215822e-05,
+      "loss": 0.9299,
+      "step": 4975
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.746674334932655e-05,
+      "loss": 1.2393,
+      "step": 4976
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7457752497195397e-05,
+      "loss": 0.8058,
+      "step": 4977
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.744876271810148e-05,
+      "loss": 1.0541,
+      "step": 4978
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7439774013323785e-05,
+      "loss": 0.9178,
+      "step": 4979
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.743078638414114e-05,
+      "loss": 0.8457,
+      "step": 4980
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7421799831832213e-05,
+      "loss": 1.1372,
+      "step": 4981
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.741281435767552e-05,
+      "loss": 1.137,
+      "step": 4982
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.740382996294943e-05,
+      "loss": 0.9568,
+      "step": 4983
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7394846648932155e-05,
+      "loss": 1.0339,
+      "step": 4984
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7385864416901763e-05,
+      "loss": 1.1548,
+      "step": 4985
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7376883268136136e-05,
+      "loss": 1.144,
+      "step": 4986
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.736790320391304e-05,
+      "loss": 1.2333,
+      "step": 4987
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.735892422551006e-05,
+      "loss": 0.9902,
+      "step": 4988
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7349946334204646e-05,
+      "loss": 0.9744,
+      "step": 4989
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7340969531274088e-05,
+      "loss": 0.6835,
+      "step": 4990
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7331993817995506e-05,
+      "loss": 0.9523,
+      "step": 4991
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7323019195645897e-05,
+      "loss": 1.1332,
+      "step": 4992
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7314045665502054e-05,
+      "loss": 1.0887,
+      "step": 4993
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7305073228840663e-05,
+      "loss": 1.2921,
+      "step": 4994
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7296101886938236e-05,
+      "loss": 0.8899,
+      "step": 4995
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7287131641071113e-05,
+      "loss": 1.0973,
+      "step": 4996
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7278162492515502e-05,
+      "loss": 1.0298,
+      "step": 4997
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.726919444254744e-05,
+      "loss": 0.8606,
+      "step": 4998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7260227492442833e-05,
+      "loss": 0.7057,
+      "step": 4999
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7251261643477383e-05,
+      "loss": 1.2054,
+      "step": 5000
+    },
+    {
+      "epoch": 0.6,
+      "eval_loss": 1.0751953125,
+      "eval_runtime": 3.6315,
+      "eval_samples_per_second": 27.537,
+      "eval_steps_per_second": 4.681,
+      "step": 5000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.724229689692668e-05,
+      "loss": 1.0157,
+      "step": 5001
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7233333254066133e-05,
+      "loss": 0.9614,
+      "step": 5002
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.722437071617101e-05,
+      "loss": 0.9426,
+      "step": 5003
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7215409284516408e-05,
+      "loss": 1.3408,
+      "step": 5004
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7206448960377275e-05,
+      "loss": 0.9082,
+      "step": 5005
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.71974897450284e-05,
+      "loss": 1.093,
+      "step": 5006
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7188531639744403e-05,
+      "loss": 1.0232,
+      "step": 5007
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7179574645799762e-05,
+      "loss": 0.9607,
+      "step": 5008
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7170618764468787e-05,
+      "loss": 0.9002,
+      "step": 5009
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7161663997025644e-05,
+      "loss": 0.8,
+      "step": 5010
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.715271034474432e-05,
+      "loss": 1.1681,
+      "step": 5011
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.714375780889865e-05,
+      "loss": 1.1782,
+      "step": 5012
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.713480639076232e-05,
+      "loss": 0.9131,
+      "step": 5013
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7125856091608854e-05,
+      "loss": 0.9213,
+      "step": 5014
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7116906912711603e-05,
+      "loss": 0.7463,
+      "step": 5015
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7107958855343768e-05,
+      "loss": 1.088,
+      "step": 5016
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.70990119207784e-05,
+      "loss": 1.2485,
+      "step": 5017
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7090066110288377e-05,
+      "loss": 1.2123,
+      "step": 5018
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7081121425146425e-05,
+      "loss": 1.1533,
+      "step": 5019
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.707217786662511e-05,
+      "loss": 0.7894,
+      "step": 5020
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.706323543599682e-05,
+      "loss": 1.0518,
+      "step": 5021
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7054294134533803e-05,
+      "loss": 1.1349,
+      "step": 5022
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7045353963508146e-05,
+      "loss": 0.8027,
+      "step": 5023
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7036414924191767e-05,
+      "loss": 1.1732,
+      "step": 5024
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.702747701785642e-05,
+      "loss": 1.1573,
+      "step": 5025
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.70185402457737e-05,
+      "loss": 1.1006,
+      "step": 5026
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7009604609215052e-05,
+      "loss": 1.0356,
+      "step": 5027
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7000670109451754e-05,
+      "loss": 1.0082,
+      "step": 5028
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.69917367477549e-05,
+      "loss": 1.0247,
+      "step": 5029
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6982804525395466e-05,
+      "loss": 0.9752,
+      "step": 5030
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.697387344364423e-05,
+      "loss": 0.923,
+      "step": 5031
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6964943503771814e-05,
+      "loss": 1.1692,
+      "step": 5032
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6956014707048687e-05,
+      "loss": 1.0452,
+      "step": 5033
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.694708705474516e-05,
+      "loss": 1.226,
+      "step": 5034
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6938160548131358e-05,
+      "loss": 1.293,
+      "step": 5035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.692923518847726e-05,
+      "loss": 0.8342,
+      "step": 5036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6920310977052683e-05,
+      "loss": 0.8544,
+      "step": 5037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.6911387915127284e-05,
+      "loss": 0.9451,
+      "step": 5038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.690246600397053e-05,
+      "loss": 0.9646,
+      "step": 5039
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6893545244851756e-05,
+      "loss": 0.9259,
+      "step": 5040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.688462563904012e-05,
+      "loss": 1.3623,
+      "step": 5041
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6875707187804617e-05,
+      "loss": 1.0765,
+      "step": 5042
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6866789892414054e-05,
+      "loss": 1.2798,
+      "step": 5043
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.685787375413714e-05,
+      "loss": 1.1755,
+      "step": 5044
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6848958774242347e-05,
+      "loss": 0.778,
+      "step": 5045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6840044953998014e-05,
+      "loss": 0.9468,
+      "step": 5046
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6831132294672313e-05,
+      "loss": 1.2732,
+      "step": 5047
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.682222079753326e-05,
+      "loss": 0.9072,
+      "step": 5048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6813310463848687e-05,
+      "loss": 0.8378,
+      "step": 5049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6804401294886268e-05,
+      "loss": 1.2645,
+      "step": 5050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.679549329191351e-05,
+      "loss": 1.5779,
+      "step": 5051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6786586456197774e-05,
+      "loss": 0.9465,
+      "step": 5052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6777680789006216e-05,
+      "loss": 1.2625,
+      "step": 5053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6768776291605853e-05,
+      "loss": 1.0262,
+      "step": 5054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6759872965263536e-05,
+      "loss": 1.0834,
+      "step": 5055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6750970811245935e-05,
+      "loss": 1.0992,
+      "step": 5056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.674206983081957e-05,
+      "loss": 0.7985,
+      "step": 5057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.673317002525079e-05,
+      "loss": 1.275,
+      "step": 5058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6724271395805756e-05,
+      "loss": 0.9423,
+      "step": 5059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6715373943750488e-05,
+      "loss": 0.7994,
+      "step": 5060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.670647767035083e-05,
+      "loss": 0.8032,
+      "step": 5061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6697582576872454e-05,
+      "loss": 0.9784,
+      "step": 5062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6688688664580863e-05,
+      "loss": 0.8787,
+      "step": 5063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.66797959347414e-05,
+      "loss": 1.1455,
+      "step": 5064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6670904388619235e-05,
+      "loss": 1.1185,
+      "step": 5065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6662014027479374e-05,
+      "loss": 1.1246,
+      "step": 5066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6653124852586645e-05,
+      "loss": 1.0813,
+      "step": 5067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6644236865205715e-05,
+      "loss": 1.2148,
+      "step": 5068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6635350066601076e-05,
+      "loss": 1.2288,
+      "step": 5069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6626464458037056e-05,
+      "loss": 1.0829,
+      "step": 5070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6617580040777823e-05,
+      "loss": 0.9888,
+      "step": 5071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.660869681608736e-05,
+      "loss": 0.7164,
+      "step": 5072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6599814785229478e-05,
+      "loss": 0.9581,
+      "step": 5073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.659093394946783e-05,
+      "loss": 0.9489,
+      "step": 5074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6582054310065896e-05,
+      "loss": 1.1169,
+      "step": 5075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6573175868286987e-05,
+      "loss": 1.0096,
+      "step": 5076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.656429862539423e-05,
+      "loss": 1.0571,
+      "step": 5077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6555422582650604e-05,
+      "loss": 0.8921,
+      "step": 5078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6546547741318895e-05,
+      "loss": 1.0107,
+      "step": 5079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6537674102661743e-05,
+      "loss": 0.7959,
+      "step": 5080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.652880166794159e-05,
+      "loss": 0.9447,
+      "step": 5081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6519930438420723e-05,
+      "loss": 1.2622,
+      "step": 5082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6511060415361253e-05,
+      "loss": 1.1824,
+      "step": 5083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6502191600025117e-05,
+      "loss": 1.2662,
+      "step": 5084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6493323993674093e-05,
+      "loss": 0.8518,
+      "step": 5085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6484457597569778e-05,
+      "loss": 0.9745,
+      "step": 5086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6475592412973583e-05,
+      "loss": 1.2537,
+      "step": 5087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6466728441146766e-05,
+      "loss": 1.1912,
+      "step": 5088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.645786568335041e-05,
+      "loss": 1.0554,
+      "step": 5089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6449004140845427e-05,
+      "loss": 0.9125,
+      "step": 5090
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6440143814892532e-05,
+      "loss": 1.2433,
+      "step": 5091
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6431284706752296e-05,
+      "loss": 1.2136,
+      "step": 5092
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.642242681768511e-05,
+      "loss": 1.1272,
+      "step": 5093
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6413570148951184e-05,
+      "loss": 1.0022,
+      "step": 5094
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.640471470181056e-05,
+      "loss": 1.1141,
+      "step": 5095
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6395860477523102e-05,
+      "loss": 1.3384,
+      "step": 5096
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6387007477348494e-05,
+      "loss": 1.0472,
+      "step": 5097
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.637815570254627e-05,
+      "loss": 1.1509,
+      "step": 5098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.636930515437577e-05,
+      "loss": 0.7628,
+      "step": 5099
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6360455834096156e-05,
+      "loss": 0.8589,
+      "step": 5100
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.635160774296644e-05,
+      "loss": 0.6637,
+      "step": 5101
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6342760882245417e-05,
+      "loss": 0.8895,
+      "step": 5102
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6333915253191752e-05,
+      "loss": 0.6302,
+      "step": 5103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6325070857063903e-05,
+      "loss": 0.776,
+      "step": 5104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6316227695120175e-05,
+      "loss": 1.0361,
+      "step": 5105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6307385768618677e-05,
+      "loss": 0.9244,
+      "step": 5106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.629854507881735e-05,
+      "loss": 1.0327,
+      "step": 5107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6289705626973972e-05,
+      "loss": 0.8984,
+      "step": 5108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.628086741434613e-05,
+      "loss": 0.9111,
+      "step": 5109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.627203044219123e-05,
+      "loss": 1.2975,
+      "step": 5110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.626319471176651e-05,
+      "loss": 1.0001,
+      "step": 5111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6254360224329045e-05,
+      "loss": 1.0326,
+      "step": 5112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6245526981135715e-05,
+      "loss": 0.9084,
+      "step": 5113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.623669498344322e-05,
+      "loss": 0.8459,
+      "step": 5114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6227864232508104e-05,
+      "loss": 1.0844,
+      "step": 5115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.62190347295867e-05,
+      "loss": 1.1588,
+      "step": 5116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6210206475935198e-05,
+      "loss": 0.9354,
+      "step": 5117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6201379472809593e-05,
+      "loss": 1.1892,
+      "step": 5118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.619255372146571e-05,
+      "loss": 0.942,
+      "step": 5119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6183729223159173e-05,
+      "loss": 1.1931,
+      "step": 5120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.617490597914546e-05,
+      "loss": 1.0625,
+      "step": 5121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.6166083990679853e-05,
+      "loss": 1.0095,
+      "step": 5122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.615726325901746e-05,
+      "loss": 0.9712,
+      "step": 5123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.614844378541319e-05,
+      "loss": 1.135,
+      "step": 5124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.613962557112183e-05,
+      "loss": 1.1488,
+      "step": 5125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6130808617397917e-05,
+      "loss": 0.8905,
+      "step": 5126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6121992925495853e-05,
+      "loss": 0.9113,
+      "step": 5127
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6113178496669844e-05,
+      "loss": 0.8673,
+      "step": 5128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.610436533217393e-05,
+      "loss": 0.8463,
+      "step": 5129
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6095553433261947e-05,
+      "loss": 0.9458,
+      "step": 5130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6086742801187575e-05,
+      "loss": 1.2592,
+      "step": 5131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6077933437204307e-05,
+      "loss": 1.0662,
+      "step": 5132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6069125342565454e-05,
+      "loss": 0.8198,
+      "step": 5133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6060318518524133e-05,
+      "loss": 1.3057,
+      "step": 5134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6051512966333303e-05,
+      "loss": 1.1443,
+      "step": 5135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6042708687245728e-05,
+      "loss": 1.1107,
+      "step": 5136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6033905682514005e-05,
+      "loss": 1.0679,
+      "step": 5137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.6025103953390512e-05,
+      "loss": 0.9689,
+      "step": 5138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.601630350112751e-05,
+      "loss": 0.6763,
+      "step": 5139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.600750432697702e-05,
+      "loss": 1.2378,
+      "step": 5140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5998706432190907e-05,
+      "loss": 1.1857,
+      "step": 5141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5989909818020844e-05,
+      "loss": 0.937,
+      "step": 5142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.598111448571834e-05,
+      "loss": 0.8616,
+      "step": 5143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5972320436534693e-05,
+      "loss": 0.9905,
+      "step": 5144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5963527671721045e-05,
+      "loss": 0.9161,
+      "step": 5145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5954736192528343e-05,
+      "loss": 0.7655,
+      "step": 5146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5945946000207352e-05,
+      "loss": 1.0881,
+      "step": 5147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5937157096008646e-05,
+      "loss": 0.8728,
+      "step": 5148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5928369481182636e-05,
+      "loss": 1.0352,
+      "step": 5149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5919583156979538e-05,
+      "loss": 1.0396,
+      "step": 5150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.591079812464938e-05,
+      "loss": 0.949,
+      "step": 5151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5902014385442e-05,
+      "loss": 0.891,
+      "step": 5152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5893231940607086e-05,
+      "loss": 0.9735,
+      "step": 5153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5884450791394104e-05,
+      "loss": 0.8173,
+      "step": 5154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.587567093905235e-05,
+      "loss": 1.015,
+      "step": 5155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5866892384830938e-05,
+      "loss": 0.8799,
+      "step": 5156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5858115129978802e-05,
+      "loss": 0.9669,
+      "step": 5157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5849339175744667e-05,
+      "loss": 0.9426,
+      "step": 5158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5840564523377098e-05,
+      "loss": 0.9291,
+      "step": 5159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5831791174124473e-05,
+      "loss": 0.8887,
+      "step": 5160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5823019129234977e-05,
+      "loss": 1.1926,
+      "step": 5161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5814248389956605e-05,
+      "loss": 1.2605,
+      "step": 5162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5805478957537173e-05,
+      "loss": 1.0382,
+      "step": 5163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.579671083322431e-05,
+      "loss": 0.8701,
+      "step": 5164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5787944018265458e-05,
+      "loss": 1.2893,
+      "step": 5165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.577917851390788e-05,
+      "loss": 1.051,
+      "step": 5166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5770414321398648e-05,
+      "loss": 1.2412,
+      "step": 5167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5761651441984636e-05,
+      "loss": 1.1775,
+      "step": 5168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5752889876912542e-05,
+      "loss": 1.0995,
+      "step": 5169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.574412962742888e-05,
+      "loss": 1.0277,
+      "step": 5170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.573537069477998e-05,
+      "loss": 1.2531,
+      "step": 5171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5726613080211962e-05,
+      "loss": 1.0641,
+      "step": 5172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.571785678497078e-05,
+      "loss": 0.9177,
+      "step": 5173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5709101810302195e-05,
+      "loss": 1.0688,
+      "step": 5174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5700348157451784e-05,
+      "loss": 1.0603,
+      "step": 5175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.569159582766492e-05,
+      "loss": 0.8235,
+      "step": 5176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5682844822186807e-05,
+      "loss": 0.8311,
+      "step": 5177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.567409514226245e-05,
+      "loss": 1.2184,
+      "step": 5178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5665346789136663e-05,
+      "loss": 0.9706,
+      "step": 5179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5656599764054085e-05,
+      "loss": 0.7943,
+      "step": 5180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.564785406825916e-05,
+      "loss": 0.9552,
+      "step": 5181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5639109702996128e-05,
+      "loss": 1.1185,
+      "step": 5182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.563036666950906e-05,
+      "loss": 1.1414,
+      "step": 5183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5621624969041828e-05,
+      "loss": 1.0022,
+      "step": 5184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.561288460283812e-05,
+      "loss": 1.0895,
+      "step": 5185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5604145572141416e-05,
+      "loss": 0.8905,
+      "step": 5186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.559540787819503e-05,
+      "loss": 1.0182,
+      "step": 5187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5586671522242075e-05,
+      "loss": 0.8237,
+      "step": 5188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5577936505525482e-05,
+      "loss": 1.2986,
+      "step": 5189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5569202829287965e-05,
+      "loss": 1.2582,
+      "step": 5190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.556047049477208e-05,
+      "loss": 0.8241,
+      "step": 5191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.555173950322017e-05,
+      "loss": 1.1599,
+      "step": 5192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5543009855874397e-05,
+      "loss": 1.0795,
+      "step": 5193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5534281553976738e-05,
+      "loss": 1.5002,
+      "step": 5194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5525554598768966e-05,
+      "loss": 1.0292,
+      "step": 5195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5516828991492665e-05,
+      "loss": 0.8755,
+      "step": 5196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5508104733389233e-05,
+      "loss": 0.9999,
+      "step": 5197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5499381825699864e-05,
+      "loss": 1.0671,
+      "step": 5198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5490660269665574e-05,
+      "loss": 0.9595,
+      "step": 5199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.548194006652719e-05,
+      "loss": 1.1807,
+      "step": 5200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5473221217525315e-05,
+      "loss": 1.134,
+      "step": 5201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5464503723900397e-05,
+      "loss": 1.1382,
+      "step": 5202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5455787586892673e-05,
+      "loss": 0.9546,
+      "step": 5203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5447072807742192e-05,
+      "loss": 0.9604,
+      "step": 5204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.5438359387688805e-05,
+      "loss": 1.2089,
+      "step": 5205
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5429647327972164e-05,
+      "loss": 1.1589,
+      "step": 5206
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5420936629831746e-05,
+      "loss": 1.1179,
+      "step": 5207
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5412227294506825e-05,
+      "loss": 1.2306,
+      "step": 5208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5403519323236475e-05,
+      "loss": 0.8727,
+      "step": 5209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5394812717259595e-05,
+      "loss": 0.8333,
+      "step": 5210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5386107477814847e-05,
+      "loss": 1.1111,
+      "step": 5211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5377403606140752e-05,
+      "loss": 1.2676,
+      "step": 5212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.53687011034756e-05,
+      "loss": 0.6912,
+      "step": 5213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5359999971057513e-05,
+      "loss": 1.0538,
+      "step": 5214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.535130021012438e-05,
+      "loss": 1.2349,
+      "step": 5215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5342601821913935e-05,
+      "loss": 0.9103,
+      "step": 5216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.533390480766369e-05,
+      "loss": 0.8369,
+      "step": 5217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5325209168610987e-05,
+      "loss": 1.0785,
+      "step": 5218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5316514905992935e-05,
+      "loss": 1.1003,
+      "step": 5219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5307822021046475e-05,
+      "loss": 1.1527,
+      "step": 5220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5299130515008354e-05,
+      "loss": 0.996,
+      "step": 5221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.529044038911511e-05,
+      "loss": 1.2434,
+      "step": 5222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.528175164460309e-05,
+      "loss": 0.9982,
+      "step": 5223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.527306428270845e-05,
+      "loss": 0.9869,
+      "step": 5224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5264378304667125e-05,
+      "loss": 1.0415,
+      "step": 5225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5255693711714887e-05,
+      "loss": 0.9222,
+      "step": 5226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5247010505087289e-05,
+      "loss": 1.017,
+      "step": 5227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5238328686019698e-05,
+      "loss": 1.0646,
+      "step": 5228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.522964825574727e-05,
+      "loss": 0.842,
+      "step": 5229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.522096921550497e-05,
+      "loss": 0.9938,
+      "step": 5230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.521229156652758e-05,
+      "loss": 1.0004,
+      "step": 5231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5203615310049665e-05,
+      "loss": 0.9772,
+      "step": 5232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5194940447305584e-05,
+      "loss": 1.334,
+      "step": 5233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5186266979529537e-05,
+      "loss": 1.1847,
+      "step": 5234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5177594907955484e-05,
+      "loss": 1.1687,
+      "step": 5235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5168924233817206e-05,
+      "loss": 1.2156,
+      "step": 5236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.516025495834828e-05,
+      "loss": 0.8702,
+      "step": 5237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5151587082782095e-05,
+      "loss": 1.0148,
+      "step": 5238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5142920608351819e-05,
+      "loss": 1.2089,
+      "step": 5239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5134255536290437e-05,
+      "loss": 1.0159,
+      "step": 5240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5125591867830735e-05,
+      "loss": 1.2075,
+      "step": 5241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5116929604205297e-05,
+      "loss": 1.3789,
+      "step": 5242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5108268746646492e-05,
+      "loss": 0.8756,
+      "step": 5243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5099609296386513e-05,
+      "loss": 0.9794,
+      "step": 5244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5090951254657343e-05,
+      "loss": 1.0912,
+      "step": 5245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.508229462269076e-05,
+      "loss": 1.0792,
+      "step": 5246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.507363940171833e-05,
+      "loss": 0.8315,
+      "step": 5247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5064985592971467e-05,
+      "loss": 0.9679,
+      "step": 5248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5056333197681327e-05,
+      "loss": 1.1664,
+      "step": 5249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.504768221707889e-05,
+      "loss": 1.0562,
+      "step": 5250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5039032652394936e-05,
+      "loss": 1.1626,
+      "step": 5251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5030384504860045e-05,
+      "loss": 0.9547,
+      "step": 5252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5021737775704584e-05,
+      "loss": 0.8877,
+      "step": 5253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.5013092466158723e-05,
+      "loss": 1.3387,
+      "step": 5254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.500444857745244e-05,
+      "loss": 0.9325,
+      "step": 5255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4995806110815502e-05,
+      "loss": 1.1727,
+      "step": 5256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.498716506747747e-05,
+      "loss": 1.129,
+      "step": 5257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4978525448667707e-05,
+      "loss": 1.291,
+      "step": 5258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4969887255615378e-05,
+      "loss": 1.1427,
+      "step": 5259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4961250489549441e-05,
+      "loss": 1.0557,
+      "step": 5260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4952615151698635e-05,
+      "loss": 0.9465,
+      "step": 5261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4943981243291538e-05,
+      "loss": 0.9746,
+      "step": 5262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4935348765556478e-05,
+      "loss": 1.1775,
+      "step": 5263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4926717719721612e-05,
+      "loss": 1.0999,
+      "step": 5264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4918088107014872e-05,
+      "loss": 1.3208,
+      "step": 5265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4909459928663999e-05,
+      "loss": 1.3345,
+      "step": 5266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4900833185896524e-05,
+      "loss": 0.807,
+      "step": 5267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4892207879939773e-05,
+      "loss": 1.304,
+      "step": 5268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4883584012020879e-05,
+      "loss": 0.8385,
+      "step": 5269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4874961583366758e-05,
+      "loss": 1.0239,
+      "step": 5270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.486634059520412e-05,
+      "loss": 0.8639,
+      "step": 5271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4857721048759476e-05,
+      "loss": 0.8574,
+      "step": 5272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4849102945259133e-05,
+      "loss": 0.9482,
+      "step": 5273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4840486285929183e-05,
+      "loss": 1.0762,
+      "step": 5274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4831871071995537e-05,
+      "loss": 1.0815,
+      "step": 5275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4823257304683877e-05,
+      "loss": 0.9266,
+      "step": 5276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4814644985219678e-05,
+      "loss": 1.2103,
+      "step": 5277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4806034114828221e-05,
+      "loss": 1.1401,
+      "step": 5278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4797424694734574e-05,
+      "loss": 1.0994,
+      "step": 5279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4788816726163613e-05,
+      "loss": 1.0554,
+      "step": 5280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4780210210339979e-05,
+      "loss": 1.1571,
+      "step": 5281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4771605148488126e-05,
+      "loss": 1.2102,
+      "step": 5282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4763001541832308e-05,
+      "loss": 1.0035,
+      "step": 5283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.475439939159656e-05,
+      "loss": 1.2695,
+      "step": 5284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4745798699004704e-05,
+      "loss": 1.3511,
+      "step": 5285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4737199465280366e-05,
+      "loss": 1.1915,
+      "step": 5286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.472860169164696e-05,
+      "loss": 0.9401,
+      "step": 5287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.472000537932769e-05,
+      "loss": 1.0411,
+      "step": 5288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.4711410529545566e-05,
+      "loss": 1.2118,
+      "step": 5289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4702817143523372e-05,
+      "loss": 0.9856,
+      "step": 5290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4694225222483699e-05,
+      "loss": 0.8499,
+      "step": 5291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.468563476764891e-05,
+      "loss": 1.0171,
+      "step": 5292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4677045780241176e-05,
+      "loss": 0.8862,
+      "step": 5293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.466845826148245e-05,
+      "loss": 1.0643,
+      "step": 5294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4659872212594492e-05,
+      "loss": 0.9749,
+      "step": 5295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4651287634798826e-05,
+      "loss": 0.8177,
+      "step": 5296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4642704529316791e-05,
+      "loss": 1.0781,
+      "step": 5297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4634122897369513e-05,
+      "loss": 0.9581,
+      "step": 5298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4625542740177883e-05,
+      "loss": 1.0428,
+      "step": 5299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4616964058962618e-05,
+      "loss": 0.6865,
+      "step": 5300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.46083868549442e-05,
+      "loss": 0.9581,
+      "step": 5301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.459981112934292e-05,
+      "loss": 1.0477,
+      "step": 5302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.459123688337884e-05,
+      "loss": 0.9475,
+      "step": 5303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4582664118271832e-05,
+      "loss": 1.1019,
+      "step": 5304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.457409283524152e-05,
+      "loss": 0.75,
+      "step": 5305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4565523035507373e-05,
+      "loss": 0.899,
+      "step": 5306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4556954720288604e-05,
+      "loss": 0.7679,
+      "step": 5307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4548387890804216e-05,
+      "loss": 1.0767,
+      "step": 5308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.453982254827304e-05,
+      "loss": 0.6591,
+      "step": 5309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4531258693913644e-05,
+      "loss": 0.8395,
+      "step": 5310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.452269632894443e-05,
+      "loss": 0.7855,
+      "step": 5311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4514135454583561e-05,
+      "loss": 0.9608,
+      "step": 5312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4505576072048981e-05,
+      "loss": 0.858,
+      "step": 5313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4497018182558458e-05,
+      "loss": 1.2039,
+      "step": 5314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4488461787329502e-05,
+      "loss": 1.0001,
+      "step": 5315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4479906887579447e-05,
+      "loss": 0.9521,
+      "step": 5316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4471353484525407e-05,
+      "loss": 1.3383,
+      "step": 5317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4462801579384266e-05,
+      "loss": 1.0034,
+      "step": 5318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4454251173372698e-05,
+      "loss": 1.0087,
+      "step": 5319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4445702267707192e-05,
+      "loss": 0.739,
+      "step": 5320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.443715486360399e-05,
+      "loss": 1.0167,
+      "step": 5321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4428608962279125e-05,
+      "loss": 0.7428,
+      "step": 5322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4420064564948444e-05,
+      "loss": 1.0524,
+      "step": 5323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4411521672827538e-05,
+      "loss": 1.162,
+      "step": 5324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4402980287131829e-05,
+      "loss": 0.9121,
+      "step": 5325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4394440409076493e-05,
+      "loss": 1.1522,
+      "step": 5326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4385902039876486e-05,
+      "loss": 0.8488,
+      "step": 5327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4377365180746588e-05,
+      "loss": 0.9073,
+      "step": 5328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4368829832901315e-05,
+      "loss": 1.011,
+      "step": 5329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.436029599755501e-05,
+      "loss": 0.7827,
+      "step": 5330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4351763675921786e-05,
+      "loss": 1.2126,
+      "step": 5331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4343232869215533e-05,
+      "loss": 1.2523,
+      "step": 5332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.433470357864992e-05,
+      "loss": 1.0465,
+      "step": 5333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.432617580543843e-05,
+      "loss": 1.1013,
+      "step": 5334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.43176495507943e-05,
+      "loss": 0.6983,
+      "step": 5335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4309124815930555e-05,
+      "loss": 1.0591,
+      "step": 5336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4300601602060032e-05,
+      "loss": 0.942,
+      "step": 5337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4292079910395307e-05,
+      "loss": 1.155,
+      "step": 5338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4283559742148784e-05,
+      "loss": 0.9495,
+      "step": 5339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.427504109853262e-05,
+      "loss": 1.2191,
+      "step": 5340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4266523980758756e-05,
+      "loss": 0.8826,
+      "step": 5341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.425800839003893e-05,
+      "loss": 0.9583,
+      "step": 5342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4249494327584668e-05,
+      "loss": 0.9088,
+      "step": 5343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.424098179460725e-05,
+      "loss": 0.9551,
+      "step": 5344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4232470792317776e-05,
+      "loss": 1.1356,
+      "step": 5345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4223961321927099e-05,
+      "loss": 0.7781,
+      "step": 5346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4215453384645847e-05,
+      "loss": 1.2838,
+      "step": 5347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.420694698168447e-05,
+      "loss": 1.1403,
+      "step": 5348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4198442114253168e-05,
+      "loss": 0.9554,
+      "step": 5349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4189938783561921e-05,
+      "loss": 0.9967,
+      "step": 5350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4181436990820515e-05,
+      "loss": 0.8793,
+      "step": 5351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4172936737238485e-05,
+      "loss": 0.77,
+      "step": 5352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4164438024025183e-05,
+      "loss": 0.9941,
+      "step": 5353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4155940852389715e-05,
+      "loss": 0.7581,
+      "step": 5354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4147445223540962e-05,
+      "loss": 0.7278,
+      "step": 5355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4138951138687612e-05,
+      "loss": 0.9417,
+      "step": 5356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4130458599038127e-05,
+      "loss": 1.0569,
+      "step": 5357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4121967605800726e-05,
+      "loss": 0.8361,
+      "step": 5358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4113478160183444e-05,
+      "loss": 0.901,
+      "step": 5359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4104990263394064e-05,
+      "loss": 1.1144,
+      "step": 5360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4096503916640153e-05,
+      "loss": 0.8403,
+      "step": 5361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4088019121129088e-05,
+      "loss": 1.1892,
+      "step": 5362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4079535878067984e-05,
+      "loss": 1.1676,
+      "step": 5363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4071054188663752e-05,
+      "loss": 1.1061,
+      "step": 5364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4062574054123102e-05,
+      "loss": 1.2695,
+      "step": 5365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4054095475652484e-05,
+      "loss": 1.0562,
+      "step": 5366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4045618454458168e-05,
+      "loss": 0.7814,
+      "step": 5367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4037142991746171e-05,
+      "loss": 0.8842,
+      "step": 5368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4028669088722288e-05,
+      "loss": 1.3253,
+      "step": 5369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4020196746592113e-05,
+      "loss": 0.9603,
+      "step": 5370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.401172596656102e-05,
+      "loss": 0.9187,
+      "step": 5371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.4003256749834131e-05,
+      "loss": 0.8333,
+      "step": 5372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3994789097616377e-05,
+      "loss": 1.0883,
+      "step": 5373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3986323011112445e-05,
+      "loss": 1.2168,
+      "step": 5374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3977858491526801e-05,
+      "loss": 0.9655,
+      "step": 5375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.396939554006371e-05,
+      "loss": 1.3323,
+      "step": 5376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3960934157927189e-05,
+      "loss": 0.9844,
+      "step": 5377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3952474346321031e-05,
+      "loss": 0.8569,
+      "step": 5378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3944016106448837e-05,
+      "loss": 0.845,
+      "step": 5379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3935559439513936e-05,
+      "loss": 0.9479,
+      "step": 5380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3927104346719488e-05,
+      "loss": 1.0898,
+      "step": 5381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.391865082926839e-05,
+      "loss": 1.2484,
+      "step": 5382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3910198888363302e-05,
+      "loss": 1.063,
+      "step": 5383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3901748525206723e-05,
+      "loss": 1.1475,
+      "step": 5384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.389329974100087e-05,
+      "loss": 0.9706,
+      "step": 5385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3884852536947745e-05,
+      "loss": 0.8756,
+      "step": 5386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3876406914249155e-05,
+      "loss": 0.8463,
+      "step": 5387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3867962874106644e-05,
+      "loss": 0.8262,
+      "step": 5388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3859520417721542e-05,
+      "loss": 1.0486,
+      "step": 5389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.385107954629498e-05,
+      "loss": 1.2339,
+      "step": 5390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3842640261027829e-05,
+      "loss": 1.1262,
+      "step": 5391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.383420256312074e-05,
+      "loss": 0.7983,
+      "step": 5392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3825766453774169e-05,
+      "loss": 0.7654,
+      "step": 5393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.38173319341883e-05,
+      "loss": 1.0353,
+      "step": 5394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3808899005563131e-05,
+      "loss": 0.9945,
+      "step": 5395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3800467669098412e-05,
+      "loss": 1.0865,
+      "step": 5396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.379203792599365e-05,
+      "loss": 0.9058,
+      "step": 5397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3783609777448187e-05,
+      "loss": 0.9001,
+      "step": 5398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3775183224661076e-05,
+      "loss": 1.1864,
+      "step": 5399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3766758268831154e-05,
+      "loss": 0.7322,
+      "step": 5400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3758334911157061e-05,
+      "loss": 0.9214,
+      "step": 5401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3749913152837179e-05,
+      "loss": 0.8324,
+      "step": 5402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3741492995069672e-05,
+      "loss": 0.8867,
+      "step": 5403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3733074439052484e-05,
+      "loss": 0.7419,
+      "step": 5404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3724657485983317e-05,
+      "loss": 1.095,
+      "step": 5405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3716242137059671e-05,
+      "loss": 0.8146,
+      "step": 5406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.370782839347878e-05,
+      "loss": 1.213,
+      "step": 5407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.369941625643767e-05,
+      "loss": 0.9155,
+      "step": 5408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3691005727133148e-05,
+      "loss": 1.3611,
+      "step": 5409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3682596806761775e-05,
+      "loss": 1.1036,
+      "step": 5410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3674189496519887e-05,
+      "loss": 1.1382,
+      "step": 5411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.366578379760361e-05,
+      "loss": 0.8741,
+      "step": 5412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3657379711208817e-05,
+      "loss": 1.1807,
+      "step": 5413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.364897723853114e-05,
+      "loss": 0.8162,
+      "step": 5414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3640576380766035e-05,
+      "loss": 0.7183,
+      "step": 5415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.363217713910867e-05,
+      "loss": 0.7839,
+      "step": 5416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3623779514754003e-05,
+      "loss": 1.2358,
+      "step": 5417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3615383508896787e-05,
+      "loss": 0.9311,
+      "step": 5418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.36069891227315e-05,
+      "loss": 1.045,
+      "step": 5419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3598596357452437e-05,
+      "loss": 1.0747,
+      "step": 5420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3590205214253626e-05,
+      "loss": 1.1222,
+      "step": 5421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.358181569432887e-05,
+      "loss": 1.1339,
+      "step": 5422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3573427798871765e-05,
+      "loss": 1.2762,
+      "step": 5423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3565041529075639e-05,
+      "loss": 1.0574,
+      "step": 5424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3556656886133621e-05,
+      "loss": 1.0042,
+      "step": 5425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3548273871238604e-05,
+      "loss": 1.0984,
+      "step": 5426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3539892485583228e-05,
+      "loss": 1.1453,
+      "step": 5427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3531512730359907e-05,
+      "loss": 0.9983,
+      "step": 5428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3523134606760856e-05,
+      "loss": 1.0332,
+      "step": 5429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3514758115978016e-05,
+      "loss": 1.0726,
+      "step": 5430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3506383259203107e-05,
+      "loss": 1.1021,
+      "step": 5431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3498010037627637e-05,
+      "loss": 1.1879,
+      "step": 5432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3489638452442848e-05,
+      "loss": 0.817,
+      "step": 5433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3481268504839788e-05,
+      "loss": 0.889,
+      "step": 5434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.347290019600924e-05,
+      "loss": 1.3424,
+      "step": 5435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.346453352714176e-05,
+      "loss": 0.8918,
+      "step": 5436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3456168499427679e-05,
+      "loss": 1.3269,
+      "step": 5437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3447805114057105e-05,
+      "loss": 0.7834,
+      "step": 5438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3439443372219879e-05,
+      "loss": 1.1136,
+      "step": 5439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3431083275105643e-05,
+      "loss": 1.0614,
+      "step": 5440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3422724823903788e-05,
+      "loss": 0.9454,
+      "step": 5441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3414368019803458e-05,
+      "loss": 1.0574,
+      "step": 5442
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3406012863993594e-05,
+      "loss": 1.016,
+      "step": 5443
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3397659357662884e-05,
+      "loss": 0.9988,
+      "step": 5444
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.338930750199977e-05,
+      "loss": 1.0229,
+      "step": 5445
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3380957298192488e-05,
+      "loss": 0.9358,
+      "step": 5446
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3372608747429008e-05,
+      "loss": 0.8022,
+      "step": 5447
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3364261850897098e-05,
+      "loss": 0.7655,
+      "step": 5448
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3355916609784266e-05,
+      "loss": 0.9519,
+      "step": 5449
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3347573025277776e-05,
+      "loss": 1.01,
+      "step": 5450
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.333923109856469e-05,
+      "loss": 1.0424,
+      "step": 5451
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3330890830831817e-05,
+      "loss": 1.0305,
+      "step": 5452
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.332255222326571e-05,
+      "loss": 0.9658,
+      "step": 5453
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3314215277052733e-05,
+      "loss": 1.0585,
+      "step": 5454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.3305879993378966e-05,
+      "loss": 1.2522,
+      "step": 5455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3297546373430264e-05,
+      "loss": 0.917,
+      "step": 5456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3289214418392273e-05,
+      "loss": 1.0635,
+      "step": 5457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3280884129450376e-05,
+      "loss": 1.1794,
+      "step": 5458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3272555507789709e-05,
+      "loss": 1.0144,
+      "step": 5459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3264228554595213e-05,
+      "loss": 1.193,
+      "step": 5460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.325590327105154e-05,
+      "loss": 0.7892,
+      "step": 5461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.324757965834315e-05,
+      "loss": 0.6935,
+      "step": 5462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3239257717654241e-05,
+      "loss": 0.9629,
+      "step": 5463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3230937450168765e-05,
+      "loss": 1.0115,
+      "step": 5464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3222618857070457e-05,
+      "loss": 0.9294,
+      "step": 5465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3214301939542812e-05,
+      "loss": 0.8519,
+      "step": 5466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3205986698769063e-05,
+      "loss": 1.0497,
+      "step": 5467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3197673135932242e-05,
+      "loss": 1.0269,
+      "step": 5468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3189361252215115e-05,
+      "loss": 0.7529,
+      "step": 5469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3181051048800196e-05,
+      "loss": 1.0563,
+      "step": 5470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3172742526869808e-05,
+      "loss": 1.093,
+      "step": 5471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3164435687605991e-05,
+      "loss": 0.9725,
+      "step": 5472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3156130532190553e-05,
+      "loss": 1.0336,
+      "step": 5473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.314782706180509e-05,
+      "loss": 1.0004,
+      "step": 5474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.313952527763092e-05,
+      "loss": 0.8669,
+      "step": 5475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.313122518084916e-05,
+      "loss": 1.2072,
+      "step": 5476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3122926772640653e-05,
+      "loss": 0.9067,
+      "step": 5477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3114630054185997e-05,
+      "loss": 0.9078,
+      "step": 5478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3106335026665611e-05,
+      "loss": 0.8213,
+      "step": 5479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3098041691259605e-05,
+      "loss": 0.9949,
+      "step": 5480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.308975004914787e-05,
+      "loss": 0.9697,
+      "step": 5481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3081460101510074e-05,
+      "loss": 0.9989,
+      "step": 5482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.307317184952562e-05,
+      "loss": 0.8137,
+      "step": 5483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3064885294373675e-05,
+      "loss": 0.9316,
+      "step": 5484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3056600437233186e-05,
+      "loss": 1.321,
+      "step": 5485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3048317279282829e-05,
+      "loss": 0.8533,
+      "step": 5486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3040035821701044e-05,
+      "loss": 1.2312,
+      "step": 5487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3031756065666054e-05,
+      "loss": 1.2292,
+      "step": 5488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3023478012355803e-05,
+      "loss": 1.142,
+      "step": 5489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3015201662948029e-05,
+      "loss": 0.9447,
+      "step": 5490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.3006927018620205e-05,
+      "loss": 0.9655,
+      "step": 5491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2998654080549544e-05,
+      "loss": 1.2439,
+      "step": 5492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2990382849913074e-05,
+      "loss": 1.0732,
+      "step": 5493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2982113327887536e-05,
+      "loss": 0.9128,
+      "step": 5494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2973845515649414e-05,
+      "loss": 1.1515,
+      "step": 5495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2965579414375e-05,
+      "loss": 1.3212,
+      "step": 5496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2957315025240303e-05,
+      "loss": 1.0063,
+      "step": 5497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2949052349421082e-05,
+      "loss": 1.1677,
+      "step": 5498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2940791388092899e-05,
+      "loss": 0.9446,
+      "step": 5499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.293253214243102e-05,
+      "loss": 1.1265,
+      "step": 5500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.292427461361051e-05,
+      "loss": 1.0323,
+      "step": 5501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2916018802806158e-05,
+      "loss": 0.9463,
+      "step": 5502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.290776471119251e-05,
+      "loss": 1.1658,
+      "step": 5503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.28995123399439e-05,
+      "loss": 1.1993,
+      "step": 5504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2891261690234382e-05,
+      "loss": 0.6984,
+      "step": 5505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2883012763237761e-05,
+      "loss": 0.809,
+      "step": 5506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.287476556012765e-05,
+      "loss": 1.0405,
+      "step": 5507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.286652008207736e-05,
+      "loss": 0.9956,
+      "step": 5508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2858276330259968e-05,
+      "loss": 1.0869,
+      "step": 5509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.285003430584834e-05,
+      "loss": 0.9861,
+      "step": 5510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2841794010015051e-05,
+      "loss": 1.145,
+      "step": 5511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2833555443932448e-05,
+      "loss": 0.9018,
+      "step": 5512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2825318608772652e-05,
+      "loss": 0.8259,
+      "step": 5513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2817083505707494e-05,
+      "loss": 0.9164,
+      "step": 5514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2808850135908607e-05,
+      "loss": 0.9698,
+      "step": 5515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2800618500547349e-05,
+      "loss": 1.2798,
+      "step": 5516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2792388600794818e-05,
+      "loss": 0.9255,
+      "step": 5517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2784160437821912e-05,
+      "loss": 0.808,
+      "step": 5518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2775934012799229e-05,
+      "loss": 0.8293,
+      "step": 5519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.276770932689716e-05,
+      "loss": 0.7297,
+      "step": 5520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2759486381285829e-05,
+      "loss": 0.7733,
+      "step": 5521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2751265177135122e-05,
+      "loss": 0.8905,
+      "step": 5522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2743045715614655e-05,
+      "loss": 0.7903,
+      "step": 5523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2734827997893828e-05,
+      "loss": 1.0542,
+      "step": 5524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2726612025141777e-05,
+      "loss": 0.9733,
+      "step": 5525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2718397798527374e-05,
+      "loss": 1.0466,
+      "step": 5526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.271018531921928e-05,
+      "loss": 0.8276,
+      "step": 5527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2701974588385867e-05,
+      "loss": 0.7577,
+      "step": 5528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2693765607195297e-05,
+      "loss": 0.8262,
+      "step": 5529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2685558376815454e-05,
+      "loss": 0.9486,
+      "step": 5530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.267735289841397e-05,
+      "loss": 1.0474,
+      "step": 5531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2669149173158265e-05,
+      "loss": 0.8218,
+      "step": 5532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2660947202215462e-05,
+      "loss": 1.2198,
+      "step": 5533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2652746986752468e-05,
+      "loss": 0.928,
+      "step": 5534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.264454852793594e-05,
+      "loss": 0.9843,
+      "step": 5535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.263635182693226e-05,
+      "loss": 1.0186,
+      "step": 5536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2628156884907572e-05,
+      "loss": 1.0378,
+      "step": 5537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.2619963703027787e-05,
+      "loss": 1.1788,
+      "step": 5538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2611772282458542e-05,
+      "loss": 0.7749,
+      "step": 5539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2603582624365224e-05,
+      "loss": 0.8365,
+      "step": 5540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2595394729912993e-05,
+      "loss": 1.1619,
+      "step": 5541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2587208600266728e-05,
+      "loss": 0.9805,
+      "step": 5542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2579024236591086e-05,
+      "loss": 0.8284,
+      "step": 5543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2570841640050457e-05,
+      "loss": 0.7277,
+      "step": 5544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.256266081180896e-05,
+      "loss": 0.8477,
+      "step": 5545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.25544817530305e-05,
+      "loss": 0.8759,
+      "step": 5546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2546304464878722e-05,
+      "loss": 1.0376,
+      "step": 5547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2538128948516992e-05,
+      "loss": 0.7166,
+      "step": 5548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.252995520510846e-05,
+      "loss": 0.865,
+      "step": 5549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2521783235815998e-05,
+      "loss": 1.0222,
+      "step": 5550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2513613041802222e-05,
+      "loss": 0.844,
+      "step": 5551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2505444624229529e-05,
+      "loss": 0.993,
+      "step": 5552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.249727798426003e-05,
+      "loss": 1.3477,
+      "step": 5553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.248911312305559e-05,
+      "loss": 0.9567,
+      "step": 5554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.248095004177784e-05,
+      "loss": 0.795,
+      "step": 5555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2472788741588124e-05,
+      "loss": 0.8802,
+      "step": 5556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2464629223647573e-05,
+      "loss": 0.7191,
+      "step": 5557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2456471489117033e-05,
+      "loss": 0.9724,
+      "step": 5558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2448315539157096e-05,
+      "loss": 1.1768,
+      "step": 5559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2440161374928124e-05,
+      "loss": 0.8195,
+      "step": 5560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2432008997590219e-05,
+      "loss": 0.9619,
+      "step": 5561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2423858408303204e-05,
+      "loss": 1.0829,
+      "step": 5562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2415709608226683e-05,
+      "loss": 1.1309,
+      "step": 5563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2407562598519978e-05,
+      "loss": 0.8186,
+      "step": 5564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2399417380342154e-05,
+      "loss": 0.9048,
+      "step": 5565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2391273954852056e-05,
+      "loss": 0.9045,
+      "step": 5566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2383132323208241e-05,
+      "loss": 0.9855,
+      "step": 5567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2374992486569009e-05,
+      "loss": 0.942,
+      "step": 5568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2366854446092439e-05,
+      "loss": 0.9559,
+      "step": 5569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2358718202936309e-05,
+      "loss": 0.9442,
+      "step": 5570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2350583758258182e-05,
+      "loss": 1.13,
+      "step": 5571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2342451113215345e-05,
+      "loss": 0.9746,
+      "step": 5572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2334320268964814e-05,
+      "loss": 1.0536,
+      "step": 5573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.232619122666338e-05,
+      "loss": 0.7795,
+      "step": 5574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2318063987467571e-05,
+      "loss": 0.9948,
+      "step": 5575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2309938552533635e-05,
+      "loss": 1.272,
+      "step": 5576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2301814923017593e-05,
+      "loss": 0.8816,
+      "step": 5577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2293693100075192e-05,
+      "loss": 1.2141,
+      "step": 5578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2285573084861913e-05,
+      "loss": 0.8455,
+      "step": 5579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2277454878533013e-05,
+      "loss": 0.9917,
+      "step": 5580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2269338482243458e-05,
+      "loss": 0.7887,
+      "step": 5581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2261223897147963e-05,
+      "loss": 1.2169,
+      "step": 5582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2253111124401011e-05,
+      "loss": 0.9778,
+      "step": 5583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2245000165156792e-05,
+      "loss": 1.1154,
+      "step": 5584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2236891020569267e-05,
+      "loss": 1.1063,
+      "step": 5585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2228783691792114e-05,
+      "loss": 0.719,
+      "step": 5586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2220678179978756e-05,
+      "loss": 0.9315,
+      "step": 5587
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2212574486282397e-05,
+      "loss": 0.7714,
+      "step": 5588
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2204472611855933e-05,
+      "loss": 1.0085,
+      "step": 5589
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.219637255785201e-05,
+      "loss": 1.1873,
+      "step": 5590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2188274325423044e-05,
+      "loss": 0.849,
+      "step": 5591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.218017791572116e-05,
+      "loss": 1.2166,
+      "step": 5592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2172083329898235e-05,
+      "loss": 0.8195,
+      "step": 5593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2163990569105896e-05,
+      "loss": 1.0576,
+      "step": 5594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.215589963449549e-05,
+      "loss": 1.0629,
+      "step": 5595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2147810527218131e-05,
+      "loss": 0.7137,
+      "step": 5596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2139723248424653e-05,
+      "loss": 1.0078,
+      "step": 5597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2131637799265622e-05,
+      "loss": 1.084,
+      "step": 5598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2123554180891378e-05,
+      "loss": 1.1239,
+      "step": 5599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2115472394451966e-05,
+      "loss": 0.9458,
+      "step": 5600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2107392441097168e-05,
+      "loss": 1.1919,
+      "step": 5601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2099314321976557e-05,
+      "loss": 1.1047,
+      "step": 5602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2091238038239392e-05,
+      "loss": 0.7406,
+      "step": 5603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2083163591034675e-05,
+      "loss": 0.991,
+      "step": 5604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2075090981511181e-05,
+      "loss": 1.1289,
+      "step": 5605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.206702021081739e-05,
+      "loss": 0.9637,
+      "step": 5606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2058951280101524e-05,
+      "loss": 0.6956,
+      "step": 5607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2050884190511572e-05,
+      "loss": 0.9398,
+      "step": 5608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2042818943195222e-05,
+      "loss": 0.9071,
+      "step": 5609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2034755539299935e-05,
+      "loss": 0.8225,
+      "step": 5610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2026693979972887e-05,
+      "loss": 0.9722,
+      "step": 5611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2018634266360981e-05,
+      "loss": 1.0626,
+      "step": 5612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2010576399610904e-05,
+      "loss": 1.0197,
+      "step": 5613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.2002520380869022e-05,
+      "loss": 0.8402,
+      "step": 5614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1994466211281483e-05,
+      "loss": 0.9493,
+      "step": 5615
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.198641389199416e-05,
+      "loss": 0.8348,
+      "step": 5616
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1978363424152652e-05,
+      "loss": 0.9857,
+      "step": 5617
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1970314808902292e-05,
+      "loss": 1.0762,
+      "step": 5618
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1962268047388173e-05,
+      "loss": 0.9761,
+      "step": 5619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1954223140755105e-05,
+      "loss": 0.6373,
+      "step": 5620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1946180090147624e-05,
+      "loss": 0.8558,
+      "step": 5621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.1938138896710041e-05,
+      "loss": 1.045,
+      "step": 5622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1930099561586356e-05,
+      "loss": 0.8635,
+      "step": 5623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1922062085920346e-05,
+      "loss": 0.8751,
+      "step": 5624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1914026470855496e-05,
+      "loss": 0.7386,
+      "step": 5625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1905992717535026e-05,
+      "loss": 0.9143,
+      "step": 5626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1897960827101917e-05,
+      "loss": 1.2551,
+      "step": 5627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.188993080069885e-05,
+      "loss": 1.1111,
+      "step": 5628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.188190263946827e-05,
+      "loss": 0.9215,
+      "step": 5629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1873876344552352e-05,
+      "loss": 1.0245,
+      "step": 5630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1865851917092987e-05,
+      "loss": 1.1132,
+      "step": 5631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1857829358231809e-05,
+      "loss": 1.0164,
+      "step": 5632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1849808669110202e-05,
+      "loss": 1.1295,
+      "step": 5633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1841789850869267e-05,
+      "loss": 0.83,
+      "step": 5634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1833772904649831e-05,
+      "loss": 0.7289,
+      "step": 5635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1825757831592486e-05,
+      "loss": 1.0339,
+      "step": 5636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1817744632837518e-05,
+      "loss": 1.02,
+      "step": 5637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1809733309524987e-05,
+      "loss": 0.6667,
+      "step": 5638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1801723862794655e-05,
+      "loss": 1.1151,
+      "step": 5639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1793716293786023e-05,
+      "loss": 0.8125,
+      "step": 5640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1785710603638342e-05,
+      "loss": 0.9532,
+      "step": 5641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1777706793490567e-05,
+      "loss": 0.8151,
+      "step": 5642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1769704864481412e-05,
+      "loss": 1.0734,
+      "step": 5643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1761704817749325e-05,
+      "loss": 1.1511,
+      "step": 5644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1753706654432458e-05,
+      "loss": 0.897,
+      "step": 5645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1745710375668708e-05,
+      "loss": 0.7974,
+      "step": 5646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1737715982595726e-05,
+      "loss": 0.8523,
+      "step": 5647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1729723476350865e-05,
+      "loss": 0.9067,
+      "step": 5648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.172173285807121e-05,
+      "loss": 1.1605,
+      "step": 5649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1713744128893611e-05,
+      "loss": 0.8057,
+      "step": 5650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1705757289954602e-05,
+      "loss": 1.1367,
+      "step": 5651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1697772342390497e-05,
+      "loss": 0.9806,
+      "step": 5652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1689789287337302e-05,
+      "loss": 0.967,
+      "step": 5653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.168180812593076e-05,
+      "loss": 1.0352,
+      "step": 5654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1673828859306363e-05,
+      "loss": 1.04,
+      "step": 5655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1665851488599333e-05,
+      "loss": 0.8749,
+      "step": 5656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1657876014944594e-05,
+      "loss": 1.1998,
+      "step": 5657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1649902439476831e-05,
+      "loss": 0.9017,
+      "step": 5658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1641930763330447e-05,
+      "loss": 1.1608,
+      "step": 5659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1633960987639556e-05,
+      "loss": 0.8888,
+      "step": 5660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1625993113538042e-05,
+      "loss": 0.843,
+      "step": 5661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1618027142159488e-05,
+      "loss": 0.7659,
+      "step": 5662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1610063074637204e-05,
+      "loss": 0.6469,
+      "step": 5663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1602100912104261e-05,
+      "loss": 0.8138,
+      "step": 5664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1594140655693411e-05,
+      "loss": 0.9554,
+      "step": 5665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1586182306537192e-05,
+      "loss": 1.1688,
+      "step": 5666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1578225865767819e-05,
+      "loss": 1.0396,
+      "step": 5667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1570271334517255e-05,
+      "loss": 0.7803,
+      "step": 5668
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1562318713917197e-05,
+      "loss": 1.4854,
+      "step": 5669
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.155436800509908e-05,
+      "loss": 0.7366,
+      "step": 5670
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1546419209194036e-05,
+      "loss": 0.817,
+      "step": 5671
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1538472327332955e-05,
+      "loss": 0.9442,
+      "step": 5672
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1530527360646434e-05,
+      "loss": 0.8387,
+      "step": 5673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1522584310264797e-05,
+      "loss": 0.8833,
+      "step": 5674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1514643177318124e-05,
+      "loss": 1.1833,
+      "step": 5675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1506703962936188e-05,
+      "loss": 0.7958,
+      "step": 5676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1498766668248498e-05,
+      "loss": 1.0094,
+      "step": 5677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1490831294384309e-05,
+      "loss": 0.9733,
+      "step": 5678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1482897842472573e-05,
+      "loss": 0.8094,
+      "step": 5679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1474966313641999e-05,
+      "loss": 0.9987,
+      "step": 5680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1467036709021003e-05,
+      "loss": 0.9659,
+      "step": 5681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.145910902973771e-05,
+      "loss": 0.979,
+      "step": 5682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1451183276920032e-05,
+      "loss": 0.8408,
+      "step": 5683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1443259451695546e-05,
+      "loss": 0.8871,
+      "step": 5684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1435337555191569e-05,
+      "loss": 0.8716,
+      "step": 5685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.142741758853517e-05,
+      "loss": 1.1685,
+      "step": 5686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1419499552853113e-05,
+      "loss": 0.9974,
+      "step": 5687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1411583449271892e-05,
+      "loss": 0.7463,
+      "step": 5688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.140366927891775e-05,
+      "loss": 1.1022,
+      "step": 5689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.139575704291662e-05,
+      "loss": 0.9703,
+      "step": 5690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1387846742394192e-05,
+      "loss": 1.1995,
+      "step": 5691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1379938378475866e-05,
+      "loss": 1.1516,
+      "step": 5692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1372031952286749e-05,
+      "loss": 0.9674,
+      "step": 5693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.136412746495171e-05,
+      "loss": 0.9724,
+      "step": 5694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1356224917595317e-05,
+      "loss": 0.8617,
+      "step": 5695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1348324311341846e-05,
+      "loss": 0.7471,
+      "step": 5696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1340425647315352e-05,
+      "loss": 1.1322,
+      "step": 5697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1332528926639569e-05,
+      "loss": 1.0471,
+      "step": 5698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1324634150437949e-05,
+      "loss": 1.0001,
+      "step": 5699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1316741319833705e-05,
+      "loss": 1.0703,
+      "step": 5700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1308850435949741e-05,
+      "loss": 1.0152,
+      "step": 5701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1300961499908689e-05,
+      "loss": 0.9009,
+      "step": 5702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1293074512832924e-05,
+      "loss": 0.701,
+      "step": 5703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1285189475844512e-05,
+      "loss": 0.7483,
+      "step": 5704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.1277306390065278e-05,
+      "loss": 0.6803,
+      "step": 5705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1269425256616742e-05,
+      "loss": 0.7748,
+      "step": 5706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1261546076620144e-05,
+      "loss": 1.1506,
+      "step": 5707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1253668851196478e-05,
+      "loss": 1.0564,
+      "step": 5708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1245793581466423e-05,
+      "loss": 1.1824,
+      "step": 5709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1237920268550386e-05,
+      "loss": 0.866,
+      "step": 5710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1230048913568533e-05,
+      "loss": 0.9143,
+      "step": 5711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1222179517640713e-05,
+      "loss": 1.2649,
+      "step": 5712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1214312081886492e-05,
+      "loss": 1.0568,
+      "step": 5713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.120644660742519e-05,
+      "loss": 0.7899,
+      "step": 5714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1198583095375822e-05,
+      "loss": 1.47,
+      "step": 5715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1190721546857124e-05,
+      "loss": 1.0898,
+      "step": 5716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.118286196298758e-05,
+      "loss": 0.9772,
+      "step": 5717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1175004344885351e-05,
+      "loss": 1.0654,
+      "step": 5718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1167148693668364e-05,
+      "loss": 0.9697,
+      "step": 5719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1159295010454234e-05,
+      "loss": 0.9523,
+      "step": 5720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.11514432963603e-05,
+      "loss": 0.9998,
+      "step": 5721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1143593552503639e-05,
+      "loss": 1.0454,
+      "step": 5722
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1135745780001025e-05,
+      "loss": 0.938,
+      "step": 5723
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1127899979968967e-05,
+      "loss": 0.9774,
+      "step": 5724
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1120056153523697e-05,
+      "loss": 1.0209,
+      "step": 5725
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1112214301781154e-05,
+      "loss": 0.9329,
+      "step": 5726
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1104374425856984e-05,
+      "loss": 1.1193,
+      "step": 5727
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1096536526866594e-05,
+      "loss": 0.8596,
+      "step": 5728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1088700605925067e-05,
+      "loss": 1.1964,
+      "step": 5729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1080866664147215e-05,
+      "loss": 0.8473,
+      "step": 5730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1073034702647594e-05,
+      "loss": 0.7586,
+      "step": 5731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.106520472254044e-05,
+      "loss": 0.8287,
+      "step": 5732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1057376724939744e-05,
+      "loss": 0.9801,
+      "step": 5733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.104955071095919e-05,
+      "loss": 1.2256,
+      "step": 5734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1041726681712173e-05,
+      "loss": 0.7784,
+      "step": 5735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1033904638311842e-05,
+      "loss": 1.093,
+      "step": 5736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1026084581871018e-05,
+      "loss": 0.8969,
+      "step": 5737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1018266513502279e-05,
+      "loss": 1.1862,
+      "step": 5738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1010450434317904e-05,
+      "loss": 1.0359,
+      "step": 5739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.1002636345429884e-05,
+      "loss": 1.0493,
+      "step": 5740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0994824247949923e-05,
+      "loss": 1.1304,
+      "step": 5741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0987014142989463e-05,
+      "loss": 0.9912,
+      "step": 5742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0979206031659645e-05,
+      "loss": 1.1011,
+      "step": 5743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.097139991507132e-05,
+      "loss": 1.1371,
+      "step": 5744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0963595794335083e-05,
+      "loss": 1.1295,
+      "step": 5745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0955793670561212e-05,
+      "loss": 1.1198,
+      "step": 5746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0947993544859734e-05,
+      "loss": 1.3583,
+      "step": 5747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0940195418340366e-05,
+      "loss": 0.9359,
+      "step": 5748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0932399292112539e-05,
+      "loss": 0.8957,
+      "step": 5749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0924605167285427e-05,
+      "loss": 1.1343,
+      "step": 5750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0916813044967886e-05,
+      "loss": 0.9047,
+      "step": 5751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0909022926268514e-05,
+      "loss": 0.9719,
+      "step": 5752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0901234812295616e-05,
+      "loss": 1.0026,
+      "step": 5753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0893448704157204e-05,
+      "loss": 1.0216,
+      "step": 5754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0885664602961e-05,
+      "loss": 0.783,
+      "step": 5755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0877882509814466e-05,
+      "loss": 0.7902,
+      "step": 5756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0870102425824755e-05,
+      "loss": 1.073,
+      "step": 5757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0862324352098731e-05,
+      "loss": 0.9658,
+      "step": 5758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0854548289743002e-05,
+      "loss": 0.7671,
+      "step": 5759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0846774239863852e-05,
+      "loss": 0.9913,
+      "step": 5760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0839002203567311e-05,
+      "loss": 1.0028,
+      "step": 5761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0831232181959103e-05,
+      "loss": 0.9844,
+      "step": 5762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0823464176144663e-05,
+      "loss": 1.0145,
+      "step": 5763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.081569818722915e-05,
+      "loss": 0.8915,
+      "step": 5764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0807934216317451e-05,
+      "loss": 0.9331,
+      "step": 5765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0800172264514119e-05,
+      "loss": 0.9625,
+      "step": 5766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0792412332923477e-05,
+      "loss": 0.7172,
+      "step": 5767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0784654422649514e-05,
+      "loss": 1.2715,
+      "step": 5768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0776898534795948e-05,
+      "loss": 0.8872,
+      "step": 5769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0769144670466224e-05,
+      "loss": 0.963,
+      "step": 5770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0761392830763484e-05,
+      "loss": 1.1038,
+      "step": 5771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0753643016790566e-05,
+      "loss": 1.2422,
+      "step": 5772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.074589522965006e-05,
+      "loss": 0.6876,
+      "step": 5773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0738149470444229e-05,
+      "loss": 0.9958,
+      "step": 5774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0730405740275076e-05,
+      "loss": 0.9741,
+      "step": 5775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0722664040244302e-05,
+      "loss": 0.8137,
+      "step": 5776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0714924371453308e-05,
+      "loss": 1.1173,
+      "step": 5777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0707186735003224e-05,
+      "loss": 0.7644,
+      "step": 5778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.06994511319949e-05,
+      "loss": 1.1377,
+      "step": 5779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0691717563528861e-05,
+      "loss": 0.9712,
+      "step": 5780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0683986030705381e-05,
+      "loss": 0.8822,
+      "step": 5781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.067625653462442e-05,
+      "loss": 0.9269,
+      "step": 5782
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0668529076385644e-05,
+      "loss": 0.8038,
+      "step": 5783
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0660803657088461e-05,
+      "loss": 0.8909,
+      "step": 5784
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0653080277831948e-05,
+      "loss": 0.9006,
+      "step": 5785
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0645358939714928e-05,
+      "loss": 0.891,
+      "step": 5786
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0637639643835911e-05,
+      "loss": 0.7549,
+      "step": 5787
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.0629922391293118e-05,
+      "loss": 0.9587,
+      "step": 5788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0622207183184494e-05,
+      "loss": 1.1095,
+      "step": 5789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.061449402060768e-05,
+      "loss": 0.6231,
+      "step": 5790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0606782904660006e-05,
+      "loss": 1.0316,
+      "step": 5791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0599073836438577e-05,
+      "loss": 0.8577,
+      "step": 5792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0591366817040144e-05,
+      "loss": 0.9456,
+      "step": 5793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0583661847561174e-05,
+      "loss": 0.7722,
+      "step": 5794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0575958929097877e-05,
+      "loss": 0.7977,
+      "step": 5795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0568258062746134e-05,
+      "loss": 0.9696,
+      "step": 5796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0560559249601546e-05,
+      "loss": 0.9822,
+      "step": 5797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.055286249075944e-05,
+      "loss": 0.7591,
+      "step": 5798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0545167787314819e-05,
+      "loss": 0.8198,
+      "step": 5799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0537475140362426e-05,
+      "loss": 1.046,
+      "step": 5800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.052978455099669e-05,
+      "loss": 0.871,
+      "step": 5801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0522096020311741e-05,
+      "loss": 0.9418,
+      "step": 5802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.051440954940145e-05,
+      "loss": 1.2445,
+      "step": 5803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0506725139359358e-05,
+      "loss": 0.8959,
+      "step": 5804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0499042791278713e-05,
+      "loss": 0.8263,
+      "step": 5805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.049136250625252e-05,
+      "loss": 0.8,
+      "step": 5806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0483684285373435e-05,
+      "loss": 1.2355,
+      "step": 5807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0476008129733836e-05,
+      "loss": 0.8966,
+      "step": 5808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0468334040425822e-05,
+      "loss": 0.9552,
+      "step": 5809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0460662018541188e-05,
+      "loss": 1.2494,
+      "step": 5810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.045299206517142e-05,
+      "loss": 0.9773,
+      "step": 5811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0445324181407739e-05,
+      "loss": 0.8154,
+      "step": 5812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.043765836834104e-05,
+      "loss": 0.9767,
+      "step": 5813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0429994627061964e-05,
+      "loss": 0.9952,
+      "step": 5814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0422332958660818e-05,
+      "loss": 1.0836,
+      "step": 5815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0414673364227623e-05,
+      "loss": 0.6875,
+      "step": 5816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0407015844852127e-05,
+      "loss": 0.6583,
+      "step": 5817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0399360401623762e-05,
+      "loss": 0.9583,
+      "step": 5818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0391707035631645e-05,
+      "loss": 0.9049,
+      "step": 5819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0384055747964663e-05,
+      "loss": 1.3317,
+      "step": 5820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.037640653971135e-05,
+      "loss": 0.8401,
+      "step": 5821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0368759411959947e-05,
+      "loss": 0.9653,
+      "step": 5822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0361114365798432e-05,
+      "loss": 0.9825,
+      "step": 5823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.035347140231446e-05,
+      "loss": 0.8591,
+      "step": 5824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0345830522595387e-05,
+      "loss": 1.0088,
+      "step": 5825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.03381917277283e-05,
+      "loss": 1.4204,
+      "step": 5826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0330555018799956e-05,
+      "loss": 0.8679,
+      "step": 5827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.032292039689685e-05,
+      "loss": 0.8062,
+      "step": 5828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0315287863105147e-05,
+      "loss": 0.9075,
+      "step": 5829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0307657418510724e-05,
+      "loss": 0.595,
+      "step": 5830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0300029064199184e-05,
+      "loss": 1.0406,
+      "step": 5831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0292402801255799e-05,
+      "loss": 1.005,
+      "step": 5832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0284778630765564e-05,
+      "loss": 0.965,
+      "step": 5833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0277156553813184e-05,
+      "loss": 0.9555,
+      "step": 5834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.026953657148304e-05,
+      "loss": 0.6395,
+      "step": 5835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0261918684859223e-05,
+      "loss": 0.99,
+      "step": 5836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.025430289502555e-05,
+      "loss": 0.9725,
+      "step": 5837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.024668920306551e-05,
+      "loss": 0.8693,
+      "step": 5838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0239077610062297e-05,
+      "loss": 1.0668,
+      "step": 5839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.023146811709883e-05,
+      "loss": 0.843,
+      "step": 5840
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.02238607252577e-05,
+      "loss": 0.8528,
+      "step": 5841
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0216255435621224e-05,
+      "loss": 0.9768,
+      "step": 5842
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0208652249271405e-05,
+      "loss": 0.739,
+      "step": 5843
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0201051167289938e-05,
+      "loss": 0.8907,
+      "step": 5844
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0193452190758249e-05,
+      "loss": 0.9458,
+      "step": 5845
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0185855320757425e-05,
+      "loss": 0.9,
+      "step": 5846
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.017826055836829e-05,
+      "loss": 0.879,
+      "step": 5847
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0170667904671361e-05,
+      "loss": 0.8951,
+      "step": 5848
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.016307736074683e-05,
+      "loss": 0.9136,
+      "step": 5849
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0155488927674602e-05,
+      "loss": 0.875,
+      "step": 5850
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0147902606534304e-05,
+      "loss": 1.0347,
+      "step": 5851
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.014031839840523e-05,
+      "loss": 1.2948,
+      "step": 5852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0132736304366382e-05,
+      "loss": 0.8014,
+      "step": 5853
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0125156325496485e-05,
+      "loss": 0.9979,
+      "step": 5854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0117578462873922e-05,
+      "loss": 1.0156,
+      "step": 5855
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.011000271757682e-05,
+      "loss": 1.0831,
+      "step": 5856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0102429090682969e-05,
+      "loss": 0.8616,
+      "step": 5857
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0094857583269862e-05,
+      "loss": 1.0765,
+      "step": 5858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0087288196414709e-05,
+      "loss": 1.0095,
+      "step": 5859
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.007972093119442e-05,
+      "loss": 1.0707,
+      "step": 5860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0072155788685569e-05,
+      "loss": 1.059,
+      "step": 5861
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.006459276996447e-05,
+      "loss": 1.0691,
+      "step": 5862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.005703187610711e-05,
+      "loss": 1.0531,
+      "step": 5863
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0049473108189164e-05,
+      "loss": 0.7825,
+      "step": 5864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0041916467286042e-05,
+      "loss": 1.0952,
+      "step": 5865
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0034361954472819e-05,
+      "loss": 0.6697,
+      "step": 5866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0026809570824266e-05,
+      "loss": 1.0337,
+      "step": 5867
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0019259317414884e-05,
+      "loss": 0.9008,
+      "step": 5868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.001171119531883e-05,
+      "loss": 1.0127,
+      "step": 5869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.0004165205609991e-05,
+      "loss": 1.1779,
+      "step": 5870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.996621349361935e-06,
+      "loss": 1.2712,
+      "step": 5871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.989079627647912e-06,
+      "loss": 0.7593,
+      "step": 5872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.981540041540898e-06,
+      "loss": 0.6849,
+      "step": 5873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.97400259211356e-06,
+      "loss": 0.7862,
+      "step": 5874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.966467280438235e-06,
+      "loss": 0.9255,
+      "step": 5875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.958934107586988e-06,
+      "loss": 1.1917,
+      "step": 5876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.95140307463156e-06,
+      "loss": 1.0868,
+      "step": 5877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.943874182643382e-06,
+      "loss": 0.858,
+      "step": 5878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.936347432693607e-06,
+      "loss": 1.1379,
+      "step": 5879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.928822825853062e-06,
+      "loss": 0.9551,
+      "step": 5880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.921300363192265e-06,
+      "loss": 1.0798,
+      "step": 5881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.913780045781457e-06,
+      "loss": 1.1449,
+      "step": 5882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.906261874690533e-06,
+      "loss": 1.4707,
+      "step": 5883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.89874585098913e-06,
+      "loss": 0.8794,
+      "step": 5884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.891231975746537e-06,
+      "loss": 0.8336,
+      "step": 5885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.883720250031752e-06,
+      "loss": 0.9482,
+      "step": 5886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.876210674913477e-06,
+      "loss": 0.8781,
+      "step": 5887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.868703251460112e-06,
+      "loss": 0.8439,
+      "step": 5888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.861197980739717e-06,
+      "loss": 0.9441,
+      "step": 5889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.85369486382009e-06,
+      "loss": 0.9843,
+      "step": 5890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.84619390176869e-06,
+      "loss": 0.8677,
+      "step": 5891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.838695095652675e-06,
+      "loss": 1.3348,
+      "step": 5892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.831198446538917e-06,
+      "loss": 0.8906,
+      "step": 5893
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.823703955493948e-06,
+      "loss": 1.0071,
+      "step": 5894
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.816211623584029e-06,
+      "loss": 0.9813,
+      "step": 5895
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.808721451875088e-06,
+      "loss": 0.8485,
+      "step": 5896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.801233441432744e-06,
+      "loss": 0.955,
+      "step": 5897
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.793747593322337e-06,
+      "loss": 1.183,
+      "step": 5898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.78626390860887e-06,
+      "loss": 0.9926,
+      "step": 5899
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.778782388357028e-06,
+      "loss": 1.0184,
+      "step": 5900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.771303033631249e-06,
+      "loss": 0.7809,
+      "step": 5901
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.7638258454956e-06,
+      "loss": 0.9288,
+      "step": 5902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.756350825013857e-06,
+      "loss": 0.6547,
+      "step": 5903
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.748877973249513e-06,
+      "loss": 1.0154,
+      "step": 5904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.741407291265713e-06,
+      "loss": 0.7977,
+      "step": 5905
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.733938780125316e-06,
+      "loss": 0.5981,
+      "step": 5906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.72647244089088e-06,
+      "loss": 0.8425,
+      "step": 5907
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.719008274624624e-06,
+      "loss": 1.0809,
+      "step": 5908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.7115462823885e-06,
+      "loss": 0.9913,
+      "step": 5909
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.704086465244114e-06,
+      "loss": 0.9978,
+      "step": 5910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.696628824252765e-06,
+      "loss": 0.7822,
+      "step": 5911
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.689173360475475e-06,
+      "loss": 0.8546,
+      "step": 5912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.681720074972928e-06,
+      "loss": 0.8896,
+      "step": 5913
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.67426896880548e-06,
+      "loss": 1.0262,
+      "step": 5914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.66682004303324e-06,
+      "loss": 1.2815,
+      "step": 5915
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.659373298715951e-06,
+      "loss": 0.8904,
+      "step": 5916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.651928736913057e-06,
+      "loss": 1.2275,
+      "step": 5917
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.64448635868371e-06,
+      "loss": 1.073,
+      "step": 5918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.637046165086734e-06,
+      "loss": 0.7579,
+      "step": 5919
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.629608157180633e-06,
+      "loss": 0.8008,
+      "step": 5920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.622172336023638e-06,
+      "loss": 0.7543,
+      "step": 5921
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.614738702673625e-06,
+      "loss": 0.9026,
+      "step": 5922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.607307258188195e-06,
+      "loss": 0.8597,
+      "step": 5923
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.599878003624612e-06,
+      "loss": 0.8442,
+      "step": 5924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.592450940039833e-06,
+      "loss": 1.0176,
+      "step": 5925
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.585026068490518e-06,
+      "loss": 0.7849,
+      "step": 5926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.577603390032996e-06,
+      "loss": 1.2612,
+      "step": 5927
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.570182905723299e-06,
+      "loss": 0.9093,
+      "step": 5928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.562764616617145e-06,
+      "loss": 0.8987,
+      "step": 5929
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.555348523769936e-06,
+      "loss": 0.8215,
+      "step": 5930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.547934628236743e-06,
+      "loss": 0.7307,
+      "step": 5931
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.540522931072366e-06,
+      "loss": 1.0651,
+      "step": 5932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.533113433331259e-06,
+      "loss": 0.946,
+      "step": 5933
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.525706136067563e-06,
+      "loss": 1.0729,
+      "step": 5934
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.518301040335134e-06,
+      "loss": 0.8564,
+      "step": 5935
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.510898147187477e-06,
+      "loss": 0.993,
+      "step": 5936
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.503497457677826e-06,
+      "loss": 0.8457,
+      "step": 5937
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.496098972859066e-06,
+      "loss": 0.7941,
+      "step": 5938
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.488702693783774e-06,
+      "loss": 1.1532,
+      "step": 5939
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.481308621504234e-06,
+      "loss": 0.7034,
+      "step": 5940
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.47391675707239e-06,
+      "loss": 1.0061,
+      "step": 5941
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.466527101539891e-06,
+      "loss": 0.9567,
+      "step": 5942
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.459139655958075e-06,
+      "loss": 1.0137,
+      "step": 5943
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.451754421377943e-06,
+      "loss": 0.9321,
+      "step": 5944
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.444371398850186e-06,
+      "loss": 1.0107,
+      "step": 5945
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.436990589425209e-06,
+      "loss": 1.0029,
+      "step": 5946
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.42961199415307e-06,
+      "loss": 0.8551,
+      "step": 5947
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.422235614083517e-06,
+      "loss": 0.7588,
+      "step": 5948
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.414861450266005e-06,
+      "loss": 0.6599,
+      "step": 5949
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.40748950374964e-06,
+      "loss": 1.0721,
+      "step": 5950
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.400119775583249e-06,
+      "loss": 0.9122,
+      "step": 5951
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.392752266815317e-06,
+      "loss": 0.8754,
+      "step": 5952
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.385386978494009e-06,
+      "loss": 0.8389,
+      "step": 5953
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.378023911667208e-06,
+      "loss": 0.6675,
+      "step": 5954
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.370663067382437e-06,
+      "loss": 0.8344,
+      "step": 5955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.363304446686941e-06,
+      "loss": 1.1754,
+      "step": 5956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.355948050627633e-06,
+      "loss": 0.7792,
+      "step": 5957
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.348593880251106e-06,
+      "loss": 0.8726,
+      "step": 5958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.341241936603625e-06,
+      "loss": 0.984,
+      "step": 5959
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.333892220731177e-06,
+      "loss": 0.9508,
+      "step": 5960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.326544733679397e-06,
+      "loss": 1.0145,
+      "step": 5961
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.319199476493602e-06,
+      "loss": 0.9133,
+      "step": 5962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.311856450218822e-06,
+      "loss": 0.7068,
+      "step": 5963
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.304515655899737e-06,
+      "loss": 1.0614,
+      "step": 5964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.297177094580736e-06,
+      "loss": 0.7043,
+      "step": 5965
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.289840767305871e-06,
+      "loss": 1.0797,
+      "step": 5966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.282506675118877e-06,
+      "loss": 0.7187,
+      "step": 5967
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.275174819063182e-06,
+      "loss": 0.8602,
+      "step": 5968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.2678452001819e-06,
+      "loss": 1.1384,
+      "step": 5969
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.2605178195178e-06,
+      "loss": 1.2527,
+      "step": 5970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.253192678113367e-06,
+      "loss": 1.1818,
+      "step": 5971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.245869777010745e-06,
+      "loss": 1.0863,
+      "step": 5972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.238549117251754e-06,
+      "loss": 0.9584,
+      "step": 5973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.231230699877921e-06,
+      "loss": 1.0925,
+      "step": 5974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.223914525930436e-06,
+      "loss": 0.9551,
+      "step": 5975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.216600596450158e-06,
+      "loss": 0.689,
+      "step": 5976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.209288912477663e-06,
+      "loss": 1.1078,
+      "step": 5977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.201979475053165e-06,
+      "loss": 0.9324,
+      "step": 5978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.194672285216602e-06,
+      "loss": 1.3727,
+      "step": 5979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.187367344007555e-06,
+      "loss": 1.2574,
+      "step": 5980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.180064652465292e-06,
+      "loss": 1.1257,
+      "step": 5981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.172764211628781e-06,
+      "loss": 0.7699,
+      "step": 5982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.165466022536661e-06,
+      "loss": 0.7607,
+      "step": 5983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.158170086227235e-06,
+      "loss": 0.9553,
+      "step": 5984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.150876403738512e-06,
+      "loss": 0.7679,
+      "step": 5985
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.143584976108156e-06,
+      "loss": 0.9052,
+      "step": 5986
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.136295804373512e-06,
+      "loss": 0.8336,
+      "step": 5987
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.129008889571633e-06,
+      "loss": 0.9395,
+      "step": 5988
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.121724232739207e-06,
+      "loss": 1.1277,
+      "step": 5989
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.114441834912644e-06,
+      "loss": 1.0759,
+      "step": 5990
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.107161697128005e-06,
+      "loss": 0.8184,
+      "step": 5991
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.099883820421026e-06,
+      "loss": 0.949,
+      "step": 5992
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.09260820582715e-06,
+      "loss": 0.8911,
+      "step": 5993
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.085334854381472e-06,
+      "loss": 0.9484,
+      "step": 5994
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.078063767118756e-06,
+      "loss": 0.9032,
+      "step": 5995
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.070794945073494e-06,
+      "loss": 0.5798,
+      "step": 5996
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.06352838927981e-06,
+      "loss": 1.0164,
+      "step": 5997
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.056264100771506e-06,
+      "loss": 1.1224,
+      "step": 5998
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.049002080582091e-06,
+      "loss": 0.811,
+      "step": 5999
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.041742329744729e-06,
+      "loss": 0.9788,
+      "step": 6000
+    },
+    {
+      "epoch": 0.72,
+      "eval_loss": 0.99658203125,
+      "eval_runtime": 3.5953,
+      "eval_samples_per_second": 27.814,
+      "eval_steps_per_second": 4.728,
+      "step": 6000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.034484849292258e-06,
+      "loss": 1.1228,
+      "step": 6001
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.027229640257212e-06,
+      "loss": 0.7657,
+      "step": 6002
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.019976703671782e-06,
+      "loss": 0.7806,
+      "step": 6003
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.012726040567856e-06,
+      "loss": 1.1221,
+      "step": 6004
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.005477651976981e-06,
+      "loss": 1.0441,
+      "step": 6005
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.998231538930382e-06,
+      "loss": 1.1293,
+      "step": 6006
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.990987702458977e-06,
+      "loss": 0.9325,
+      "step": 6007
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.983746143593341e-06,
+      "loss": 0.9533,
+      "step": 6008
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.976506863363718e-06,
+      "loss": 1.0502,
+      "step": 6009
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.96926986280007e-06,
+      "loss": 1.0546,
+      "step": 6010
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.962035142931996e-06,
+      "loss": 0.772,
+      "step": 6011
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.95480270478877e-06,
+      "loss": 1.1056,
+      "step": 6012
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.947572549399372e-06,
+      "loss": 0.8082,
+      "step": 6013
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.940344677792426e-06,
+      "loss": 1.0864,
+      "step": 6014
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.933119090996234e-06,
+      "loss": 1.2655,
+      "step": 6015
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.925895790038802e-06,
+      "loss": 0.8019,
+      "step": 6016
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.918674775947773e-06,
+      "loss": 1.2166,
+      "step": 6017
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.911456049750494e-06,
+      "loss": 1.0117,
+      "step": 6018
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.904239612473975e-06,
+      "loss": 0.8915,
+      "step": 6019
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.897025465144887e-06,
+      "loss": 0.9546,
+      "step": 6020
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.889813608789604e-06,
+      "loss": 0.9492,
+      "step": 6021
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.882604044434154e-06,
+      "loss": 0.7465,
+      "step": 6022
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.875396773104222e-06,
+      "loss": 0.6626,
+      "step": 6023
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.868191795825223e-06,
+      "loss": 0.649,
+      "step": 6024
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.860989113622198e-06,
+      "loss": 0.8636,
+      "step": 6025
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.85378872751986e-06,
+      "loss": 0.7814,
+      "step": 6026
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.846590638542632e-06,
+      "loss": 1.0883,
+      "step": 6027
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.839394847714572e-06,
+      "loss": 1.011,
+      "step": 6028
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.832201356059425e-06,
+      "loss": 0.8151,
+      "step": 6029
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.825010164600625e-06,
+      "loss": 0.7263,
+      "step": 6030
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.817821274361249e-06,
+      "loss": 1.0436,
+      "step": 6031
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.810634686364078e-06,
+      "loss": 0.9856,
+      "step": 6032
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.803450401631539e-06,
+      "loss": 0.8382,
+      "step": 6033
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.796268421185735e-06,
+      "loss": 0.7944,
+      "step": 6034
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.789088746048465e-06,
+      "loss": 1.0104,
+      "step": 6035
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.781911377241168e-06,
+      "loss": 0.9115,
+      "step": 6036
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.774736315784973e-06,
+      "loss": 1.0238,
+      "step": 6037
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.767563562700693e-06,
+      "loss": 0.6687,
+      "step": 6038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.76039311900878e-06,
+      "loss": 0.7845,
+      "step": 6039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.753224985729375e-06,
+      "loss": 0.7794,
+      "step": 6040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.746059163882304e-06,
+      "loss": 1.183,
+      "step": 6041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.738895654487039e-06,
+      "loss": 0.9989,
+      "step": 6042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.731734458562727e-06,
+      "loss": 0.9646,
+      "step": 6043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.72457557712821e-06,
+      "loss": 0.8101,
+      "step": 6044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.71741901120197e-06,
+      "loss": 0.8004,
+      "step": 6045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.710264761802187e-06,
+      "loss": 0.9086,
+      "step": 6046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.703112829946686e-06,
+      "loss": 0.8192,
+      "step": 6047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.695963216652969e-06,
+      "loss": 1.155,
+      "step": 6048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.688815922938231e-06,
+      "loss": 0.9818,
+      "step": 6049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.681670949819304e-06,
+      "loss": 1.1752,
+      "step": 6050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.674528298312706e-06,
+      "loss": 0.9388,
+      "step": 6051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.66738796943464e-06,
+      "loss": 1.1808,
+      "step": 6052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.660249964200948e-06,
+      "loss": 0.948,
+      "step": 6053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.653114283627151e-06,
+      "loss": 0.8378,
+      "step": 6054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.645980928728455e-06,
+      "loss": 0.8819,
+      "step": 6055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.638849900519724e-06,
+      "loss": 1.01,
+      "step": 6056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.631721200015477e-06,
+      "loss": 0.7138,
+      "step": 6057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.624594828229935e-06,
+      "loss": 0.8914,
+      "step": 6058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.617470786176952e-06,
+      "loss": 0.9933,
+      "step": 6059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.610349074870078e-06,
+      "loss": 0.8522,
+      "step": 6060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.603229695322518e-06,
+      "loss": 1.0883,
+      "step": 6061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.596112648547139e-06,
+      "loss": 0.7976,
+      "step": 6062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.5889979355565e-06,
+      "loss": 0.7808,
+      "step": 6063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.581885557362798e-06,
+      "loss": 0.7755,
+      "step": 6064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.574775514977917e-06,
+      "loss": 0.6423,
+      "step": 6065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.567667809413418e-06,
+      "loss": 0.9211,
+      "step": 6066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.560562441680502e-06,
+      "loss": 0.7554,
+      "step": 6067
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.55345941279005e-06,
+      "loss": 0.9917,
+      "step": 6068
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.546358723752621e-06,
+      "loss": 0.9473,
+      "step": 6069
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.539260375578426e-06,
+      "loss": 0.7826,
+      "step": 6070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.53216436927734e-06,
+      "loss": 0.9611,
+      "step": 6071
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.525070705858929e-06,
+      "loss": 0.9584,
+      "step": 6072
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.517979386332398e-06,
+      "loss": 1.257,
+      "step": 6073
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.510890411706643e-06,
+      "loss": 0.8735,
+      "step": 6074
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.503803782990207e-06,
+      "loss": 0.9714,
+      "step": 6075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.496719501191297e-06,
+      "loss": 1.0205,
+      "step": 6076
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.489637567317806e-06,
+      "loss": 1.0796,
+      "step": 6077
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.48255798237729e-06,
+      "loss": 0.8777,
+      "step": 6078
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.475480747376943e-06,
+      "loss": 0.9802,
+      "step": 6079
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.468405863323665e-06,
+      "loss": 0.9519,
+      "step": 6080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.461333331223992e-06,
+      "loss": 0.899,
+      "step": 6081
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.454263152084127e-06,
+      "loss": 0.8263,
+      "step": 6082
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.447195326909963e-06,
+      "loss": 0.7782,
+      "step": 6083
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.44012985670702e-06,
+      "loss": 0.6943,
+      "step": 6084
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.433066742480527e-06,
+      "loss": 0.8815,
+      "step": 6085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.426005985235343e-06,
+      "loss": 0.8442,
+      "step": 6086
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.418947585975997e-06,
+      "loss": 0.8746,
+      "step": 6087
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.411891545706704e-06,
+      "loss": 1.0825,
+      "step": 6088
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.404837865431322e-06,
+      "loss": 0.8759,
+      "step": 6089
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.39778654615337e-06,
+      "loss": 0.8313,
+      "step": 6090
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.390737588876047e-06,
+      "loss": 0.9404,
+      "step": 6091
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.383690994602223e-06,
+      "loss": 0.9454,
+      "step": 6092
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.376646764334394e-06,
+      "loss": 1.0037,
+      "step": 6093
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.36960489907477e-06,
+      "loss": 0.9862,
+      "step": 6094
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.362565399825187e-06,
+      "loss": 1.0579,
+      "step": 6095
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.355528267587145e-06,
+      "loss": 0.8595,
+      "step": 6096
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.34849350336184e-06,
+      "loss": 1.2039,
+      "step": 6097
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.34146110815009e-06,
+      "loss": 0.9252,
+      "step": 6098
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.334431082952409e-06,
+      "loss": 0.8389,
+      "step": 6099
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.327403428768956e-06,
+      "loss": 0.8959,
+      "step": 6100
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.32037814659955e-06,
+      "loss": 0.796,
+      "step": 6101
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.313355237443694e-06,
+      "loss": 0.9526,
+      "step": 6102
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.306334702300528e-06,
+      "loss": 1.066,
+      "step": 6103
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.29931654216885e-06,
+      "loss": 0.965,
+      "step": 6104
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.292300758047173e-06,
+      "loss": 1.0579,
+      "step": 6105
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.285287350933613e-06,
+      "loss": 1.0649,
+      "step": 6106
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.27827632182596e-06,
+      "loss": 0.9828,
+      "step": 6107
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.271267671721694e-06,
+      "loss": 0.8408,
+      "step": 6108
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.264261401617926e-06,
+      "loss": 0.8633,
+      "step": 6109
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.257257512511436e-06,
+      "loss": 0.5751,
+      "step": 6110
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.250256005398679e-06,
+      "loss": 0.8823,
+      "step": 6111
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.24325688127575e-06,
+      "loss": 0.9619,
+      "step": 6112
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.236260141138434e-06,
+      "loss": 0.9592,
+      "step": 6113
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.229265785982143e-06,
+      "loss": 1.0867,
+      "step": 6114
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.222273816801962e-06,
+      "loss": 0.9835,
+      "step": 6115
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.215284234592658e-06,
+      "loss": 0.9301,
+      "step": 6116
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.208297040348625e-06,
+      "loss": 0.9229,
+      "step": 6117
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.201312235063927e-06,
+      "loss": 0.8066,
+      "step": 6118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.194329819732316e-06,
+      "loss": 1.1401,
+      "step": 6119
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.187349795347174e-06,
+      "loss": 1.0227,
+      "step": 6120
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.180372162901536e-06,
+      "loss": 0.9446,
+      "step": 6121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.173396923388132e-06,
+      "loss": 1.0748,
+      "step": 6122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.16642407779932e-06,
+      "loss": 0.9393,
+      "step": 6123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.159453627127123e-06,
+      "loss": 0.8495,
+      "step": 6124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.152485572363242e-06,
+      "loss": 1.155,
+      "step": 6125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.145519914499009e-06,
+      "loss": 1.1508,
+      "step": 6126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.138556654525445e-06,
+      "loss": 0.964,
+      "step": 6127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.131595793433209e-06,
+      "loss": 0.9299,
+      "step": 6128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.124637332212612e-06,
+      "loss": 1.0568,
+      "step": 6129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.117681271853652e-06,
+      "loss": 1.0239,
+      "step": 6130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.110727613345967e-06,
+      "loss": 0.9144,
+      "step": 6131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.103776357678835e-06,
+      "loss": 1.3105,
+      "step": 6132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.096827505841245e-06,
+      "loss": 1.0078,
+      "step": 6133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.0898810588218e-06,
+      "loss": 0.9127,
+      "step": 6134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.082937017608758e-06,
+      "loss": 0.9835,
+      "step": 6135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.075995383190067e-06,
+      "loss": 0.8721,
+      "step": 6136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.069056156553308e-06,
+      "loss": 0.9703,
+      "step": 6137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.06211933868572e-06,
+      "loss": 0.9392,
+      "step": 6138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.05518493057422e-06,
+      "loss": 1.1613,
+      "step": 6139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.048252933205352e-06,
+      "loss": 0.8058,
+      "step": 6140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.04132334756535e-06,
+      "loss": 0.7271,
+      "step": 6141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.034396174640074e-06,
+      "loss": 1.0668,
+      "step": 6142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.027471415415055e-06,
+      "loss": 0.8784,
+      "step": 6143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.020549070875492e-06,
+      "loss": 0.6448,
+      "step": 6144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.01362914200621e-06,
+      "loss": 0.9568,
+      "step": 6145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.00671162979172e-06,
+      "loss": 0.8802,
+      "step": 6146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.999796535216184e-06,
+      "loss": 1.0955,
+      "step": 6147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.992883859263408e-06,
+      "loss": 0.8018,
+      "step": 6148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.985973602916855e-06,
+      "loss": 0.8011,
+      "step": 6149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.979065767159657e-06,
+      "loss": 0.9028,
+      "step": 6150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.97216035297459e-06,
+      "loss": 0.6978,
+      "step": 6151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.965257361344081e-06,
+      "loss": 0.9332,
+      "step": 6152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.958356793250236e-06,
+      "loss": 0.8424,
+      "step": 6153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.95145864967478e-06,
+      "loss": 0.8444,
+      "step": 6154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.944562931599137e-06,
+      "loss": 0.7462,
+      "step": 6155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.937669640004346e-06,
+      "loss": 0.6997,
+      "step": 6156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.930778775871118e-06,
+      "loss": 0.9188,
+      "step": 6157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.923890340179829e-06,
+      "loss": 0.8281,
+      "step": 6158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.91700433391048e-06,
+      "loss": 0.933,
+      "step": 6159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.910120758042758e-06,
+      "loss": 0.7556,
+      "step": 6160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.903239613555995e-06,
+      "loss": 0.9618,
+      "step": 6161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.896360901429168e-06,
+      "loss": 1.2708,
+      "step": 6162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.8894846226409e-06,
+      "loss": 0.6755,
+      "step": 6163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.882610778169503e-06,
+      "loss": 1.0104,
+      "step": 6164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.875739368992907e-06,
+      "loss": 1.001,
+      "step": 6165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.868870396088705e-06,
+      "loss": 0.7372,
+      "step": 6166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.862003860434159e-06,
+      "loss": 0.8049,
+      "step": 6167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.855139763006158e-06,
+      "loss": 0.8487,
+      "step": 6168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.848278104781278e-06,
+      "loss": 0.9176,
+      "step": 6169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.84141888673572e-06,
+      "loss": 0.8778,
+      "step": 6170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.834562109845334e-06,
+      "loss": 1.3883,
+      "step": 6171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.827707775085649e-06,
+      "loss": 1.075,
+      "step": 6172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.820855883431835e-06,
+      "loss": 1.0962,
+      "step": 6173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.8140064358587e-06,
+      "loss": 1.285,
+      "step": 6174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.807159433340733e-06,
+      "loss": 0.9849,
+      "step": 6175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.800314876852046e-06,
+      "loss": 0.7445,
+      "step": 6176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.793472767366411e-06,
+      "loss": 1.0079,
+      "step": 6177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.786633105857275e-06,
+      "loss": 0.7993,
+      "step": 6178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.779795893297698e-06,
+      "loss": 0.6902,
+      "step": 6179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.772961130660428e-06,
+      "loss": 0.8328,
+      "step": 6180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.76612881891784e-06,
+      "loss": 0.8353,
+      "step": 6181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.759298959041963e-06,
+      "loss": 1.1953,
+      "step": 6182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.752471552004497e-06,
+      "loss": 1.1129,
+      "step": 6183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.745646598776774e-06,
+      "loss": 1.1691,
+      "step": 6184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.738824100329767e-06,
+      "loss": 0.9314,
+      "step": 6185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.732004057634127e-06,
+      "loss": 0.9652,
+      "step": 6186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.725186471660151e-06,
+      "loss": 0.9442,
+      "step": 6187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.718371343377761e-06,
+      "loss": 1.0728,
+      "step": 6188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.711558673756563e-06,
+      "loss": 0.8817,
+      "step": 6189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.704748463765791e-06,
+      "loss": 1.0781,
+      "step": 6190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.697940714374324e-06,
+      "loss": 0.9196,
+      "step": 6191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.691135426550723e-06,
+      "loss": 1.0194,
+      "step": 6192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.684332601263152e-06,
+      "loss": 0.588,
+      "step": 6193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.677532239479479e-06,
+      "loss": 1.0367,
+      "step": 6194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.670734342167178e-06,
+      "loss": 0.9727,
+      "step": 6195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.663938910293378e-06,
+      "loss": 0.7659,
+      "step": 6196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.657145944824884e-06,
+      "loss": 0.8705,
+      "step": 6197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.650355446728128e-06,
+      "loss": 0.8804,
+      "step": 6198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.643567416969186e-06,
+      "loss": 0.7598,
+      "step": 6199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.636781856513796e-06,
+      "loss": 0.9608,
+      "step": 6200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.629998766327356e-06,
+      "loss": 0.9369,
+      "step": 6201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.623218147374881e-06,
+      "loss": 0.8422,
+      "step": 6202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.616440000621061e-06,
+      "loss": 1.1425,
+      "step": 6203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.609664327030222e-06,
+      "loss": 0.7012,
+      "step": 6204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.602891127566331e-06,
+      "loss": 0.8465,
+      "step": 6205
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.59612040319303e-06,
+      "loss": 0.8707,
+      "step": 6206
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.589352154873572e-06,
+      "loss": 1.0564,
+      "step": 6207
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.582586383570897e-06,
+      "loss": 0.8712,
+      "step": 6208
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.575823090247561e-06,
+      "loss": 0.8578,
+      "step": 6209
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.569062275865777e-06,
+      "loss": 0.9509,
+      "step": 6210
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.562303941387416e-06,
+      "loss": 0.7006,
+      "step": 6211
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.555548087773986e-06,
+      "loss": 0.7571,
+      "step": 6212
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.548794715986623e-06,
+      "loss": 0.8992,
+      "step": 6213
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5420438269861694e-06,
+      "loss": 0.8525,
+      "step": 6214
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.535295421733052e-06,
+      "loss": 1.0388,
+      "step": 6215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.528549501187365e-06,
+      "loss": 1.0309,
+      "step": 6216
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5218060663088675e-06,
+      "loss": 0.8033,
+      "step": 6217
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5150651180569386e-06,
+      "loss": 0.9095,
+      "step": 6218
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.50832665739061e-06,
+      "loss": 0.8657,
+      "step": 6219
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.501590685268575e-06,
+      "loss": 0.6106,
+      "step": 6220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.494857202649153e-06,
+      "loss": 0.6937,
+      "step": 6221
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.488126210490329e-06,
+      "loss": 1.3419,
+      "step": 6222
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.481397709749716e-06,
+      "loss": 0.9452,
+      "step": 6223
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.474671701384573e-06,
+      "loss": 0.8177,
+      "step": 6224
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.467948186351822e-06,
+      "loss": 1.1788,
+      "step": 6225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.461227165608017e-06,
+      "loss": 0.9602,
+      "step": 6226
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.45450864010934e-06,
+      "loss": 0.7946,
+      "step": 6227
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.447792610811669e-06,
+      "loss": 1.1155,
+      "step": 6228
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.4410790786704794e-06,
+      "loss": 1.1254,
+      "step": 6229
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.434368044640899e-06,
+      "loss": 1.0223,
+      "step": 6230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.427659509677723e-06,
+      "loss": 0.903,
+      "step": 6231
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.4209534747353705e-06,
+      "loss": 0.8989,
+      "step": 6232
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.414249940767901e-06,
+      "loss": 0.9255,
+      "step": 6233
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.407548908729045e-06,
+      "loss": 1.0686,
+      "step": 6234
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.400850379572141e-06,
+      "loss": 1.1421,
+      "step": 6235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.39415435425021e-06,
+      "loss": 0.9005,
+      "step": 6236
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3874608337158865e-06,
+      "loss": 1.0769,
+      "step": 6237
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.380769818921452e-06,
+      "loss": 1.2068,
+      "step": 6238
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3740813108188534e-06,
+      "loss": 0.7618,
+      "step": 6239
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.367395310359651e-06,
+      "loss": 1.1434,
+      "step": 6240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3607118184950755e-06,
+      "loss": 1.0635,
+      "step": 6241
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3540308361759904e-06,
+      "loss": 0.9575,
+      "step": 6242
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3473523643528965e-06,
+      "loss": 0.9467,
+      "step": 6243
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.340676403975933e-06,
+      "loss": 0.7395,
+      "step": 6244
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.334002955994904e-06,
+      "loss": 0.7574,
+      "step": 6245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3273320213592344e-06,
+      "loss": 1.3928,
+      "step": 6246
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3206636010179965e-06,
+      "loss": 0.9381,
+      "step": 6247
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.313997695919919e-06,
+      "loss": 0.944,
+      "step": 6248
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.30733430701335e-06,
+      "loss": 0.9968,
+      "step": 6249
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.3006734352463025e-06,
+      "loss": 0.7469,
+      "step": 6250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.294015081566416e-06,
+      "loss": 0.9083,
+      "step": 6251
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.287359246920963e-06,
+      "loss": 0.9781,
+      "step": 6252
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.280705932256895e-06,
+      "loss": 0.6562,
+      "step": 6253
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.274055138520758e-06,
+      "loss": 0.9932,
+      "step": 6254
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.267406866658768e-06,
+      "loss": 1.0526,
+      "step": 6255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.260761117616794e-06,
+      "loss": 0.6936,
+      "step": 6256
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.254117892340309e-06,
+      "loss": 1.0208,
+      "step": 6257
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.2474771917744435e-06,
+      "loss": 0.8928,
+      "step": 6258
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.24083901686399e-06,
+      "loss": 0.7806,
+      "step": 6259
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.234203368553349e-06,
+      "loss": 0.9205,
+      "step": 6260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.227570247786572e-06,
+      "loss": 1.1615,
+      "step": 6261
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.2209396555073695e-06,
+      "loss": 0.8983,
+      "step": 6262
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.214311592659062e-06,
+      "loss": 1.0084,
+      "step": 6263
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.207686060184637e-06,
+      "loss": 0.6996,
+      "step": 6264
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.201063059026711e-06,
+      "loss": 0.9883,
+      "step": 6265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.194442590127526e-06,
+      "loss": 0.9846,
+      "step": 6266
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.187824654428993e-06,
+      "loss": 1.1246,
+      "step": 6267
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.181209252872634e-06,
+      "loss": 0.7786,
+      "step": 6268
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.174596386399629e-06,
+      "loss": 0.9684,
+      "step": 6269
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.1679860559508035e-06,
+      "loss": 0.9758,
+      "step": 6270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.161378262466601e-06,
+      "loss": 1.1355,
+      "step": 6271
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.154773006887103e-06,
+      "loss": 0.9225,
+      "step": 6272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.148170290152059e-06,
+      "loss": 1.1716,
+      "step": 6273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.141570113200826e-06,
+      "loss": 1.1831,
+      "step": 6274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.1349724769724234e-06,
+      "loss": 1.0638,
+      "step": 6275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.128377382405496e-06,
+      "loss": 0.8466,
+      "step": 6276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.121784830438316e-06,
+      "loss": 0.97,
+      "step": 6277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.1151948220088245e-06,
+      "loss": 0.6085,
+      "step": 6278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.108607358054578e-06,
+      "loss": 0.6867,
+      "step": 6279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.102022439512765e-06,
+      "loss": 0.9892,
+      "step": 6280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.095440067320236e-06,
+      "loss": 1.111,
+      "step": 6281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.088860242413469e-06,
+      "loss": 0.829,
+      "step": 6282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.082282965728568e-06,
+      "loss": 0.9138,
+      "step": 6283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.075708238201293e-06,
+      "loss": 0.8513,
+      "step": 6284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.069136060767026e-06,
+      "loss": 0.9683,
+      "step": 6285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.062566434360784e-06,
+      "loss": 1.1029,
+      "step": 6286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.055999359917248e-06,
+      "loss": 0.9379,
+      "step": 6287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.049434838370697e-06,
+      "loss": 0.9126,
+      "step": 6288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.042872870655088e-06,
+      "loss": 1.1047,
+      "step": 6289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.03631345770398e-06,
+      "loss": 0.8766,
+      "step": 6290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.029756600450577e-06,
+      "loss": 1.0487,
+      "step": 6291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.023202299827742e-06,
+      "loss": 0.7487,
+      "step": 6292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.016650556767945e-06,
+      "loss": 0.928,
+      "step": 6293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.010101372203301e-06,
+      "loss": 0.9192,
+      "step": 6294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.003554747065566e-06,
+      "loss": 0.808,
+      "step": 6295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.997010682286143e-06,
+      "loss": 0.7162,
+      "step": 6296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.990469178796038e-06,
+      "loss": 0.8284,
+      "step": 6297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.983930237525932e-06,
+      "loss": 1.1399,
+      "step": 6298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.977393859406109e-06,
+      "loss": 0.7627,
+      "step": 6299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.970860045366495e-06,
+      "loss": 0.8448,
+      "step": 6300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.964328796336675e-06,
+      "loss": 0.6112,
+      "step": 6301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.957800113245832e-06,
+      "loss": 0.6932,
+      "step": 6302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.951273997022823e-06,
+      "loss": 0.9429,
+      "step": 6303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.94475044859611e-06,
+      "loss": 0.8268,
+      "step": 6304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.93822946889379e-06,
+      "loss": 0.9761,
+      "step": 6305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.931711058843624e-06,
+      "loss": 1.0011,
+      "step": 6306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.925195219372974e-06,
+      "loss": 0.9122,
+      "step": 6307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.91868195140884e-06,
+      "loss": 0.8055,
+      "step": 6308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.912171255877895e-06,
+      "loss": 0.908,
+      "step": 6309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.905663133706402e-06,
+      "loss": 0.8636,
+      "step": 6310
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.899157585820265e-06,
+      "loss": 0.9218,
+      "step": 6311
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.892654613145044e-06,
+      "loss": 1.0975,
+      "step": 6312
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.886154216605914e-06,
+      "loss": 0.9696,
+      "step": 6313
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.879656397127676e-06,
+      "loss": 1.2222,
+      "step": 6314
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.873161155634794e-06,
+      "loss": 0.9935,
+      "step": 6315
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.866668493051334e-06,
+      "loss": 0.9026,
+      "step": 6316
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.860178410301019e-06,
+      "loss": 0.9482,
+      "step": 6317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.853690908307189e-06,
+      "loss": 0.7584,
+      "step": 6318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.847205987992817e-06,
+      "loss": 0.883,
+      "step": 6319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.8407236502805284e-06,
+      "loss": 0.8932,
+      "step": 6320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.834243896092557e-06,
+      "loss": 1.255,
+      "step": 6321
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.827766726350765e-06,
+      "loss": 0.8323,
+      "step": 6322
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.82129214197669e-06,
+      "loss": 0.9221,
+      "step": 6323
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.814820143891459e-06,
+      "loss": 1.1346,
+      "step": 6324
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.808350733015839e-06,
+      "loss": 0.9506,
+      "step": 6325
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.801883910270246e-06,
+      "loss": 0.9152,
+      "step": 6326
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.79541967657471e-06,
+      "loss": 1.194,
+      "step": 6327
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.788958032848891e-06,
+      "loss": 0.7386,
+      "step": 6328
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.782498980012103e-06,
+      "loss": 0.8217,
+      "step": 6329
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.776042518983267e-06,
+      "loss": 0.9041,
+      "step": 6330
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.769588650680952e-06,
+      "loss": 1.059,
+      "step": 6331
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.763137376023352e-06,
+      "loss": 0.9663,
+      "step": 6332
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.756688695928276e-06,
+      "loss": 0.9443,
+      "step": 6333
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.750242611313199e-06,
+      "loss": 1.1068,
+      "step": 6334
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.743799123095199e-06,
+      "loss": 0.819,
+      "step": 6335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.737358232190974e-06,
+      "loss": 0.8335,
+      "step": 6336
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.730919939516905e-06,
+      "loss": 0.7639,
+      "step": 6337
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.724484245988954e-06,
+      "loss": 0.949,
+      "step": 6338
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.718051152522723e-06,
+      "loss": 0.931,
+      "step": 6339
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.711620660033457e-06,
+      "loss": 0.8716,
+      "step": 6340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.705192769436025e-06,
+      "loss": 0.6453,
+      "step": 6341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.69876748164491e-06,
+      "loss": 1.0251,
+      "step": 6342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.692344797574257e-06,
+      "loss": 1.0321,
+      "step": 6343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.68592471813781e-06,
+      "loss": 1.0586,
+      "step": 6344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.679507244248967e-06,
+      "loss": 0.9258,
+      "step": 6345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.673092376820736e-06,
+      "loss": 0.7256,
+      "step": 6346
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.666680116765755e-06,
+      "loss": 1.0452,
+      "step": 6347
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.660270464996312e-06,
+      "loss": 0.7917,
+      "step": 6348
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.653863422424297e-06,
+      "loss": 0.9743,
+      "step": 6349
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.647458989961244e-06,
+      "loss": 0.7871,
+      "step": 6350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.641057168518328e-06,
+      "loss": 1.3411,
+      "step": 6351
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.634657959006321e-06,
+      "loss": 0.9452,
+      "step": 6352
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.628261362335642e-06,
+      "loss": 0.822,
+      "step": 6353
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.621867379416344e-06,
+      "loss": 1.0593,
+      "step": 6354
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.615476011158095e-06,
+      "loss": 0.8047,
+      "step": 6355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.60908725847019e-06,
+      "loss": 0.6807,
+      "step": 6356
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.6027011222615705e-06,
+      "loss": 0.8013,
+      "step": 6357
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.596317603440783e-06,
+      "loss": 0.998,
+      "step": 6358
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.589936702916022e-06,
+      "loss": 0.7971,
+      "step": 6359
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.583558421595099e-06,
+      "loss": 0.9028,
+      "step": 6360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.577182760385436e-06,
+      "loss": 0.8097,
+      "step": 6361
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.570809720194127e-06,
+      "loss": 1.0668,
+      "step": 6362
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.564439301927841e-06,
+      "loss": 1.0228,
+      "step": 6363
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.558071506492908e-06,
+      "loss": 0.899,
+      "step": 6364
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.5517063347952875e-06,
+      "loss": 0.8411,
+      "step": 6365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.5453437877405435e-06,
+      "loss": 1.0232,
+      "step": 6366
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.538983866233869e-06,
+      "loss": 1.0022,
+      "step": 6367
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.532626571180109e-06,
+      "loss": 0.8535,
+      "step": 6368
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.5262719034837e-06,
+      "loss": 0.6579,
+      "step": 6369
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.5199198640487375e-06,
+      "loss": 0.7816,
+      "step": 6370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 6.51357045377892e-06,
+      "loss": 0.9518,
+      "step": 6371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.50722367357757e-06,
+      "loss": 0.9076,
+      "step": 6372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.500879524347667e-06,
+      "loss": 0.7747,
+      "step": 6373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.494538006991779e-06,
+      "loss": 0.8898,
+      "step": 6374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.488199122412111e-06,
+      "loss": 0.6866,
+      "step": 6375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.481862871510516e-06,
+      "loss": 0.9407,
+      "step": 6376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4755292551884335e-06,
+      "loss": 0.8505,
+      "step": 6377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4691982743469575e-06,
+      "loss": 0.7462,
+      "step": 6378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.462869929886808e-06,
+      "loss": 0.9708,
+      "step": 6379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.456544222708308e-06,
+      "loss": 0.7071,
+      "step": 6380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.450221153711414e-06,
+      "loss": 0.8781,
+      "step": 6381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.4439007237957235e-06,
+      "loss": 0.7666,
+      "step": 6382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.437582933860431e-06,
+      "loss": 0.9475,
+      "step": 6383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.431267784804385e-06,
+      "loss": 1.0815,
+      "step": 6384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.424955277526034e-06,
+      "loss": 0.9691,
+      "step": 6385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.418645412923455e-06,
+      "loss": 1.0118,
+      "step": 6386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.412338191894368e-06,
+      "loss": 0.8877,
+      "step": 6387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.406033615336093e-06,
+      "loss": 0.7937,
+      "step": 6388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.39973168414558e-06,
+      "loss": 0.8411,
+      "step": 6389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3934323992194094e-06,
+      "loss": 1.0824,
+      "step": 6390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.387135761453791e-06,
+      "loss": 0.8148,
+      "step": 6391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.380841771744536e-06,
+      "loss": 0.7531,
+      "step": 6392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.374550430987106e-06,
+      "loss": 0.8848,
+      "step": 6393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.368261740076562e-06,
+      "loss": 1.0048,
+      "step": 6394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.361975699907591e-06,
+      "loss": 0.8535,
+      "step": 6395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.355692311374523e-06,
+      "loss": 1.1219,
+      "step": 6396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.349411575371287e-06,
+      "loss": 0.8682,
+      "step": 6397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.34313349279145e-06,
+      "loss": 0.8657,
+      "step": 6398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.3368580645282e-06,
+      "loss": 0.5568,
+      "step": 6399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.330585291474331e-06,
+      "loss": 0.8928,
+      "step": 6400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.324315174522283e-06,
+      "loss": 1.0261,
+      "step": 6401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.318047714564107e-06,
+      "loss": 1.059,
+      "step": 6402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.311782912491462e-06,
+      "loss": 0.9282,
+      "step": 6403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.305520769195656e-06,
+      "loss": 0.8259,
+      "step": 6404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.299261285567606e-06,
+      "loss": 1.0142,
+      "step": 6405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.293004462497843e-06,
+      "loss": 0.7798,
+      "step": 6406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.2867503008765345e-06,
+      "loss": 0.8705,
+      "step": 6407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.28049880159346e-06,
+      "loss": 1.1367,
+      "step": 6408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.274249965538009e-06,
+      "loss": 0.8856,
+      "step": 6409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.268003793599223e-06,
+      "loss": 0.8739,
+      "step": 6410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.26176028666573e-06,
+      "loss": 0.6722,
+      "step": 6411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.255519445625812e-06,
+      "loss": 0.9432,
+      "step": 6412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.249281271367346e-06,
+      "loss": 1.0851,
+      "step": 6413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.24304576477783e-06,
+      "loss": 0.9337,
+      "step": 6414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.236812926744406e-06,
+      "loss": 1.0648,
+      "step": 6415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.2305827581538166e-06,
+      "loss": 1.271,
+      "step": 6416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.2243552598924116e-06,
+      "loss": 0.9783,
+      "step": 6417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.218130432846209e-06,
+      "loss": 0.9995,
+      "step": 6418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.211908277900802e-06,
+      "loss": 1.053,
+      "step": 6419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.205688795941411e-06,
+      "loss": 1.0746,
+      "step": 6420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.199471987852895e-06,
+      "loss": 0.8206,
+      "step": 6421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.193257854519716e-06,
+      "loss": 0.9086,
+      "step": 6422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.187046396825955e-06,
+      "loss": 1.0564,
+      "step": 6423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.180837615655325e-06,
+      "loss": 1.0839,
+      "step": 6424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.174631511891144e-06,
+      "loss": 0.8579,
+      "step": 6425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.168428086416364e-06,
+      "loss": 1.005,
+      "step": 6426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.162227340113544e-06,
+      "loss": 1.0625,
+      "step": 6427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.156029273864858e-06,
+      "loss": 0.8124,
+      "step": 6428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.14983388855212e-06,
+      "loss": 0.6741,
+      "step": 6429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.1436411850567425e-06,
+      "loss": 0.7462,
+      "step": 6430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.137451164259747e-06,
+      "loss": 0.8577,
+      "step": 6431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.131263827041819e-06,
+      "loss": 0.8467,
+      "step": 6432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.125079174283218e-06,
+      "loss": 0.8226,
+      "step": 6433
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.118897206863827e-06,
+      "loss": 0.9418,
+      "step": 6434
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.112717925663175e-06,
+      "loss": 1.0072,
+      "step": 6435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.106541331560378e-06,
+      "loss": 1.0381,
+      "step": 6436
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.100367425434172e-06,
+      "loss": 0.9879,
+      "step": 6437
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.094196208162942e-06,
+      "loss": 0.9043,
+      "step": 6438
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.08802768062465e-06,
+      "loss": 0.823,
+      "step": 6439
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0818618436969064e-06,
+      "loss": 1.1919,
+      "step": 6440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.075698698256921e-06,
+      "loss": 0.7474,
+      "step": 6441
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.069538245181519e-06,
+      "loss": 1.0594,
+      "step": 6442
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.063380485347161e-06,
+      "loss": 0.976,
+      "step": 6443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0572254196299115e-06,
+      "loss": 0.8386,
+      "step": 6444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.051073048905434e-06,
+      "loss": 0.7791,
+      "step": 6445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.044923374049055e-06,
+      "loss": 0.6846,
+      "step": 6446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.038776395935683e-06,
+      "loss": 0.8239,
+      "step": 6447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.032632115439835e-06,
+      "loss": 0.7626,
+      "step": 6448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.026490533435677e-06,
+      "loss": 0.912,
+      "step": 6449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.020351650796965e-06,
+      "loss": 0.8406,
+      "step": 6450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.014215468397074e-06,
+      "loss": 0.8782,
+      "step": 6451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.008081987109012e-06,
+      "loss": 0.6902,
+      "step": 6452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.0019512078053784e-06,
+      "loss": 0.9564,
+      "step": 6453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.995823131358413e-06,
+      "loss": 1.2406,
+      "step": 6454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 5.989697758639956e-06,
+      "loss": 1.0287,
+      "step": 6455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.983575090521451e-06,
+      "loss": 0.8195,
+      "step": 6456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.97745512787399e-06,
+      "loss": 1.1108,
+      "step": 6457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.971337871568247e-06,
+      "loss": 0.8897,
+      "step": 6458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.965223322474534e-06,
+      "loss": 0.8349,
+      "step": 6459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.959111481462773e-06,
+      "loss": 0.9645,
+      "step": 6460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.953002349402493e-06,
+      "loss": 1.1063,
+      "step": 6461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.946895927162832e-06,
+      "loss": 1.1012,
+      "step": 6462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.940792215612568e-06,
+      "loss": 1.0138,
+      "step": 6463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.934691215620067e-06,
+      "loss": 1.2435,
+      "step": 6464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.928592928053317e-06,
+      "loss": 0.8466,
+      "step": 6465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.922497353779935e-06,
+      "loss": 0.9641,
+      "step": 6466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.916404493667127e-06,
+      "loss": 0.8425,
+      "step": 6467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.910314348581736e-06,
+      "loss": 1.0365,
+      "step": 6468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.9042269193902044e-06,
+      "loss": 0.8798,
+      "step": 6469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.898142206958584e-06,
+      "loss": 0.6777,
+      "step": 6470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.892060212152564e-06,
+      "loss": 0.7731,
+      "step": 6471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.885980935837412e-06,
+      "loss": 1.5369,
+      "step": 6472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.879904378878043e-06,
+      "loss": 0.8755,
+      "step": 6473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.87383054213897e-06,
+      "loss": 0.9547,
+      "step": 6474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.867759426484318e-06,
+      "loss": 0.9615,
+      "step": 6475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.861691032777811e-06,
+      "loss": 0.8936,
+      "step": 6476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.855625361882825e-06,
+      "loss": 1.0095,
+      "step": 6477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.849562414662302e-06,
+      "loss": 0.8782,
+      "step": 6478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.843502191978834e-06,
+      "loss": 1.1793,
+      "step": 6479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.837444694694608e-06,
+      "loss": 0.9148,
+      "step": 6480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.831389923671416e-06,
+      "loss": 0.8864,
+      "step": 6481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.825337879770681e-06,
+      "loss": 0.8563,
+      "step": 6482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.819288563853431e-06,
+      "loss": 0.8215,
+      "step": 6483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.81324197678029e-06,
+      "loss": 0.8548,
+      "step": 6484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.807198119411514e-06,
+      "loss": 0.811,
+      "step": 6485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.801156992606974e-06,
+      "loss": 0.9508,
+      "step": 6486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.795118597226127e-06,
+      "loss": 0.7197,
+      "step": 6487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7890829341280675e-06,
+      "loss": 1.0071,
+      "step": 6488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.783050004171492e-06,
+      "loss": 0.7684,
+      "step": 6489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.777019808214693e-06,
+      "loss": 0.7502,
+      "step": 6490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.770992347115603e-06,
+      "loss": 0.9907,
+      "step": 6491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.764967621731737e-06,
+      "loss": 0.9937,
+      "step": 6492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.758945632920248e-06,
+      "loss": 0.937,
+      "step": 6493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.752926381537879e-06,
+      "loss": 0.7897,
+      "step": 6494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.746909868440983e-06,
+      "loss": 1.1602,
+      "step": 6495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7408960944855455e-06,
+      "loss": 0.9022,
+      "step": 6496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.7348850605271395e-06,
+      "loss": 0.8623,
+      "step": 6497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.728876767420949e-06,
+      "loss": 0.8672,
+      "step": 6498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.722871216021783e-06,
+      "loss": 0.8564,
+      "step": 6499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.716868407184056e-06,
+      "loss": 0.9424,
+      "step": 6500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.71086834176178e-06,
+      "loss": 0.7061,
+      "step": 6501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.704871020608599e-06,
+      "loss": 0.8074,
+      "step": 6502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6988764445777445e-06,
+      "loss": 1.0662,
+      "step": 6503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6928846145220565e-06,
+      "loss": 0.7979,
+      "step": 6504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.686895531294015e-06,
+      "loss": 1.1338,
+      "step": 6505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.680909195745668e-06,
+      "loss": 1.1288,
+      "step": 6506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6749256087287094e-06,
+      "loss": 1.0387,
+      "step": 6507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6689447710944195e-06,
+      "loss": 0.6048,
+      "step": 6508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.6629666836936855e-06,
+      "loss": 1.0243,
+      "step": 6509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.656991347377022e-06,
+      "loss": 0.9673,
+      "step": 6510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.651018762994539e-06,
+      "loss": 0.8879,
+      "step": 6511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.64504893139595e-06,
+      "loss": 1.0712,
+      "step": 6512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.639081853430586e-06,
+      "loss": 0.7949,
+      "step": 6513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.633117529947399e-06,
+      "loss": 0.9805,
+      "step": 6514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.627155961794917e-06,
+      "loss": 0.8857,
+      "step": 6515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.621197149821305e-06,
+      "loss": 0.8401,
+      "step": 6516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.615241094874321e-06,
+      "loss": 0.9772,
+      "step": 6517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.60928779780133e-06,
+      "loss": 0.953,
+      "step": 6518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.603337259449315e-06,
+      "loss": 0.7505,
+      "step": 6519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.597389480664852e-06,
+      "loss": 1.0334,
+      "step": 6520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.591444462294148e-06,
+      "loss": 0.8949,
+      "step": 6521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.585502205182991e-06,
+      "loss": 0.9619,
+      "step": 6522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.579562710176783e-06,
+      "loss": 0.962,
+      "step": 6523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.57362597812055e-06,
+      "loss": 0.6853,
+      "step": 6524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.567692009858904e-06,
+      "loss": 0.8907,
+      "step": 6525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.561760806236061e-06,
+      "loss": 0.6572,
+      "step": 6526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5558323680958815e-06,
+      "loss": 0.7314,
+      "step": 6527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.54990669628179e-06,
+      "loss": 1.1305,
+      "step": 6528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.543983791636831e-06,
+      "loss": 0.9128,
+      "step": 6529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.538063655003664e-06,
+      "loss": 0.8677,
+      "step": 6530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.532146287224546e-06,
+      "loss": 0.6749,
+      "step": 6531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5262316891413365e-06,
+      "loss": 0.8577,
+      "step": 6532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.520319861595516e-06,
+      "loss": 0.8936,
+      "step": 6533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.5144108054281525e-06,
+      "loss": 0.5862,
+      "step": 6534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.50850452147994e-06,
+      "loss": 0.6956,
+      "step": 6535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.50260101059116e-06,
+      "loss": 0.9551,
+      "step": 6536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.496700273601699e-06,
+      "loss": 0.9742,
+      "step": 6537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 5.49080231135107e-06,
+      "loss": 0.95,
+      "step": 6538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.484907124678373e-06,
+      "loss": 0.8849,
+      "step": 6539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.479014714422301e-06,
+      "loss": 0.8208,
+      "step": 6540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.473125081421199e-06,
+      "loss": 1.0958,
+      "step": 6541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.467238226512969e-06,
+      "loss": 0.9045,
+      "step": 6542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.461354150535128e-06,
+      "loss": 0.6315,
+      "step": 6543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.455472854324822e-06,
+      "loss": 1.312,
+      "step": 6544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.449594338718778e-06,
+      "loss": 0.75,
+      "step": 6545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.443718604553322e-06,
+      "loss": 0.96,
+      "step": 6546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.437845652664414e-06,
+      "loss": 0.847,
+      "step": 6547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4319754838875854e-06,
+      "loss": 0.9741,
+      "step": 6548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4261080990579975e-06,
+      "loss": 0.7127,
+      "step": 6549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.420243499010402e-06,
+      "loss": 1.011,
+      "step": 6550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.414381684579148e-06,
+      "loss": 0.8394,
+      "step": 6551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.408522656598211e-06,
+      "loss": 0.8048,
+      "step": 6552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.4026664159011395e-06,
+      "loss": 1.0129,
+      "step": 6553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3968129633211155e-06,
+      "loss": 0.9524,
+      "step": 6554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.390962299690911e-06,
+      "loss": 0.8914,
+      "step": 6555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.385114425842899e-06,
+      "loss": 0.9778,
+      "step": 6556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.37926934260905e-06,
+      "loss": 0.8872,
+      "step": 6557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.373427050820956e-06,
+      "loss": 0.8152,
+      "step": 6558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.367587551309799e-06,
+      "loss": 1.1007,
+      "step": 6559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3617508449063575e-06,
+      "loss": 0.8715,
+      "step": 6560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.355916932441032e-06,
+      "loss": 1.0297,
+      "step": 6561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.350085814743805e-06,
+      "loss": 0.8427,
+      "step": 6562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3442574926442806e-06,
+      "loss": 1.0146,
+      "step": 6563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.338431966971652e-06,
+      "loss": 0.7042,
+      "step": 6564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3326092385547086e-06,
+      "loss": 1.1562,
+      "step": 6565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.3267893082218686e-06,
+      "loss": 1.0133,
+      "step": 6566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.320972176801117e-06,
+      "loss": 0.916,
+      "step": 6567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.315157845120067e-06,
+      "loss": 1.0266,
+      "step": 6568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.309346314005931e-06,
+      "loss": 0.6532,
+      "step": 6569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.303537584285512e-06,
+      "loss": 0.9434,
+      "step": 6570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2977316567852095e-06,
+      "loss": 0.6915,
+      "step": 6571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.29192853233105e-06,
+      "loss": 0.8174,
+      "step": 6572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.286128211748631e-06,
+      "loss": 0.8235,
+      "step": 6573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.280330695863181e-06,
+      "loss": 0.8324,
+      "step": 6574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.274535985499502e-06,
+      "loss": 0.6519,
+      "step": 6575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.268744081482005e-06,
+      "loss": 0.858,
+      "step": 6576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.262954984634721e-06,
+      "loss": 0.9814,
+      "step": 6577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.257168695781259e-06,
+      "loss": 0.8998,
+      "step": 6578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.251385215744825e-06,
+      "loss": 0.962,
+      "step": 6579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.245604545348251e-06,
+      "loss": 0.8538,
+      "step": 6580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.239826685413943e-06,
+      "loss": 0.7792,
+      "step": 6581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.234051636763923e-06,
+      "loss": 0.8999,
+      "step": 6582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2282794002198154e-06,
+      "loss": 0.7867,
+      "step": 6583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.222509976602833e-06,
+      "loss": 0.9896,
+      "step": 6584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.216743366733784e-06,
+      "loss": 1.0171,
+      "step": 6585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.2109795714331e-06,
+      "loss": 1.0703,
+      "step": 6586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.205218591520783e-06,
+      "loss": 0.8937,
+      "step": 6587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.199460427816463e-06,
+      "loss": 0.7132,
+      "step": 6588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.193705081139347e-06,
+      "loss": 1.0844,
+      "step": 6589
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.187952552308245e-06,
+      "loss": 0.8134,
+      "step": 6590
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.182202842141584e-06,
+      "loss": 1.0879,
+      "step": 6591
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.176455951457368e-06,
+      "loss": 0.8368,
+      "step": 6592
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.170711881073201e-06,
+      "loss": 1.1488,
+      "step": 6593
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.164970631806304e-06,
+      "loss": 0.9041,
+      "step": 6594
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.159232204473488e-06,
+      "loss": 1.0518,
+      "step": 6595
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.15349659989115e-06,
+      "loss": 1.2271,
+      "step": 6596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.147763818875306e-06,
+      "loss": 0.7125,
+      "step": 6597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.1420338622415605e-06,
+      "loss": 0.8131,
+      "step": 6598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.136306730805099e-06,
+      "loss": 0.8063,
+      "step": 6599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.130582425380742e-06,
+      "loss": 0.8922,
+      "step": 6600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.124860946782872e-06,
+      "loss": 0.7563,
+      "step": 6601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.119142295825499e-06,
+      "loss": 0.7964,
+      "step": 6602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.113426473322211e-06,
+      "loss": 1.0835,
+      "step": 6603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.107713480086191e-06,
+      "loss": 1.0111,
+      "step": 6604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.102003316930243e-06,
+      "loss": 1.0975,
+      "step": 6605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.096295984666743e-06,
+      "loss": 1.0378,
+      "step": 6606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.090591484107671e-06,
+      "loss": 1.1599,
+      "step": 6607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.084889816064611e-06,
+      "loss": 1.0659,
+      "step": 6608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.079190981348753e-06,
+      "loss": 1.0049,
+      "step": 6609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.073494980770854e-06,
+      "loss": 1.0624,
+      "step": 6610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.067801815141299e-06,
+      "loss": 0.7377,
+      "step": 6611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.062111485270049e-06,
+      "loss": 0.942,
+      "step": 6612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0564239919666635e-06,
+      "loss": 0.6735,
+      "step": 6613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.050739336040316e-06,
+      "loss": 0.7723,
+      "step": 6614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.04505751829975e-06,
+      "loss": 0.8772,
+      "step": 6615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0393785395533325e-06,
+      "loss": 1.0195,
+      "step": 6616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.033702400609006e-06,
+      "loss": 0.8838,
+      "step": 6617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.028029102274309e-06,
+      "loss": 1.1689,
+      "step": 6618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.022358645356398e-06,
+      "loss": 1.0886,
+      "step": 6619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.016691030662002e-06,
+      "loss": 0.8184,
+      "step": 6620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 5.0110262589974475e-06,
+      "loss": 0.8131,
+      "step": 6621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.005364331168671e-06,
+      "loss": 1.0748,
+      "step": 6622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.999705247981201e-06,
+      "loss": 1.1853,
+      "step": 6623
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.994049010240145e-06,
+      "loss": 0.7635,
+      "step": 6624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.988395618750228e-06,
+      "loss": 0.95,
+      "step": 6625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.982745074315753e-06,
+      "loss": 0.9371,
+      "step": 6626
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.977097377740622e-06,
+      "loss": 1.1633,
+      "step": 6627
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.971452529828341e-06,
+      "loss": 0.907,
+      "step": 6628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.965810531381995e-06,
+      "loss": 1.0011,
+      "step": 6629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.960171383204282e-06,
+      "loss": 0.9645,
+      "step": 6630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.954535086097481e-06,
+      "loss": 0.7366,
+      "step": 6631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.948901640863462e-06,
+      "loss": 0.7665,
+      "step": 6632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.943271048303713e-06,
+      "loss": 0.9001,
+      "step": 6633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.937643309219286e-06,
+      "loss": 0.7468,
+      "step": 6634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.932018424410833e-06,
+      "loss": 0.7314,
+      "step": 6635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.926396394678631e-06,
+      "loss": 0.9508,
+      "step": 6636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.920777220822518e-06,
+      "loss": 1.0876,
+      "step": 6637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9151609036419215e-06,
+      "loss": 1.111,
+      "step": 6638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.9095474439358965e-06,
+      "loss": 1.0475,
+      "step": 6639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.903936842503062e-06,
+      "loss": 1.2258,
+      "step": 6640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.898329100141635e-06,
+      "loss": 0.8951,
+      "step": 6641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.892724217649439e-06,
+      "loss": 0.9229,
+      "step": 6642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.88712219582387e-06,
+      "loss": 1.4042,
+      "step": 6643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.881523035461946e-06,
+      "loss": 0.9924,
+      "step": 6644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.875926737360251e-06,
+      "loss": 1.1794,
+      "step": 6645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.870333302314964e-06,
+      "loss": 0.884,
+      "step": 6646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.86474273112188e-06,
+      "loss": 1.1777,
+      "step": 6647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.859155024576361e-06,
+      "loss": 0.8922,
+      "step": 6648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.853570183473363e-06,
+      "loss": 1.0065,
+      "step": 6649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.847988208607465e-06,
+      "loss": 0.7455,
+      "step": 6650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.842409100772805e-06,
+      "loss": 0.6893,
+      "step": 6651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.836832860763113e-06,
+      "loss": 1.0095,
+      "step": 6652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.83125948937174e-06,
+      "loss": 0.8407,
+      "step": 6653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.825688987391602e-06,
+      "loss": 0.7607,
+      "step": 6654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.820121355615209e-06,
+      "loss": 1.056,
+      "step": 6655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.814556594834682e-06,
+      "loss": 0.9395,
+      "step": 6656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.80899470584171e-06,
+      "loss": 1.2563,
+      "step": 6657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.8034356894275964e-06,
+      "loss": 1.078,
+      "step": 6658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.797879546383213e-06,
+      "loss": 1.2101,
+      "step": 6659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.792326277499035e-06,
+      "loss": 1.0785,
+      "step": 6660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.786775883565131e-06,
+      "loss": 0.8403,
+      "step": 6661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7812283653711516e-06,
+      "loss": 0.7371,
+      "step": 6662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.775683723706345e-06,
+      "loss": 0.7668,
+      "step": 6663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.770141959359558e-06,
+      "loss": 0.9153,
+      "step": 6664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.764603073119211e-06,
+      "loss": 0.7981,
+      "step": 6665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.7590670657733175e-06,
+      "loss": 0.8076,
+      "step": 6666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.753533938109497e-06,
+      "loss": 1.079,
+      "step": 6667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.748003690914934e-06,
+      "loss": 0.8464,
+      "step": 6668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.742476324976438e-06,
+      "loss": 1.0052,
+      "step": 6669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.736951841080378e-06,
+      "loss": 0.9235,
+      "step": 6670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.731430240012716e-06,
+      "loss": 0.7914,
+      "step": 6671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.725911522559023e-06,
+      "loss": 0.812,
+      "step": 6672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.720395689504448e-06,
+      "loss": 1.0986,
+      "step": 6673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.714882741633719e-06,
+      "loss": 0.9762,
+      "step": 6674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.709372679731177e-06,
+      "loss": 0.6825,
+      "step": 6675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.703865504580726e-06,
+      "loss": 1.1895,
+      "step": 6676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.698361216965885e-06,
+      "loss": 0.855,
+      "step": 6677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6928598176697505e-06,
+      "loss": 0.8911,
+      "step": 6678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.687361307475002e-06,
+      "loss": 0.8508,
+      "step": 6679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.681865687163911e-06,
+      "loss": 1.108,
+      "step": 6680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.676372957518352e-06,
+      "loss": 0.8608,
+      "step": 6681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.670883119319763e-06,
+      "loss": 0.7941,
+      "step": 6682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6653961733491985e-06,
+      "loss": 0.7178,
+      "step": 6683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.65991212038728e-06,
+      "loss": 0.929,
+      "step": 6684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.65443096121422e-06,
+      "loss": 1.0288,
+      "step": 6685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.648952696609834e-06,
+      "loss": 1.1771,
+      "step": 6686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.643477327353513e-06,
+      "loss": 0.9241,
+      "step": 6687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.638004854224232e-06,
+      "loss": 0.9762,
+      "step": 6688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.632535278000571e-06,
+      "loss": 0.9792,
+      "step": 6689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6270685994606795e-06,
+      "loss": 0.9236,
+      "step": 6690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.6216048193823055e-06,
+      "loss": 1.0297,
+      "step": 6691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.616143938542791e-06,
+      "loss": 0.809,
+      "step": 6692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.610685957719049e-06,
+      "loss": 1.1852,
+      "step": 6693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.60523087768758e-06,
+      "loss": 0.8411,
+      "step": 6694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.599778699224497e-06,
+      "loss": 0.8695,
+      "step": 6695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.594329423105465e-06,
+      "loss": 0.915,
+      "step": 6696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5888830501057685e-06,
+      "loss": 0.7539,
+      "step": 6697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5834395810002595e-06,
+      "loss": 1.0525,
+      "step": 6698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.5779990165633725e-06,
+      "loss": 0.8192,
+      "step": 6699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.572561357569152e-06,
+      "loss": 0.8699,
+      "step": 6700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.567126604791208e-06,
+      "loss": 1.0714,
+      "step": 6701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.561694759002735e-06,
+      "loss": 0.8748,
+      "step": 6702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.556265820976535e-06,
+      "loss": 0.9454,
+      "step": 6703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.550839791484987e-06,
+      "loss": 0.7767,
+      "step": 6704
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 4.54541667130004e-06,
+      "loss": 0.8892,
+      "step": 6705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.53999646119326e-06,
+      "loss": 1.2148,
+      "step": 6706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.534579161935767e-06,
+      "loss": 1.0662,
+      "step": 6707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.529164774298284e-06,
+      "loss": 1.0883,
+      "step": 6708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.5237532990511226e-06,
+      "loss": 1.0011,
+      "step": 6709
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.518344736964167e-06,
+      "loss": 1.0264,
+      "step": 6710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.512939088806903e-06,
+      "loss": 1.0089,
+      "step": 6711
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.507536355348391e-06,
+      "loss": 1.019,
+      "step": 6712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.502136537357271e-06,
+      "loss": 0.8689,
+      "step": 6713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.496739635601788e-06,
+      "loss": 0.7633,
+      "step": 6714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.491345650849754e-06,
+      "loss": 0.9825,
+      "step": 6715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.485954583868568e-06,
+      "loss": 0.9637,
+      "step": 6716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4805664354252235e-06,
+      "loss": 0.9025,
+      "step": 6717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.475181206286299e-06,
+      "loss": 0.7246,
+      "step": 6718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.469798897217945e-06,
+      "loss": 0.6467,
+      "step": 6719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.464419508985909e-06,
+      "loss": 1.0103,
+      "step": 6720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.459043042355515e-06,
+      "loss": 0.8888,
+      "step": 6721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.453669498091667e-06,
+      "loss": 0.8359,
+      "step": 6722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.448298876958873e-06,
+      "loss": 0.8348,
+      "step": 6723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.442931179721202e-06,
+      "loss": 0.5656,
+      "step": 6724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.437566407142326e-06,
+      "loss": 0.7585,
+      "step": 6725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.432204559985492e-06,
+      "loss": 0.9419,
+      "step": 6726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.426845639013518e-06,
+      "loss": 1.0948,
+      "step": 6727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.421489644988836e-06,
+      "loss": 1.2731,
+      "step": 6728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.416136578673438e-06,
+      "loss": 0.7416,
+      "step": 6729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4107864408288905e-06,
+      "loss": 0.7522,
+      "step": 6730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.4054392322163875e-06,
+      "loss": 0.8593,
+      "step": 6731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.400094953596662e-06,
+      "loss": 0.9363,
+      "step": 6732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3947536057300424e-06,
+      "loss": 1.0314,
+      "step": 6733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3894151893764526e-06,
+      "loss": 0.7033,
+      "step": 6734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.384079705295388e-06,
+      "loss": 0.9706,
+      "step": 6735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.378747154245919e-06,
+      "loss": 1.0504,
+      "step": 6736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3734175369867235e-06,
+      "loss": 0.8156,
+      "step": 6737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.368090854276033e-06,
+      "loss": 0.6987,
+      "step": 6738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.362767106871693e-06,
+      "loss": 0.8864,
+      "step": 6739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3574462955311006e-06,
+      "loss": 1.0043,
+      "step": 6740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.352128421011251e-06,
+      "loss": 0.9,
+      "step": 6741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.346813484068724e-06,
+      "loss": 1.059,
+      "step": 6742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3415014854596755e-06,
+      "loss": 0.9326,
+      "step": 6743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3361924259398305e-06,
+      "loss": 1.0895,
+      "step": 6744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3308863062645364e-06,
+      "loss": 1.0208,
+      "step": 6745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.325583127188682e-06,
+      "loss": 0.9603,
+      "step": 6746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.320282889466748e-06,
+      "loss": 0.8649,
+      "step": 6747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.314985593852808e-06,
+      "loss": 0.9445,
+      "step": 6748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.309691241100508e-06,
+      "loss": 0.7681,
+      "step": 6749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.3043998319630686e-06,
+      "loss": 1.0876,
+      "step": 6750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.299111367193312e-06,
+      "loss": 1.0564,
+      "step": 6751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.293825847543617e-06,
+      "loss": 0.801,
+      "step": 6752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.288543273765969e-06,
+      "loss": 0.9299,
+      "step": 6753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.283263646611915e-06,
+      "loss": 0.8391,
+      "step": 6754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.277986966832578e-06,
+      "loss": 0.8101,
+      "step": 6755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.272713235178694e-06,
+      "loss": 0.7001,
+      "step": 6756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.267442452400541e-06,
+      "loss": 0.9998,
+      "step": 6757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.262174619247991e-06,
+      "loss": 0.9049,
+      "step": 6758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.256909736470519e-06,
+      "loss": 0.9204,
+      "step": 6759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.251647804817149e-06,
+      "loss": 0.9607,
+      "step": 6760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.246388825036496e-06,
+      "loss": 0.7296,
+      "step": 6761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.241132797876762e-06,
+      "loss": 0.8256,
+      "step": 6762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.235879724085715e-06,
+      "loss": 0.8646,
+      "step": 6763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.230629604410724e-06,
+      "loss": 1.0648,
+      "step": 6764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.225382439598716e-06,
+      "loss": 0.6842,
+      "step": 6765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.2201382303961976e-06,
+      "loss": 0.7861,
+      "step": 6766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.214896977549282e-06,
+      "loss": 0.8922,
+      "step": 6767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.209658681803635e-06,
+      "loss": 0.7582,
+      "step": 6768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.204423343904504e-06,
+      "loss": 1.0826,
+      "step": 6769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.199190964596733e-06,
+      "loss": 0.884,
+      "step": 6770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.193961544624722e-06,
+      "loss": 0.8033,
+      "step": 6771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.1887350847324675e-06,
+      "loss": 0.9198,
+      "step": 6772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.183511585663543e-06,
+      "loss": 0.8092,
+      "step": 6773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.178291048161098e-06,
+      "loss": 1.0276,
+      "step": 6774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.173073472967845e-06,
+      "loss": 0.7631,
+      "step": 6775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.16785886082611e-06,
+      "loss": 0.9822,
+      "step": 6776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.162647212477757e-06,
+      "loss": 0.7319,
+      "step": 6777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.157438528664265e-06,
+      "loss": 1.0431,
+      "step": 6778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.152232810126666e-06,
+      "loss": 0.8546,
+      "step": 6779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.147030057605577e-06,
+      "loss": 1.2175,
+      "step": 6780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.141830271841201e-06,
+      "loss": 0.7589,
+      "step": 6781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.136633453573313e-06,
+      "loss": 0.7463,
+      "step": 6782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.131439603541254e-06,
+      "loss": 0.9845,
+      "step": 6783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.126248722483969e-06,
+      "loss": 0.7618,
+      "step": 6784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.121060811139954e-06,
+      "loss": 0.7917,
+      "step": 6785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.115875870247299e-06,
+      "loss": 0.8115,
+      "step": 6786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.110693900543675e-06,
+      "loss": 0.8547,
+      "step": 6787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.105514902766311e-06,
+      "loss": 0.8263,
+      "step": 6788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.100338877652024e-06,
+      "loss": 0.9943,
+      "step": 6789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.095165825937217e-06,
+      "loss": 0.7917,
+      "step": 6790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.08999574835785e-06,
+      "loss": 0.8593,
+      "step": 6791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.084828645649483e-06,
+      "loss": 0.8259,
+      "step": 6792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.079664518547235e-06,
+      "loss": 0.7382,
+      "step": 6793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.074503367785801e-06,
+      "loss": 0.9449,
+      "step": 6794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.069345194099472e-06,
+      "loss": 1.1405,
+      "step": 6795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.064189998222096e-06,
+      "loss": 0.808,
+      "step": 6796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.059037780887096e-06,
+      "loss": 1.104,
+      "step": 6797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.053888542827489e-06,
+      "loss": 0.9215,
+      "step": 6798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.048742284775861e-06,
+      "loss": 1.0299,
+      "step": 6799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.043599007464363e-06,
+      "loss": 0.995,
+      "step": 6800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.038458711624738e-06,
+      "loss": 0.9031,
+      "step": 6801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.033321397988293e-06,
+      "loss": 0.682,
+      "step": 6802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.028187067285907e-06,
+      "loss": 0.7256,
+      "step": 6803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.023055720248059e-06,
+      "loss": 0.7769,
+      "step": 6804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.01792735760477e-06,
+      "loss": 0.9062,
+      "step": 6805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.012801980085667e-06,
+      "loss": 0.8188,
+      "step": 6806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.0076795884199325e-06,
+      "loss": 1.0703,
+      "step": 6807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.002560183336326e-06,
+      "loss": 0.7195,
+      "step": 6808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.997443765563197e-06,
+      "loss": 0.8448,
+      "step": 6809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.992330335828451e-06,
+      "loss": 1.0638,
+      "step": 6810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.987219894859576e-06,
+      "loss": 0.7738,
+      "step": 6811
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.982112443383637e-06,
+      "loss": 0.965,
+      "step": 6812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.977007982127282e-06,
+      "loss": 1.1079,
+      "step": 6813
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.971906511816709e-06,
+      "loss": 1.0785,
+      "step": 6814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.966808033177716e-06,
+      "loss": 0.9246,
+      "step": 6815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9617125469356615e-06,
+      "loss": 0.715,
+      "step": 6816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.956620053815477e-06,
+      "loss": 0.8793,
+      "step": 6817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9515305545416825e-06,
+      "loss": 0.6471,
+      "step": 6818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.946444049838349e-06,
+      "loss": 0.8496,
+      "step": 6819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.941360540429148e-06,
+      "loss": 1.097,
+      "step": 6820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.936280027037304e-06,
+      "loss": 1.0031,
+      "step": 6821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.93120251038562e-06,
+      "loss": 0.7411,
+      "step": 6822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.9261279911964845e-06,
+      "loss": 0.6945,
+      "step": 6823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.921056470191847e-06,
+      "loss": 0.8258,
+      "step": 6824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.915987948093225e-06,
+      "loss": 0.8817,
+      "step": 6825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.910922425621727e-06,
+      "loss": 0.8773,
+      "step": 6826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.905859903498033e-06,
+      "loss": 0.9293,
+      "step": 6827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.900800382442371e-06,
+      "loss": 0.8861,
+      "step": 6828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.895743863174578e-06,
+      "loss": 0.8083,
+      "step": 6829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.890690346414036e-06,
+      "loss": 0.8545,
+      "step": 6830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.885639832879709e-06,
+      "loss": 0.6169,
+      "step": 6831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.880592323290144e-06,
+      "loss": 0.9523,
+      "step": 6832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.875547818363437e-06,
+      "loss": 0.9991,
+      "step": 6833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8705063188172876e-06,
+      "loss": 0.8372,
+      "step": 6834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.865467825368941e-06,
+      "loss": 0.9548,
+      "step": 6835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.86043233873522e-06,
+      "loss": 1.0055,
+      "step": 6836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.855399859632539e-06,
+      "loss": 0.9579,
+      "step": 6837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.850370388776861e-06,
+      "loss": 0.9342,
+      "step": 6838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8453439268837185e-06,
+      "loss": 1.0647,
+      "step": 6839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.840320474668252e-06,
+      "loss": 0.9827,
+      "step": 6840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.835300032845135e-06,
+      "loss": 0.837,
+      "step": 6841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.830282602128623e-06,
+      "loss": 0.8865,
+      "step": 6842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.825268183232558e-06,
+      "loss": 1.0143,
+      "step": 6843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.82025677687034e-06,
+      "loss": 1.1641,
+      "step": 6844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.815248383754933e-06,
+      "loss": 0.8685,
+      "step": 6845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.810243004598893e-06,
+      "loss": 0.8292,
+      "step": 6846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.805240640114327e-06,
+      "loss": 0.9734,
+      "step": 6847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.8002412910129346e-06,
+      "loss": 1.0504,
+      "step": 6848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7952449580059694e-06,
+      "loss": 0.9069,
+      "step": 6849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7902516418042516e-06,
+      "loss": 0.8614,
+      "step": 6850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.785261343118196e-06,
+      "loss": 0.7604,
+      "step": 6851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7802740626577683e-06,
+      "loss": 0.7796,
+      "step": 6852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7752898011324923e-06,
+      "loss": 1.1091,
+      "step": 6853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.770308559251512e-06,
+      "loss": 1.1783,
+      "step": 6854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.765330337723491e-06,
+      "loss": 0.9231,
+      "step": 6855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7603551372566796e-06,
+      "loss": 1.0565,
+      "step": 6856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7553829585589116e-06,
+      "loss": 1.001,
+      "step": 6857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7504138023375685e-06,
+      "loss": 0.8116,
+      "step": 6858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.745447669299626e-06,
+      "loss": 0.7896,
+      "step": 6859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7404845601516082e-06,
+      "loss": 0.6086,
+      "step": 6860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7355244755996168e-06,
+      "loss": 0.7113,
+      "step": 6861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7305674163493285e-06,
+      "loss": 0.911,
+      "step": 6862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7256133831059857e-06,
+      "loss": 1.0088,
+      "step": 6863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7206623765743935e-06,
+      "loss": 1.0002,
+      "step": 6864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7157143974589414e-06,
+      "loss": 0.9939,
+      "step": 6865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.710769446463569e-06,
+      "loss": 0.7872,
+      "step": 6866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.705827524291802e-06,
+      "loss": 0.7646,
+      "step": 6867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.7008886316467346e-06,
+      "loss": 0.9008,
+      "step": 6868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6959527692310176e-06,
+      "loss": 0.7814,
+      "step": 6869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.69101993774687e-06,
+      "loss": 0.7791,
+      "step": 6870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.686090137896103e-06,
+      "loss": 0.8237,
+      "step": 6871
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.681163370380064e-06,
+      "loss": 0.9691,
+      "step": 6872
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6762396358997e-06,
+      "loss": 0.8622,
+      "step": 6873
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6713189351555043e-06,
+      "loss": 0.8252,
+      "step": 6874
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6664012688475413e-06,
+      "loss": 0.9675,
+      "step": 6875
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6614866376754623e-06,
+      "loss": 0.8344,
+      "step": 6876
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6565750423384614e-06,
+      "loss": 0.857,
+      "step": 6877
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6516664835353098e-06,
+      "loss": 0.8113,
+      "step": 6878
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.646760961964363e-06,
+      "loss": 1.1433,
+      "step": 6879
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.641858478323515e-06,
+      "loss": 0.9027,
+      "step": 6880
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.63695903331025e-06,
+      "loss": 1.0472,
+      "step": 6881
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6320626276216214e-06,
+      "loss": 0.9714,
+      "step": 6882
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627169261954236e-06,
+      "loss": 0.9302,
+      "step": 6883
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622278937004264e-06,
+      "loss": 0.847,
+      "step": 6884
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6173916534674694e-06,
+      "loss": 1.0419,
+      "step": 6885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612507412039154e-06,
+      "loss": 0.9128,
+      "step": 6886
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6076262134142085e-06,
+      "loss": 1.097,
+      "step": 6887
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.602748058287081e-06,
+      "loss": 1.049,
+      "step": 6888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5978729473517797e-06,
+      "loss": 0.8806,
+      "step": 6889
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5930008813019e-06,
+      "loss": 0.903,
+      "step": 6890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.588131860830585e-06,
+      "loss": 0.6419,
+      "step": 6891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5832658866305463e-06,
+      "loss": 1.1836,
+      "step": 6892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.578402959394078e-06,
+      "loss": 0.8918,
+      "step": 6893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.573543079813019e-06,
+      "loss": 1.1124,
+      "step": 6894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.56868624857879e-06,
+      "loss": 0.7808,
+      "step": 6895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5638324663823802e-06,
+      "loss": 1.1964,
+      "step": 6896
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.558981733914335e-06,
+      "loss": 1.0338,
+      "step": 6897
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5541340518647604e-06,
+      "loss": 0.9144,
+      "step": 6898
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.549289420923346e-06,
+      "loss": 0.9907,
+      "step": 6899
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.544447841779333e-06,
+      "loss": 0.7935,
+      "step": 6900
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.539609315121542e-06,
+      "loss": 0.853,
+      "step": 6901
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5347738416383476e-06,
+      "loss": 0.7841,
+      "step": 6902
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.529941422017685e-06,
+      "loss": 0.7852,
+      "step": 6903
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5251120569470773e-06,
+      "loss": 1.1518,
+      "step": 6904
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.520285747113594e-06,
+      "loss": 0.7861,
+      "step": 6905
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5154624932038692e-06,
+      "loss": 0.9211,
+      "step": 6906
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5106422959041125e-06,
+      "loss": 1.1113,
+      "step": 6907
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5058251559001e-06,
+      "loss": 0.7689,
+      "step": 6908
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.5010110738771584e-06,
+      "loss": 1.0781,
+      "step": 6909
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4962000505201986e-06,
+      "loss": 0.8315,
+      "step": 6910
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4913920865136785e-06,
+      "loss": 0.9767,
+      "step": 6911
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4865871825416256e-06,
+      "loss": 1.0634,
+      "step": 6912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4817853392876436e-06,
+      "loss": 0.8191,
+      "step": 6913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4769865574348804e-06,
+      "loss": 1.0311,
+      "step": 6914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.472190837666073e-06,
+      "loss": 0.829,
+      "step": 6915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.467398180663503e-06,
+      "loss": 1.0203,
+      "step": 6916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.462608587109015e-06,
+      "loss": 0.7671,
+      "step": 6917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.45782205768404e-06,
+      "loss": 0.8214,
+      "step": 6918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.453038593069552e-06,
+      "loss": 0.891,
+      "step": 6919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4482581939460913e-06,
+      "loss": 1.1443,
+      "step": 6920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4434808609937697e-06,
+      "loss": 1.1487,
+      "step": 6921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.438706594892263e-06,
+      "loss": 0.9597,
+      "step": 6922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4339353963208005e-06,
+      "loss": 1.0614,
+      "step": 6923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4291672659581915e-06,
+      "loss": 0.7629,
+      "step": 6924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4244022044827933e-06,
+      "loss": 1.0474,
+      "step": 6925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4196402125725253e-06,
+      "loss": 0.9797,
+      "step": 6926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4148812909048893e-06,
+      "loss": 0.7738,
+      "step": 6927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.410125440156928e-06,
+      "loss": 0.8522,
+      "step": 6928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.4053726610052654e-06,
+      "loss": 0.7871,
+      "step": 6929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.400622954126079e-06,
+      "loss": 0.8129,
+      "step": 6930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.395876320195102e-06,
+      "loss": 1.1278,
+      "step": 6931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3911327598876524e-06,
+      "loss": 0.8981,
+      "step": 6932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3863922738785914e-06,
+      "loss": 0.7883,
+      "step": 6933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.38165486284234e-06,
+      "loss": 0.8522,
+      "step": 6934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3769205274529037e-06,
+      "loss": 0.8422,
+      "step": 6935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3721892683838352e-06,
+      "loss": 0.8387,
+      "step": 6936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3674610863082434e-06,
+      "loss": 1.104,
+      "step": 6937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.362735981898821e-06,
+      "loss": 0.861,
+      "step": 6938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3580139558278046e-06,
+      "loss": 0.9507,
+      "step": 6939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3532950087669877e-06,
+      "loss": 0.7454,
+      "step": 6940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3485791413877506e-06,
+      "loss": 0.6162,
+      "step": 6941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3438663543610084e-06,
+      "loss": 0.9514,
+      "step": 6942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.339156648357264e-06,
+      "loss": 1.1544,
+      "step": 6943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3344500240465593e-06,
+      "loss": 1.0302,
+      "step": 6944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.329746482098503e-06,
+      "loss": 1.1471,
+      "step": 6945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3250460231822795e-06,
+      "loss": 1.1027,
+      "step": 6946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.32034864796662e-06,
+      "loss": 0.7987,
+      "step": 6947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3156543571198075e-06,
+      "loss": 1.134,
+      "step": 6948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.3109631513097235e-06,
+      "loss": 0.7495,
+      "step": 6949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.306275031203779e-06,
+      "loss": 0.9749,
+      "step": 6950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.30158999746894e-06,
+      "loss": 0.8379,
+      "step": 6951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.2969080507717685e-06,
+      "loss": 0.9598,
+      "step": 6952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.292229191778354e-06,
+      "loss": 1.0417,
+      "step": 6953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.287553421154352e-06,
+      "loss": 0.9072,
+      "step": 6954
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.282880739565003e-06,
+      "loss": 0.9243,
+      "step": 6955
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2782111476750756e-06,
+      "loss": 0.8627,
+      "step": 6956
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.273544646148921e-06,
+      "loss": 1.0231,
+      "step": 6957
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.268881235650445e-06,
+      "loss": 0.8368,
+      "step": 6958
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2642209168431053e-06,
+      "loss": 0.8945,
+      "step": 6959
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2595636903899328e-06,
+      "loss": 0.9863,
+      "step": 6960
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2549095569535087e-06,
+      "loss": 0.8215,
+      "step": 6961
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2502585171959698e-06,
+      "loss": 0.7724,
+      "step": 6962
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2456105717790397e-06,
+      "loss": 0.8109,
+      "step": 6963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.24096572136397e-06,
+      "loss": 0.6461,
+      "step": 6964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.236323966611582e-06,
+      "loss": 0.7463,
+      "step": 6965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.231685308182272e-06,
+      "loss": 0.8298,
+      "step": 6966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.227049746735966e-06,
+      "loss": 1.3246,
+      "step": 6967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.222417282932183e-06,
+      "loss": 1.1732,
+      "step": 6968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2177879174299764e-06,
+      "loss": 1.0232,
+      "step": 6969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2131616508879607e-06,
+      "loss": 1.1014,
+      "step": 6970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2085384839643294e-06,
+      "loss": 0.8745,
+      "step": 6971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.2039184173168146e-06,
+      "loss": 0.9719,
+      "step": 6972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.199301451602707e-06,
+      "loss": 0.7509,
+      "step": 6973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1946875874788786e-06,
+      "loss": 0.9614,
+      "step": 6974
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1900768256017316e-06,
+      "loss": 0.7576,
+      "step": 6975
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1854691666272444e-06,
+      "loss": 1.1367,
+      "step": 6976
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.180864611210957e-06,
+      "loss": 0.8108,
+      "step": 6977
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1762631600079555e-06,
+      "loss": 0.674,
+      "step": 6978
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.171664813672881e-06,
+      "loss": 0.8701,
+      "step": 6979
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1670695728599543e-06,
+      "loss": 0.923,
+      "step": 6980
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.16247743822293e-06,
+      "loss": 0.9424,
+      "step": 6981
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.157888410415144e-06,
+      "loss": 1.1431,
+      "step": 6982
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1533024900894715e-06,
+      "loss": 0.7525,
+      "step": 6983
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1487196778983485e-06,
+      "loss": 0.6644,
+      "step": 6984
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1441399744937817e-06,
+      "loss": 1.0629,
+      "step": 6985
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1395633805273218e-06,
+      "loss": 0.5974,
+      "step": 6986
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1349898966500756e-06,
+      "loss": 0.6114,
+      "step": 6987
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.130419523512729e-06,
+      "loss": 1.1232,
+      "step": 6988
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.125852261765491e-06,
+      "loss": 0.774,
+      "step": 6989
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.12128811205816e-06,
+      "loss": 0.741,
+      "step": 6990
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1167270750400822e-06,
+      "loss": 0.9293,
+      "step": 6991
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1121691513601504e-06,
+      "loss": 0.8313,
+      "step": 6992
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1076143416668148e-06,
+      "loss": 0.9542,
+      "step": 6993
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.1030626466081013e-06,
+      "loss": 0.7994,
+      "step": 6994
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0985140668315704e-06,
+      "loss": 0.8932,
+      "step": 6995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.093968602984362e-06,
+      "loss": 0.895,
+      "step": 6996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0894262557131514e-06,
+      "loss": 0.6544,
+      "step": 6997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0848870256641765e-06,
+      "loss": 0.7754,
+      "step": 6998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.080350913483243e-06,
+      "loss": 1.0088,
+      "step": 6999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0758179198157018e-06,
+      "loss": 0.8499,
+      "step": 7000
+    },
+    {
+      "epoch": 0.84,
+      "eval_loss": 0.98095703125,
+      "eval_runtime": 3.6395,
+      "eval_samples_per_second": 27.477,
+      "eval_steps_per_second": 4.671,
+      "step": 7000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.071288045306456e-06,
+      "loss": 0.8328,
+      "step": 7001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0667612905999843e-06,
+      "loss": 0.7528,
+      "step": 7002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0622376563402945e-06,
+      "loss": 1.0342,
+      "step": 7003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.057717143170974e-06,
+      "loss": 1.0024,
+      "step": 7004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0531997517351614e-06,
+      "loss": 0.8599,
+      "step": 7005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0486854826755424e-06,
+      "loss": 0.8947,
+      "step": 7006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0441743366343587e-06,
+      "loss": 0.705,
+      "step": 7007
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.039666314253417e-06,
+      "loss": 1.3207,
+      "step": 7008
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0351614161740703e-06,
+      "loss": 0.8031,
+      "step": 7009
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0306596430372426e-06,
+      "loss": 1.116,
+      "step": 7010
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0261609954833903e-06,
+      "loss": 0.7294,
+      "step": 7011
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.021665474152538e-06,
+      "loss": 0.9835,
+      "step": 7012
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0171730796842713e-06,
+      "loss": 0.9083,
+      "step": 7013
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0126838127177204e-06,
+      "loss": 0.7622,
+      "step": 7014
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.008197673891569e-06,
+      "loss": 0.7922,
+      "step": 7015
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.0037146638440676e-06,
+      "loss": 0.9747,
+      "step": 7016
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9992347832130174e-06,
+      "loss": 0.8572,
+      "step": 7017
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9947580326357638e-06,
+      "loss": 0.8838,
+      "step": 7018
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9902844127492254e-06,
+      "loss": 0.8232,
+      "step": 7019
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9858139241898565e-06,
+      "loss": 0.7675,
+      "step": 7020
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9813465675936737e-06,
+      "loss": 0.9061,
+      "step": 7021
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.97688234359626e-06,
+      "loss": 0.896,
+      "step": 7022
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.972421252832727e-06,
+      "loss": 0.9238,
+      "step": 7023
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9679632959377673e-06,
+      "loss": 0.9739,
+      "step": 7024
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9635084735456093e-06,
+      "loss": 0.6665,
+      "step": 7025
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9590567862900405e-06,
+      "loss": 0.8414,
+      "step": 7026
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.954608234804407e-06,
+      "loss": 0.591,
+      "step": 7027
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9501628197216052e-06,
+      "loss": 0.876,
+      "step": 7028
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9457205416740797e-06,
+      "loss": 0.7309,
+      "step": 7029
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.941281401293838e-06,
+      "loss": 1.0338,
+      "step": 7030
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9368453992124427e-06,
+      "loss": 0.9238,
+      "step": 7031
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.932412536060994e-06,
+      "loss": 0.8029,
+      "step": 7032
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9279828124701703e-06,
+      "loss": 0.6667,
+      "step": 7033
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.923556229070179e-06,
+      "loss": 0.9634,
+      "step": 7034
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.9191327864907915e-06,
+      "loss": 0.6285,
+      "step": 7035
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.914712485361337e-06,
+      "loss": 0.8654,
+      "step": 7036
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.910295326310686e-06,
+      "loss": 0.7845,
+      "step": 7037
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.905881309967276e-06,
+      "loss": 0.8203,
+      "step": 7038
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.901470436959089e-06,
+      "loss": 1.1324,
+      "step": 7039
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.897062707913653e-06,
+      "loss": 0.8213,
+      "step": 7040
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.892658123458067e-06,
+      "loss": 0.9893,
+      "step": 7041
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.888256684218965e-06,
+      "loss": 0.7504,
+      "step": 7042
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.883858390822536e-06,
+      "loss": 0.9773,
+      "step": 7043
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8794632438945425e-06,
+      "loss": 1.0028,
+      "step": 7044
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8750712440602733e-06,
+      "loss": 0.8119,
+      "step": 7045
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8706823919445765e-06,
+      "loss": 1.0428,
+      "step": 7046
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.866296688171863e-06,
+      "loss": 1.0726,
+      "step": 7047
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.861914133366084e-06,
+      "loss": 0.8949,
+      "step": 7048
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8575347281507375e-06,
+      "loss": 1.093,
+      "step": 7049
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.853158473148898e-06,
+      "loss": 1.2378,
+      "step": 7050
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8487853689831647e-06,
+      "loss": 0.8685,
+      "step": 7051
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8444154162757086e-06,
+      "loss": 0.6427,
+      "step": 7052
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8400486156482387e-06,
+      "loss": 1.0747,
+      "step": 7053
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.835684967722019e-06,
+      "loss": 0.9271,
+      "step": 7054
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.831324473117872e-06,
+      "loss": 0.7979,
+      "step": 7055
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8269671324561653e-06,
+      "loss": 1.4153,
+      "step": 7056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8226129463568063e-06,
+      "loss": 0.8634,
+      "step": 7057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.818261915439288e-06,
+      "loss": 1.0803,
+      "step": 7058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8139140403226183e-06,
+      "loss": 1.1467,
+      "step": 7059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8095693216253714e-06,
+      "loss": 0.8438,
+      "step": 7060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8052277599656784e-06,
+      "loss": 0.6963,
+      "step": 7061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.800889355961203e-06,
+      "loss": 0.7418,
+      "step": 7062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7965541102291824e-06,
+      "loss": 1.267,
+      "step": 7063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7922220233863873e-06,
+      "loss": 0.8782,
+      "step": 7064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7878930960491406e-06,
+      "loss": 0.9554,
+      "step": 7065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.783567328833328e-06,
+      "loss": 1.2139,
+      "step": 7066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.779244722354377e-06,
+      "loss": 0.7294,
+      "step": 7067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7749252772272556e-06,
+      "loss": 0.8173,
+      "step": 7068
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7706089940665035e-06,
+      "loss": 0.8959,
+      "step": 7069
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7662958734861883e-06,
+      "loss": 0.8375,
+      "step": 7070
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7619859160999488e-06,
+      "loss": 0.7653,
+      "step": 7071
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.757679122520962e-06,
+      "loss": 1.1548,
+      "step": 7072
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7533754933619564e-06,
+      "loss": 1.0383,
+      "step": 7073
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.749075029235201e-06,
+      "loss": 0.9357,
+      "step": 7074
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7447777307525403e-06,
+      "loss": 1.1501,
+      "step": 7075
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.740483598525334e-06,
+      "loss": 0.905,
+      "step": 7076
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7361926331645247e-06,
+      "loss": 0.9359,
+      "step": 7077
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.73190483528058e-06,
+      "loss": 0.7955,
+      "step": 7078
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.727620205483525e-06,
+      "loss": 0.9331,
+      "step": 7079
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7233387443829417e-06,
+      "loss": 0.7,
+      "step": 7080
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7190604525879515e-06,
+      "loss": 1.1368,
+      "step": 7081
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.714785330707223e-06,
+      "loss": 0.9292,
+      "step": 7082
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.710513379348989e-06,
+      "loss": 0.9358,
+      "step": 7083
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7062445991210077e-06,
+      "loss": 1.0063,
+      "step": 7084
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.7019789906306075e-06,
+      "loss": 0.6375,
+      "step": 7085
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.697716554484661e-06,
+      "loss": 0.8815,
+      "step": 7086
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6934572912895835e-06,
+      "loss": 1.0269,
+      "step": 7087
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.689201201651334e-06,
+      "loss": 0.7516,
+      "step": 7088
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.684948286175437e-06,
+      "loss": 0.7351,
+      "step": 7089
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6806985454669474e-06,
+      "loss": 0.7478,
+      "step": 7090
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6764519801304846e-06,
+      "loss": 1.3184,
+      "step": 7091
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.672208590770206e-06,
+      "loss": 0.7493,
+      "step": 7092
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6679683779898127e-06,
+      "loss": 0.9877,
+      "step": 7093
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6637313423925697e-06,
+      "loss": 0.9172,
+      "step": 7094
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6594974845812805e-06,
+      "loss": 0.9337,
+      "step": 7095
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.655266805158288e-06,
+      "loss": 0.8546,
+      "step": 7096
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6510393047255027e-06,
+      "loss": 0.7338,
+      "step": 7097
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.646814983884363e-06,
+      "loss": 1.1346,
+      "step": 7098
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6425938432358687e-06,
+      "loss": 1.0071,
+      "step": 7099
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6383758833805646e-06,
+      "loss": 0.9827,
+      "step": 7100
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.634161104918542e-06,
+      "loss": 0.9759,
+      "step": 7101
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.629949508449428e-06,
+      "loss": 1.0352,
+      "step": 7102
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6257410945724176e-06,
+      "loss": 0.9812,
+      "step": 7103
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.621535863886235e-06,
+      "loss": 1.0018,
+      "step": 7104
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6173338169891703e-06,
+      "loss": 1.0454,
+      "step": 7105
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6131349544790407e-06,
+      "loss": 0.9469,
+      "step": 7106
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6089392769532177e-06,
+      "loss": 0.729,
+      "step": 7107
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.604746785008627e-06,
+      "loss": 0.8809,
+      "step": 7108
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.6005574792417357e-06,
+      "loss": 1.149,
+      "step": 7109
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5963713602485505e-06,
+      "loss": 1.0721,
+      "step": 7110
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5921884286246367e-06,
+      "loss": 1.1508,
+      "step": 7111
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.588008684965104e-06,
+      "loss": 0.7156,
+      "step": 7112
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5838321298645967e-06,
+      "loss": 1.0011,
+      "step": 7113
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5796587639173253e-06,
+      "loss": 0.8267,
+      "step": 7114
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5754885877170286e-06,
+      "loss": 0.7908,
+      "step": 7115
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5713216018569934e-06,
+      "loss": 0.8441,
+      "step": 7116
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5671578069300708e-06,
+      "loss": 0.9608,
+      "step": 7117
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5629972035286303e-06,
+      "loss": 0.8993,
+      "step": 7118
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5588397922446162e-06,
+      "loss": 1.0596,
+      "step": 7119
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.5546855736694935e-06,
+      "loss": 1.2981,
+      "step": 7120
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.550534548394287e-06,
+      "loss": 1.1083,
+      "step": 7121
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.546386717009569e-06,
+      "loss": 0.7553,
+      "step": 7122
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5422420801054477e-06,
+      "loss": 0.8685,
+      "step": 7123
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.538100638271576e-06,
+      "loss": 0.9067,
+      "step": 7124
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5339623920971637e-06,
+      "loss": 0.5518,
+      "step": 7125
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.529827342170965e-06,
+      "loss": 0.897,
+      "step": 7126
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5256954890812646e-06,
+      "loss": 0.9264,
+      "step": 7127
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5215668334159114e-06,
+      "loss": 0.7782,
+      "step": 7128
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.517441375762289e-06,
+      "loss": 0.9697,
+      "step": 7129
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.513319116707316e-06,
+      "loss": 0.9918,
+      "step": 7130
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5092000568374795e-06,
+      "loss": 0.7343,
+      "step": 7131
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5050841967387913e-06,
+      "loss": 0.7915,
+      "step": 7132
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.5009715369968213e-06,
+      "loss": 0.8801,
+      "step": 7133
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4968620781966794e-06,
+      "loss": 0.8944,
+      "step": 7134
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4927558209230096e-06,
+      "loss": 1.1805,
+      "step": 7135
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4886527657600216e-06,
+      "loss": 0.7356,
+      "step": 7136
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.484552913291452e-06,
+      "loss": 0.7936,
+      "step": 7137
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4804562641005834e-06,
+      "loss": 1.014,
+      "step": 7138
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.476362818770256e-06,
+      "loss": 1.0261,
+      "step": 7139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.472272577882842e-06,
+      "loss": 0.7289,
+      "step": 7140
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.468185542020257e-06,
+      "loss": 0.6796,
+      "step": 7141
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.464101711763972e-06,
+      "loss": 0.9221,
+      "step": 7142
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4600210876949915e-06,
+      "loss": 0.8144,
+      "step": 7143
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.455943670393859e-06,
+      "loss": 0.7892,
+      "step": 7144
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.451869460440684e-06,
+      "loss": 0.9015,
+      "step": 7145
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4477984584150904e-06,
+      "loss": 1.0748,
+      "step": 7146
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.443730664896271e-06,
+      "loss": 0.8507,
+      "step": 7147
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4396660804629517e-06,
+      "loss": 0.8784,
+      "step": 7148
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4356047056933922e-06,
+      "loss": 0.9373,
+      "step": 7149
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4315465411654138e-06,
+      "loss": 0.6967,
+      "step": 7150
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4274915874563734e-06,
+      "loss": 1.1289,
+      "step": 7151
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.423439845143155e-06,
+      "loss": 0.9061,
+      "step": 7152
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4193913148022216e-06,
+      "loss": 0.9913,
+      "step": 7153
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4153459970095523e-06,
+      "loss": 0.7483,
+      "step": 7154
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4113038923406663e-06,
+      "loss": 1.1534,
+      "step": 7155
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4072650013706453e-06,
+      "loss": 1.2086,
+      "step": 7156
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.4032293246740974e-06,
+      "loss": 0.9462,
+      "step": 7157
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3991968628251817e-06,
+      "loss": 0.9543,
+      "step": 7158
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3951676163976004e-06,
+      "loss": 0.9606,
+      "step": 7159
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.391141585964585e-06,
+      "loss": 0.7432,
+      "step": 7160
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3871187720989312e-06,
+      "loss": 0.9042,
+      "step": 7161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3830991753729602e-06,
+      "loss": 1.073,
+      "step": 7162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3790827963585364e-06,
+      "loss": 0.9547,
+      "step": 7163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.375069635627081e-06,
+      "loss": 0.9753,
+      "step": 7164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3710596937495414e-06,
+      "loss": 1.0111,
+      "step": 7165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3670529712964073e-06,
+      "loss": 1.0789,
+      "step": 7166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3630494688377304e-06,
+      "loss": 1.1721,
+      "step": 7167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.35904918694308e-06,
+      "loss": 0.8069,
+      "step": 7168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3550521261815783e-06,
+      "loss": 0.9509,
+      "step": 7169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3510582871218904e-06,
+      "loss": 1.0413,
+      "step": 7170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3470676703322146e-06,
+      "loss": 1.0828,
+      "step": 7171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3430802763803077e-06,
+      "loss": 0.786,
+      "step": 7172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3390961058334494e-06,
+      "loss": 0.9863,
+      "step": 7173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3351151592584665e-06,
+      "loss": 0.8727,
+      "step": 7174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3311374372217365e-06,
+      "loss": 1.1003,
+      "step": 7175
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3271629402891654e-06,
+      "loss": 0.9225,
+      "step": 7176
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.323191669026206e-06,
+      "loss": 0.9996,
+      "step": 7177
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3192236239978544e-06,
+      "loss": 1.181,
+      "step": 7178
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3152588057686414e-06,
+      "loss": 1.2466,
+      "step": 7179
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.311297214902644e-06,
+      "loss": 0.6543,
+      "step": 7180
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.3073388519634863e-06,
+      "loss": 1.2974,
+      "step": 7181
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.303383717514318e-06,
+      "loss": 0.8276,
+      "step": 7182
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.299431812117833e-06,
+      "loss": 0.9377,
+      "step": 7183
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2954831363362824e-06,
+      "loss": 0.8461,
+      "step": 7184
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2915376907314296e-06,
+      "loss": 0.8796,
+      "step": 7185
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.287595475864607e-06,
+      "loss": 0.9169,
+      "step": 7186
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2836564922966707e-06,
+      "loss": 0.9012,
+      "step": 7187
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.279720740588012e-06,
+      "loss": 0.7797,
+      "step": 7188
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.275788221298586e-06,
+      "loss": 1.0246,
+      "step": 7189
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2718589349878667e-06,
+      "loss": 0.9102,
+      "step": 7190
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2679328822148656e-06,
+      "loss": 0.6378,
+      "step": 7191
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2640100635381577e-06,
+      "loss": 1.0916,
+      "step": 7192
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2600904795158334e-06,
+      "loss": 0.8805,
+      "step": 7193
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2561741307055323e-06,
+      "loss": 0.91,
+      "step": 7194
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2522610176644448e-06,
+      "loss": 0.8688,
+      "step": 7195
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.248351140949284e-06,
+      "loss": 0.8835,
+      "step": 7196
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2444445011163053e-06,
+      "loss": 0.8595,
+      "step": 7197
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.240541098721313e-06,
+      "loss": 0.8435,
+      "step": 7198
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2366409343196363e-06,
+      "loss": 0.8718,
+      "step": 7199
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2327440084661673e-06,
+      "loss": 0.9598,
+      "step": 7200
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2288503217153118e-06,
+      "loss": 0.9403,
+      "step": 7201
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2249598746210215e-06,
+      "loss": 0.6918,
+      "step": 7202
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2210726677368037e-06,
+      "loss": 0.8392,
+      "step": 7203
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2171887016156855e-06,
+      "loss": 0.9924,
+      "step": 7204
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.213307976810236e-06,
+      "loss": 1.1575,
+      "step": 7205
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2094304938725718e-06,
+      "loss": 0.7571,
+      "step": 7206
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2055562533543407e-06,
+      "loss": 0.7566,
+      "step": 7207
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.201685255806732e-06,
+      "loss": 0.798,
+      "step": 7208
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.19781750178048e-06,
+      "loss": 0.7682,
+      "step": 7209
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1939529918258456e-06,
+      "loss": 1.126,
+      "step": 7210
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1900917264926296e-06,
+      "loss": 1.0156,
+      "step": 7211
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.186233706330182e-06,
+      "loss": 0.8761,
+      "step": 7212
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.182378931887377e-06,
+      "loss": 0.8066,
+      "step": 7213
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1785274037126405e-06,
+      "loss": 0.7782,
+      "step": 7214
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1746791223539314e-06,
+      "loss": 0.8319,
+      "step": 7215
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1708340883587363e-06,
+      "loss": 0.8915,
+      "step": 7216
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1669923022740986e-06,
+      "loss": 1.0433,
+      "step": 7217
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.163153764646589e-06,
+      "loss": 0.9618,
+      "step": 7218
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1593184760223095e-06,
+      "loss": 0.63,
+      "step": 7219
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1554864369469103e-06,
+      "loss": 0.931,
+      "step": 7220
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1516576479655852e-06,
+      "loss": 0.7636,
+      "step": 7221
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1478321096230435e-06,
+      "loss": 0.8285,
+      "step": 7222
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1440098224635608e-06,
+      "loss": 0.8516,
+      "step": 7223
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1401907870309214e-06,
+      "loss": 0.7874,
+      "step": 7224
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1363750038684655e-06,
+      "loss": 1.0771,
+      "step": 7225
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1325624735190646e-06,
+      "loss": 0.9291,
+      "step": 7226
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.128753196525127e-06,
+      "loss": 0.7236,
+      "step": 7227
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.124947173428607e-06,
+      "loss": 1.0823,
+      "step": 7228
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1211444047709804e-06,
+      "loss": 0.9371,
+      "step": 7229
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1173448910932646e-06,
+      "loss": 0.9425,
+      "step": 7230
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.11354863293603e-06,
+      "loss": 1.1227,
+      "step": 7231
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.109755630839361e-06,
+      "loss": 0.8137,
+      "step": 7232
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.105965885342889e-06,
+      "loss": 0.8767,
+      "step": 7233
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.1021793969857843e-06,
+      "loss": 0.7394,
+      "step": 7234
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.098396166306754e-06,
+      "loss": 0.8486,
+      "step": 7235
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0946161938440345e-06,
+      "loss": 0.9178,
+      "step": 7236
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.090839480135412e-06,
+      "loss": 0.6843,
+      "step": 7237
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.08706602571819e-06,
+      "loss": 1.1622,
+      "step": 7238
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0832958311292228e-06,
+      "loss": 1.0996,
+      "step": 7239
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0795288969048975e-06,
+      "loss": 1.17,
+      "step": 7240
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.075765223581133e-06,
+      "loss": 0.9828,
+      "step": 7241
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0720048116933956e-06,
+      "loss": 0.9556,
+      "step": 7242
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0682476617766762e-06,
+      "loss": 0.8092,
+      "step": 7243
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0644937743655e-06,
+      "loss": 0.7482,
+      "step": 7244
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0607431499939404e-06,
+      "loss": 0.7358,
+      "step": 7245
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0569957891955984e-06,
+      "loss": 0.7042,
+      "step": 7246
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.053251692503605e-06,
+      "loss": 0.9675,
+      "step": 7247
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.04951086045064e-06,
+      "loss": 0.9424,
+      "step": 7248
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0457732935689173e-06,
+      "loss": 0.7708,
+      "step": 7249
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.042038992390169e-06,
+      "loss": 0.6572,
+      "step": 7250
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.038307957445687e-06,
+      "loss": 1.085,
+      "step": 7251
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0345801892662775e-06,
+      "loss": 1.057,
+      "step": 7252
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0308556883822995e-06,
+      "loss": 0.7986,
+      "step": 7253
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0271344553236316e-06,
+      "loss": 0.8533,
+      "step": 7254
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.023416490619692e-06,
+      "loss": 0.9174,
+      "step": 7255
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0197017947994473e-06,
+      "loss": 0.6773,
+      "step": 7256
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0159903683913792e-06,
+      "loss": 0.8447,
+      "step": 7257
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0122822119235134e-06,
+      "loss": 1.1753,
+      "step": 7258
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0085773259234166e-06,
+      "loss": 0.8417,
+      "step": 7259
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.0048757109181806e-06,
+      "loss": 1.0326,
+      "step": 7260
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.001177367434423e-06,
+      "loss": 0.8967,
+      "step": 7261
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9974822959983263e-06,
+      "loss": 0.8239,
+      "step": 7262
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9937904971355863e-06,
+      "loss": 0.811,
+      "step": 7263
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9901019713714247e-06,
+      "loss": 0.7639,
+      "step": 7264
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9864167192306186e-06,
+      "loss": 0.9147,
+      "step": 7265
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.982734741237463e-06,
+      "loss": 1.1655,
+      "step": 7266
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9790560379158026e-06,
+      "loss": 0.9888,
+      "step": 7267
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9753806097890045e-06,
+      "loss": 1.0668,
+      "step": 7268
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9717084573799614e-06,
+      "loss": 1.001,
+      "step": 7269
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9680395812111284e-06,
+      "loss": 1.1013,
+      "step": 7270
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.964373981804468e-06,
+      "loss": 1.2163,
+      "step": 7271
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9607116596814827e-06,
+      "loss": 0.8829,
+      "step": 7272
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9570526153632196e-06,
+      "loss": 0.6902,
+      "step": 7273
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9533968493702454e-06,
+      "loss": 0.9313,
+      "step": 7274
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9497443622226692e-06,
+      "loss": 0.787,
+      "step": 7275
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9460951544401364e-06,
+      "loss": 0.8329,
+      "step": 7276
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9424492265418147e-06,
+      "loss": 1.0128,
+      "step": 7277
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9388065790464096e-06,
+      "loss": 0.8698,
+      "step": 7278
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9351672124721664e-06,
+      "loss": 0.9161,
+      "step": 7279
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9315311273368523e-06,
+      "loss": 0.8286,
+      "step": 7280
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.927898324157784e-06,
+      "loss": 0.8001,
+      "step": 7281
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9242688034517925e-06,
+      "loss": 0.6874,
+      "step": 7282
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9206425657352504e-06,
+      "loss": 1.1957,
+      "step": 7283
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.917019611524071e-06,
+      "loss": 0.895,
+      "step": 7284
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9133999413336876e-06,
+      "loss": 1.0966,
+      "step": 7285
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9097835556790677e-06,
+      "loss": 0.972,
+      "step": 7286
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.906170455074724e-06,
+      "loss": 0.6432,
+      "step": 7287
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.902560640034684e-06,
+      "loss": 0.689,
+      "step": 7288
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8989541110725207e-06,
+      "loss": 1.1553,
+      "step": 7289
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8953508687013399e-06,
+      "loss": 0.92,
+      "step": 7290
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8917509134337736e-06,
+      "loss": 1.0101,
+      "step": 7291
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.888154245781984e-06,
+      "loss": 0.9849,
+      "step": 7292
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8845608662576757e-06,
+      "loss": 0.6425,
+      "step": 7293
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8809707753720707e-06,
+      "loss": 0.8788,
+      "step": 7294
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.877383973635946e-06,
+      "loss": 0.8486,
+      "step": 7295
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8738004615595856e-06,
+      "loss": 1.0396,
+      "step": 7296
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8702202396528173e-06,
+      "loss": 0.8699,
+      "step": 7297
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8666433084250062e-06,
+      "loss": 1.0508,
+      "step": 7298
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8630696683850397e-06,
+      "loss": 0.9746,
+      "step": 7299
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.859499320041333e-06,
+      "loss": 1.0317,
+      "step": 7300
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.855932263901855e-06,
+      "loss": 0.8889,
+      "step": 7301
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8523685004740777e-06,
+      "loss": 1.158,
+      "step": 7302
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8488080302650257e-06,
+      "loss": 0.8461,
+      "step": 7303
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8452508537812524e-06,
+      "loss": 0.9805,
+      "step": 7304
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8416969715288306e-06,
+      "loss": 1.0149,
+      "step": 7305
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.83814638401337e-06,
+      "loss": 0.9459,
+      "step": 7306
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.834599091740022e-06,
+      "loss": 1.1085,
+      "step": 7307
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8310550952134548e-06,
+      "loss": 0.8566,
+      "step": 7308
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8275143949378764e-06,
+      "loss": 0.9904,
+      "step": 7309
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8239769914170203e-06,
+      "loss": 0.8542,
+      "step": 7310
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8204428851541528e-06,
+      "loss": 0.994,
+      "step": 7311
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8169120766520776e-06,
+      "loss": 1.0305,
+      "step": 7312
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8133845664131204e-06,
+      "loss": 0.9556,
+      "step": 7313
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8098603549391357e-06,
+      "loss": 0.8696,
+      "step": 7314
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.806339442731525e-06,
+      "loss": 0.8317,
+      "step": 7315
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8028218302911959e-06,
+      "loss": 0.7953,
+      "step": 7316
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7993075181186092e-06,
+      "loss": 0.9217,
+      "step": 7317
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7957965067137483e-06,
+      "loss": 0.9465,
+      "step": 7318
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.792288796576122e-06,
+      "loss": 0.9817,
+      "step": 7319
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.78878438820477e-06,
+      "loss": 0.886,
+      "step": 7320
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7852832820982713e-06,
+      "loss": 1.217,
+      "step": 7321
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7817854787547217e-06,
+      "loss": 0.9933,
+      "step": 7322
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7782909786717622e-06,
+      "loss": 0.7114,
+      "step": 7323
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7747997823465534e-06,
+      "loss": 0.8083,
+      "step": 7324
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7713118902757842e-06,
+      "loss": 1.0698,
+      "step": 7325
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7678273029556852e-06,
+      "loss": 0.9889,
+      "step": 7326
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7643460208820072e-06,
+      "loss": 0.9172,
+      "step": 7327
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7608680445500264e-06,
+      "loss": 0.842,
+      "step": 7328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7573933744545607e-06,
+      "loss": 0.9728,
+      "step": 7329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7539220110899563e-06,
+      "loss": 0.8112,
+      "step": 7330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7504539549500737e-06,
+      "loss": 0.6877,
+      "step": 7331
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7469892065283266e-06,
+      "loss": 0.8171,
+      "step": 7332
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7435277663176402e-06,
+      "loss": 0.8284,
+      "step": 7333
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7400696348104706e-06,
+      "loss": 1.301,
+      "step": 7334
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.736614812498813e-06,
+      "loss": 0.877,
+      "step": 7335
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7331632998741771e-06,
+      "loss": 0.895,
+      "step": 7336
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7297150974276226e-06,
+      "loss": 0.9279,
+      "step": 7337
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7262702056497186e-06,
+      "loss": 0.8431,
+      "step": 7338
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7228286250305703e-06,
+      "loss": 0.7775,
+      "step": 7339
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7193903560598134e-06,
+      "loss": 1.045,
+      "step": 7340
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7159553992266125e-06,
+      "loss": 0.6823,
+      "step": 7341
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.712523755019657e-06,
+      "loss": 0.9764,
+      "step": 7342
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7090954239271679e-06,
+      "loss": 0.8618,
+      "step": 7343
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.705670406436899e-06,
+      "loss": 0.7974,
+      "step": 7344
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.7022487030361217e-06,
+      "loss": 0.8955,
+      "step": 7345
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6988303142116496e-06,
+      "loss": 1.3413,
+      "step": 7346
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6954152404498102e-06,
+      "loss": 0.8636,
+      "step": 7347
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6920034822364761e-06,
+      "loss": 0.8587,
+      "step": 7348
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.688595040057031e-06,
+      "loss": 1.1176,
+      "step": 7349
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6851899143963956e-06,
+      "loss": 0.855,
+      "step": 7350
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.681788105739021e-06,
+      "loss": 0.7397,
+      "step": 7351
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6783896145688844e-06,
+      "loss": 0.745,
+      "step": 7352
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6749944413694818e-06,
+      "loss": 1.1693,
+      "step": 7353
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6716025866238549e-06,
+      "loss": 0.8335,
+      "step": 7354
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6682140508145593e-06,
+      "loss": 0.8936,
+      "step": 7355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6648288344236763e-06,
+      "loss": 0.8076,
+      "step": 7356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661446937932834e-06,
+      "loss": 1.0837,
+      "step": 7357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580683618231728e-06,
+      "loss": 0.8297,
+      "step": 7358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6546931065753528e-06,
+      "loss": 0.9949,
+      "step": 7359
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.651321172669587e-06,
+      "loss": 1.0596,
+      "step": 7360
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.647952560585586e-06,
+      "loss": 0.757,
+      "step": 7361
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6445872708026195e-06,
+      "loss": 0.7896,
+      "step": 7362
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6412253037994568e-06,
+      "loss": 0.8746,
+      "step": 7363
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6378666600544045e-06,
+      "loss": 0.7599,
+      "step": 7364
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6345113400453055e-06,
+      "loss": 0.8052,
+      "step": 7365
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6311593442495192e-06,
+      "loss": 0.9285,
+      "step": 7366
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6278106731439284e-06,
+      "loss": 0.7848,
+      "step": 7367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6244653272049576e-06,
+      "loss": 1.0745,
+      "step": 7368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6211233069085452e-06,
+      "loss": 0.9353,
+      "step": 7369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6177846127301611e-06,
+      "loss": 0.8698,
+      "step": 7370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6144492451448085e-06,
+      "loss": 0.6526,
+      "step": 7371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6111172046270052e-06,
+      "loss": 0.8337,
+      "step": 7372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6077884916507995e-06,
+      "loss": 0.6832,
+      "step": 7373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6044631066897741e-06,
+      "loss": 1.1578,
+      "step": 7374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6011410502170226e-06,
+      "loss": 1.0747,
+      "step": 7375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.597822322705189e-06,
+      "loss": 0.8126,
+      "step": 7376
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.594506924626421e-06,
+      "loss": 0.9065,
+      "step": 7377
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5911948564523966e-06,
+      "loss": 0.9208,
+      "step": 7378
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.587886118654333e-06,
+      "loss": 0.6493,
+      "step": 7379
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5845807117029627e-06,
+      "loss": 1.2303,
+      "step": 7380
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5812786360685394e-06,
+      "loss": 0.9221,
+      "step": 7381
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5779798922208627e-06,
+      "loss": 0.9968,
+      "step": 7382
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.574684480629235e-06,
+      "loss": 0.7366,
+      "step": 7383
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5713924017625003e-06,
+      "loss": 0.7294,
+      "step": 7384
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5681036560890234e-06,
+      "loss": 1.0027,
+      "step": 7385
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5648182440766962e-06,
+      "loss": 1.0502,
+      "step": 7386
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5615361661929284e-06,
+      "loss": 0.9291,
+      "step": 7387
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5582574229046715e-06,
+      "loss": 1.0303,
+      "step": 7388
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5549820146783829e-06,
+      "loss": 0.8608,
+      "step": 7389
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5517099419800646e-06,
+      "loss": 0.6239,
+      "step": 7390
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5484412052752306e-06,
+      "loss": 0.804,
+      "step": 7391
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5451758050289228e-06,
+      "loss": 0.9493,
+      "step": 7392
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5419137417057167e-06,
+      "loss": 0.9108,
+      "step": 7393
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5386550157697022e-06,
+      "loss": 0.9622,
+      "step": 7394
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5353996276844972e-06,
+      "loss": 0.6417,
+      "step": 7395
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5321475779132538e-06,
+      "loss": 0.6584,
+      "step": 7396
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5288988669186322e-06,
+      "loss": 1.2905,
+      "step": 7397
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5256534951628322e-06,
+      "loss": 0.8204,
+      "step": 7398
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.522411463107576e-06,
+      "loss": 0.9235,
+      "step": 7399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5191727712141086e-06,
+      "loss": 0.8954,
+      "step": 7400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.515937419943192e-06,
+      "loss": 0.8256,
+      "step": 7401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5127054097551275e-06,
+      "loss": 0.8768,
+      "step": 7402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5094767411097305e-06,
+      "loss": 1.0543,
+      "step": 7403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.5062514144663476e-06,
+      "loss": 0.9879,
+      "step": 7404
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.503029430283842e-06,
+      "loss": 0.9459,
+      "step": 7405
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4998107890206086e-06,
+      "loss": 0.85,
+      "step": 7406
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.496595491134567e-06,
+      "loss": 0.6486,
+      "step": 7407
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.493383537083154e-06,
+      "loss": 0.9581,
+      "step": 7408
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4901749273233344e-06,
+      "loss": 0.8268,
+      "step": 7409
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4869696623116048e-06,
+      "loss": 0.8842,
+      "step": 7410
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4837677425039691e-06,
+      "loss": 0.8196,
+      "step": 7411
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.480569168355972e-06,
+      "loss": 0.95,
+      "step": 7412
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4773739403226738e-06,
+      "loss": 0.8431,
+      "step": 7413
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4741820588586641e-06,
+      "loss": 0.9065,
+      "step": 7414
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4709935244180406e-06,
+      "loss": 0.8525,
+      "step": 7415
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4678083374544516e-06,
+      "loss": 1.0928,
+      "step": 7416
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4646264984210428e-06,
+      "loss": 0.6051,
+      "step": 7417
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.461448007770505e-06,
+      "loss": 0.8666,
+      "step": 7418
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4582728659550377e-06,
+      "loss": 1.0736,
+      "step": 7419
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4551010734263653e-06,
+      "loss": 0.7828,
+      "step": 7420
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4519326306357467e-06,
+      "loss": 0.8989,
+      "step": 7421
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4487675380339516e-06,
+      "loss": 1.0171,
+      "step": 7422
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4456057960712783e-06,
+      "loss": 0.9032,
+      "step": 7423
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.44244740519755e-06,
+      "loss": 0.8851,
+      "step": 7424
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.439292365862116e-06,
+      "loss": 0.7735,
+      "step": 7425
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4361406785138393e-06,
+      "loss": 0.9456,
+      "step": 7426
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4329923436011138e-06,
+      "loss": 0.6304,
+      "step": 7427
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.429847361571851e-06,
+      "loss": 0.885,
+      "step": 7428
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.42670573287349e-06,
+      "loss": 0.9379,
+      "step": 7429
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4235674579529928e-06,
+      "loss": 0.806,
+      "step": 7430
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4204325372568356e-06,
+      "loss": 0.736,
+      "step": 7431
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4173009712310337e-06,
+      "loss": 0.7576,
+      "step": 7432
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4141727603211113e-06,
+      "loss": 0.973,
+      "step": 7433
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4110479049721154e-06,
+      "loss": 0.9348,
+      "step": 7434
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.407926405628629e-06,
+      "loss": 1.0631,
+      "step": 7435
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4048082627347414e-06,
+      "loss": 1.1066,
+      "step": 7436
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.4016934767340727e-06,
+      "loss": 0.9908,
+      "step": 7437
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.398582048069763e-06,
+      "loss": 0.8134,
+      "step": 7438
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3954739771844806e-06,
+      "loss": 1.1294,
+      "step": 7439
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3923692645204072e-06,
+      "loss": 0.8676,
+      "step": 7440
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3892679105192569e-06,
+      "loss": 0.9517,
+      "step": 7441
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3861699156222541e-06,
+      "loss": 0.769,
+      "step": 7442
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3830752802701491e-06,
+      "loss": 0.9651,
+      "step": 7443
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3799840049032259e-06,
+      "loss": 1.2222,
+      "step": 7444
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3768960899612687e-06,
+      "loss": 1.2638,
+      "step": 7445
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3738115358836096e-06,
+      "loss": 0.9828,
+      "step": 7446
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3707303431090779e-06,
+      "loss": 0.762,
+      "step": 7447
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.367652512076037e-06,
+      "loss": 1.0527,
+      "step": 7448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3645780432223782e-06,
+      "loss": 0.7706,
+      "step": 7449
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3615069369854989e-06,
+      "loss": 1.0209,
+      "step": 7450
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3584391938023271e-06,
+      "loss": 0.7627,
+      "step": 7451
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3553748141093114e-06,
+      "loss": 1.0474,
+      "step": 7452
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3523137983424277e-06,
+      "loss": 1.0702,
+      "step": 7453
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.3492561469371584e-06,
+      "loss": 0.9683,
+      "step": 7454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3462018603285248e-06,
+      "loss": 0.9741,
+      "step": 7455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3431509389510543e-06,
+      "loss": 0.8807,
+      "step": 7456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3401033832388082e-06,
+      "loss": 0.9885,
+      "step": 7457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.337059193625359e-06,
+      "loss": 0.9229,
+      "step": 7458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.334018370543802e-06,
+      "loss": 0.9504,
+      "step": 7459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3309809144267632e-06,
+      "loss": 0.6906,
+      "step": 7460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3279468257063748e-06,
+      "loss": 1.3496,
+      "step": 7461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3249161048142972e-06,
+      "loss": 1.1907,
+      "step": 7462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3218887521817185e-06,
+      "loss": 0.9117,
+      "step": 7463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3188647682393362e-06,
+      "loss": 1.0399,
+      "step": 7464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.31584415341737e-06,
+      "loss": 0.9501,
+      "step": 7465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.312826908145573e-06,
+      "loss": 0.9232,
+      "step": 7466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.309813032853202e-06,
+      "loss": 0.9144,
+      "step": 7467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3068025279690422e-06,
+      "loss": 0.7246,
+      "step": 7468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3037953939214031e-06,
+      "loss": 1.0978,
+      "step": 7469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.3007916311381012e-06,
+      "loss": 0.9501,
+      "step": 7470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2977912400464948e-06,
+      "loss": 0.9879,
+      "step": 7471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.294794221073442e-06,
+      "loss": 1.0002,
+      "step": 7472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2918005746453299e-06,
+      "loss": 1.1056,
+      "step": 7473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2888103011880704e-06,
+      "loss": 0.8605,
+      "step": 7474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2858234011270843e-06,
+      "loss": 0.6547,
+      "step": 7475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2828398748873177e-06,
+      "loss": 0.8401,
+      "step": 7476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.279859722893248e-06,
+      "loss": 0.9275,
+      "step": 7477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2768829455688469e-06,
+      "loss": 0.983,
+      "step": 7478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2739095433376313e-06,
+      "loss": 0.946,
+      "step": 7479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2709395166226295e-06,
+      "loss": 0.8555,
+      "step": 7480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2679728658463813e-06,
+      "loss": 0.9198,
+      "step": 7481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2650095914309546e-06,
+      "loss": 1.0323,
+      "step": 7482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2620496937979399e-06,
+      "loss": 0.5256,
+      "step": 7483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2590931733684336e-06,
+      "loss": 0.7569,
+      "step": 7484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2561400305630688e-06,
+      "loss": 1.0354,
+      "step": 7485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2531902658019874e-06,
+      "loss": 0.5504,
+      "step": 7486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2502438795048477e-06,
+      "loss": 0.7245,
+      "step": 7487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2473008720908425e-06,
+      "loss": 1.1093,
+      "step": 7488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2443612439786672e-06,
+      "loss": 0.809,
+      "step": 7489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.241424995586543e-06,
+      "loss": 0.9554,
+      "step": 7490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2384921273322137e-06,
+      "loss": 0.7255,
+      "step": 7491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2355626396329367e-06,
+      "loss": 0.8943,
+      "step": 7492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2326365329054929e-06,
+      "loss": 0.8478,
+      "step": 7493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2297138075661795e-06,
+      "loss": 0.8665,
+      "step": 7494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.226794464030817e-06,
+      "loss": 0.9998,
+      "step": 7495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2238785027147315e-06,
+      "loss": 0.7264,
+      "step": 7496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2209659240327881e-06,
+      "loss": 0.7296,
+      "step": 7497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2180567283993527e-06,
+      "loss": 0.8242,
+      "step": 7498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2151509162283193e-06,
+      "loss": 1.2013,
+      "step": 7499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2122484879331015e-06,
+      "loss": 0.9905,
+      "step": 7500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2093494439266245e-06,
+      "loss": 0.8706,
+      "step": 7501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2064537846213364e-06,
+      "loss": 0.8314,
+      "step": 7502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2035615104292074e-06,
+      "loss": 0.9911,
+      "step": 7503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.2006726217617142e-06,
+      "loss": 1.022,
+      "step": 7504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1977871190298668e-06,
+      "loss": 0.7563,
+      "step": 7505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.194905002644181e-06,
+      "loss": 0.8748,
+      "step": 7506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1920262730146987e-06,
+      "loss": 1.1238,
+      "step": 7507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1891509305509836e-06,
+      "loss": 1.1107,
+      "step": 7508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1862789756621028e-06,
+      "loss": 0.9253,
+      "step": 7509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1834104087566495e-06,
+      "loss": 1.1127,
+      "step": 7510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1805452302427411e-06,
+      "loss": 0.7772,
+      "step": 7511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1776834405280018e-06,
+      "loss": 0.6899,
+      "step": 7512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1748250400195838e-06,
+      "loss": 0.721,
+      "step": 7513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1719700291241502e-06,
+      "loss": 1.1257,
+      "step": 7514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1691184082478823e-06,
+      "loss": 1.1086,
+      "step": 7515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1662701777964857e-06,
+      "loss": 1.1302,
+      "step": 7516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1634253381751753e-06,
+      "loss": 1.0685,
+      "step": 7517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1605838897886828e-06,
+      "loss": 0.6815,
+      "step": 7518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1577458330412711e-06,
+      "loss": 0.7314,
+      "step": 7519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.154911168336703e-06,
+      "loss": 0.8027,
+      "step": 7520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1520798960782704e-06,
+      "loss": 0.8241,
+      "step": 7521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1492520166687837e-06,
+      "loss": 0.8463,
+      "step": 7522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1464275305105576e-06,
+      "loss": 0.9481,
+      "step": 7523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1436064380054373e-06,
+      "loss": 0.8947,
+      "step": 7524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1407887395547795e-06,
+      "loss": 0.9048,
+      "step": 7525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1379744355594552e-06,
+      "loss": 0.9469,
+      "step": 7526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1351635264198607e-06,
+      "loss": 1.0562,
+      "step": 7527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1323560125359068e-06,
+      "loss": 0.7686,
+      "step": 7528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.12955189430701e-06,
+      "loss": 0.9235,
+      "step": 7529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1267511721321205e-06,
+      "loss": 0.9124,
+      "step": 7530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1239538464096978e-06,
+      "loss": 1.0555,
+      "step": 7531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.121159917537712e-06,
+      "loss": 0.7162,
+      "step": 7532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1183693859136563e-06,
+      "loss": 0.7555,
+      "step": 7533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.115582251934552e-06,
+      "loss": 1.0155,
+      "step": 7534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.11279851599691e-06,
+      "loss": 1.1425,
+      "step": 7535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1100181784967855e-06,
+      "loss": 0.808,
+      "step": 7536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.1072412398297289e-06,
+      "loss": 0.9252,
+      "step": 7537
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1044677003908184e-06,
+      "loss": 0.7794,
+      "step": 7538
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.1016975605746494e-06,
+      "loss": 0.7228,
+      "step": 7539
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0989308207753263e-06,
+      "loss": 1.1869,
+      "step": 7540
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0961674813864758e-06,
+      "loss": 0.9888,
+      "step": 7541
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0934075428012392e-06,
+      "loss": 0.7913,
+      "step": 7542
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0906510054122715e-06,
+      "loss": 0.7426,
+      "step": 7543
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0878978696117482e-06,
+      "loss": 0.7255,
+      "step": 7544
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0851481357913556e-06,
+      "loss": 0.8387,
+      "step": 7545
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0824018043423006e-06,
+      "loss": 0.996,
+      "step": 7546
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0796588756553039e-06,
+      "loss": 1.0892,
+      "step": 7547
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.076919350120606e-06,
+      "loss": 0.8671,
+      "step": 7548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0741832281279534e-06,
+      "loss": 0.7537,
+      "step": 7549
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0714505100666211e-06,
+      "loss": 0.804,
+      "step": 7550
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0687211963253897e-06,
+      "loss": 0.8589,
+      "step": 7551
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0659952872925622e-06,
+      "loss": 0.8337,
+      "step": 7552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0632727833559509e-06,
+      "loss": 0.8244,
+      "step": 7553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0605536849028873e-06,
+      "loss": 0.986,
+      "step": 7554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0578379923202236e-06,
+      "loss": 1.1664,
+      "step": 7555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.055125705994317e-06,
+      "loss": 0.9642,
+      "step": 7556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0524168263110423e-06,
+      "loss": 0.7352,
+      "step": 7557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0497113536557995e-06,
+      "loss": 0.9674,
+      "step": 7558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0470092884134892e-06,
+      "loss": 1.047,
+      "step": 7559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0443106309685425e-06,
+      "loss": 0.8564,
+      "step": 7560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0416153817048912e-06,
+      "loss": 1.1482,
+      "step": 7561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0389235410059983e-06,
+      "loss": 0.8668,
+      "step": 7562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.036235109254821e-06,
+      "loss": 0.8548,
+      "step": 7563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.033550086833851e-06,
+      "loss": 0.5997,
+      "step": 7564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0308684741250824e-06,
+      "loss": 0.9579,
+      "step": 7565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.028190271510035e-06,
+      "loss": 1.1592,
+      "step": 7566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0255154793697319e-06,
+      "loss": 0.8671,
+      "step": 7567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0228440980847158e-06,
+      "loss": 0.816,
+      "step": 7568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0201761280350526e-06,
+      "loss": 0.8148,
+      "step": 7569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0175115696003047e-06,
+      "loss": 0.9918,
+      "step": 7570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0148504231595635e-06,
+      "loss": 1.0125,
+      "step": 7571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0121926890914345e-06,
+      "loss": 0.8479,
+      "step": 7572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0095383677740267e-06,
+      "loss": 0.8252,
+      "step": 7573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0068874595849764e-06,
+      "loss": 0.9399,
+      "step": 7574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0042399649014294e-06,
+      "loss": 0.8322,
+      "step": 7575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.0015958841000456e-06,
+      "loss": 1.415,
+      "step": 7576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.989552175569934e-07,
+      "loss": 0.8746,
+      "step": 7577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.963179656479666e-07,
+      "loss": 1.0566,
+      "step": 7578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.936841287481623e-07,
+      "loss": 0.7878,
+      "step": 7579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.910537072323062e-07,
+      "loss": 0.9368,
+      "step": 7580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.884267014746206e-07,
+      "loss": 0.8008,
+      "step": 7581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.858031118488514e-07,
+      "loss": 1.0333,
+      "step": 7582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.83182938728261e-07,
+      "loss": 1.1145,
+      "step": 7583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.805661824856204e-07,
+      "loss": 1.0997,
+      "step": 7584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.779528434932128e-07,
+      "loss": 0.8114,
+      "step": 7585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.753429221228433e-07,
+      "loss": 0.7216,
+      "step": 7586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.727364187458233e-07,
+      "loss": 1.045,
+      "step": 7587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.701333337329782e-07,
+      "loss": 0.9266,
+      "step": 7588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.675336674546565e-07,
+      "loss": 0.9519,
+      "step": 7589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.649374202807093e-07,
+      "loss": 0.6986,
+      "step": 7590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.623445925805024e-07,
+      "loss": 0.8895,
+      "step": 7591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.597551847229242e-07,
+      "loss": 0.9139,
+      "step": 7592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.57169197076363e-07,
+      "loss": 0.9462,
+      "step": 7593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.545866300087358e-07,
+      "loss": 0.88,
+      "step": 7594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.520074838874599e-07,
+      "loss": 1.0256,
+      "step": 7595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.494317590794693e-07,
+      "loss": 1.428,
+      "step": 7596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.468594559512212e-07,
+      "loss": 0.9032,
+      "step": 7597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.442905748686698e-07,
+      "loss": 0.9463,
+      "step": 7598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.417251161972895e-07,
+      "loss": 1.0468,
+      "step": 7599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.391630803020773e-07,
+      "loss": 0.7579,
+      "step": 7600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.366044675475249e-07,
+      "loss": 0.8611,
+      "step": 7601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.340492782976524e-07,
+      "loss": 0.7463,
+      "step": 7602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.314975129159881e-07,
+      "loss": 1.0903,
+      "step": 7603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.289491717655696e-07,
+      "loss": 0.8837,
+      "step": 7604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.264042552089485e-07,
+      "loss": 0.6212,
+      "step": 7605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.238627636081937e-07,
+      "loss": 0.969,
+      "step": 7606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.213246973248823e-07,
+      "loss": 0.9937,
+      "step": 7607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.187900567201064e-07,
+      "loss": 1.2112,
+      "step": 7608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.162588421544715e-07,
+      "loss": 0.7652,
+      "step": 7609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.1373105398809e-07,
+      "loss": 0.8064,
+      "step": 7610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.112066925805934e-07,
+      "loss": 1.0221,
+      "step": 7611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.086857582911223e-07,
+      "loss": 0.8702,
+      "step": 7612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.061682514783287e-07,
+      "loss": 0.782,
+      "step": 7613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.036541725003844e-07,
+      "loss": 0.9419,
+      "step": 7614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.011435217149644e-07,
+      "loss": 0.9348,
+      "step": 7615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.986362994792552e-07,
+      "loss": 0.8901,
+      "step": 7616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.961325061499715e-07,
+      "loss": 0.6636,
+      "step": 7617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.936321420833172e-07,
+      "loss": 1.0994,
+      "step": 7618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.911352076350244e-07,
+      "loss": 0.8402,
+      "step": 7619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.886417031603339e-07,
+      "loss": 0.6521,
+      "step": 7620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.861516290139926e-07,
+      "loss": 0.9575,
+      "step": 7621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.836649855502699e-07,
+      "loss": 1.0381,
+      "step": 7622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.811817731229382e-07,
+      "loss": 0.796,
+      "step": 7623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.78701992085279e-07,
+      "loss": 0.8944,
+      "step": 7624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.762256427901017e-07,
+      "loss": 1.0066,
+      "step": 7625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.737527255897133e-07,
+      "loss": 1.099,
+      "step": 7626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.712832408359295e-07,
+      "loss": 1.0066,
+      "step": 7627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.688171888800945e-07,
+      "loss": 0.9618,
+      "step": 7628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.663545700730469e-07,
+      "loss": 1.0231,
+      "step": 7629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.638953847651454e-07,
+      "loss": 0.853,
+      "step": 7630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.61439633306263e-07,
+      "loss": 0.724,
+      "step": 7631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.589873160457784e-07,
+      "loss": 0.9089,
+      "step": 7632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.565384333325793e-07,
+      "loss": 1.1865,
+      "step": 7633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.540929855150731e-07,
+      "loss": 1.0223,
+      "step": 7634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.516509729411703e-07,
+      "loss": 0.8762,
+      "step": 7635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.492123959583014e-07,
+      "loss": 0.8823,
+      "step": 7636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.467772549134e-07,
+      "loss": 0.9514,
+      "step": 7637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.443455501529141e-07,
+      "loss": 0.7471,
+      "step": 7638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.419172820228055e-07,
+      "loss": 0.6796,
+      "step": 7639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.394924508685426e-07,
+      "loss": 0.7565,
+      "step": 7640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.370710570351048e-07,
+      "loss": 0.8662,
+      "step": 7641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.346531008669861e-07,
+      "loss": 0.8509,
+      "step": 7642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.322385827081946e-07,
+      "loss": 0.8849,
+      "step": 7643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.298275029022335e-07,
+      "loss": 1.1902,
+      "step": 7644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.274198617921391e-07,
+      "loss": 0.7954,
+      "step": 7645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.250156597204378e-07,
+      "loss": 1.1624,
+      "step": 7646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.226148970291836e-07,
+      "loss": 0.7234,
+      "step": 7647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.202175740599282e-07,
+      "loss": 0.9009,
+      "step": 7648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.178236911537407e-07,
+      "loss": 0.9786,
+      "step": 7649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.154332486511984e-07,
+      "loss": 0.8998,
+      "step": 7650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.13046246892396e-07,
+      "loss": 0.6785,
+      "step": 7651
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.106626862169231e-07,
+      "loss": 0.9479,
+      "step": 7652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.082825669638972e-07,
+      "loss": 0.9819,
+      "step": 7653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.059058894719335e-07,
+      "loss": 0.976,
+      "step": 7654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.035326540791643e-07,
+      "loss": 1.0188,
+      "step": 7655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.011628611232308e-07,
+      "loss": 0.8659,
+      "step": 7656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.987965109412854e-07,
+      "loss": 1.1158,
+      "step": 7657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.964336038699838e-07,
+      "loss": 0.8024,
+      "step": 7658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.94074140245507e-07,
+      "loss": 0.9098,
+      "step": 7659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.917181204035257e-07,
+      "loss": 0.8245,
+      "step": 7660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.893655446792381e-07,
+      "loss": 0.7738,
+      "step": 7661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.87016413407346e-07,
+      "loss": 0.8777,
+      "step": 7662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.846707269220543e-07,
+      "loss": 0.9568,
+      "step": 7663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.823284855570934e-07,
+      "loss": 0.8274,
+      "step": 7664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.799896896456882e-07,
+      "loss": 1.0098,
+      "step": 7665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.776543395205782e-07,
+      "loss": 0.9878,
+      "step": 7666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.753224355140198e-07,
+      "loss": 0.9138,
+      "step": 7667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.729939779577699e-07,
+      "loss": 0.9271,
+      "step": 7668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.706689671830969e-07,
+      "loss": 0.8877,
+      "step": 7669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.683474035207888e-07,
+      "loss": 0.9495,
+      "step": 7670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.660292873011288e-07,
+      "loss": 0.8188,
+      "step": 7671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.63714618853914e-07,
+      "loss": 0.7805,
+      "step": 7672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.61403398508459e-07,
+      "loss": 1.1514,
+      "step": 7673
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.590956265935756e-07,
+      "loss": 0.7224,
+      "step": 7674
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.567913034375956e-07,
+      "loss": 0.833,
+      "step": 7675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.544904293683514e-07,
+      "loss": 0.8162,
+      "step": 7676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.52193004713192e-07,
+      "loss": 1.068,
+      "step": 7677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.4989902979897e-07,
+      "loss": 0.8918,
+      "step": 7678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.476085049520549e-07,
+      "loss": 1.1937,
+      "step": 7679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.453214304983109e-07,
+      "loss": 1.0005,
+      "step": 7680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.430378067631305e-07,
+      "loss": 0.6679,
+      "step": 7681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.407576340713984e-07,
+      "loss": 0.9109,
+      "step": 7682
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.38480912747519e-07,
+      "loss": 0.7808,
+      "step": 7683
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.362076431154024e-07,
+      "loss": 1.0095,
+      "step": 7684
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.339378254984652e-07,
+      "loss": 0.9139,
+      "step": 7685
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.316714602196323e-07,
+      "loss": 0.8279,
+      "step": 7686
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.294085476013484e-07,
+      "loss": 1.0303,
+      "step": 7687
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.27149087965548e-07,
+      "loss": 1.0605,
+      "step": 7688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.248930816336929e-07,
+      "loss": 0.7816,
+      "step": 7689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.226405289267435e-07,
+      "loss": 0.8091,
+      "step": 7690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.203914301651682e-07,
+      "loss": 0.9958,
+      "step": 7691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.1814578566895e-07,
+      "loss": 0.9899,
+      "step": 7692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.159035957575777e-07,
+      "loss": 0.7889,
+      "step": 7693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.136648607500435e-07,
+      "loss": 0.6449,
+      "step": 7694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.114295809648563e-07,
+      "loss": 0.6993,
+      "step": 7695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.091977567200314e-07,
+      "loss": 0.92,
+      "step": 7696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.069693883330842e-07,
+      "loss": 1.0178,
+      "step": 7697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.047444761210553e-07,
+      "loss": 1.1852,
+      "step": 7698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.025230204004779e-07,
+      "loss": 0.7004,
+      "step": 7699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 7.003050214873963e-07,
+      "loss": 0.9126,
+      "step": 7700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.98090479697372e-07,
+      "loss": 0.973,
+      "step": 7701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.958793953454612e-07,
+      "loss": 0.8126,
+      "step": 7702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.936717687462429e-07,
+      "loss": 0.9139,
+      "step": 7703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.91467600213791e-07,
+      "loss": 1.1659,
+      "step": 7704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.892668900616933e-07,
+      "loss": 0.7391,
+      "step": 7705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.870696386030495e-07,
+      "loss": 1.1593,
+      "step": 7706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.848758461504596e-07,
+      "loss": 0.7612,
+      "step": 7707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.826855130160348e-07,
+      "loss": 1.0042,
+      "step": 7708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.804986395113983e-07,
+      "loss": 0.7014,
+      "step": 7709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.783152259476705e-07,
+      "loss": 0.8306,
+      "step": 7710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.761352726354891e-07,
+      "loss": 1.1179,
+      "step": 7711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.739587798850006e-07,
+      "loss": 0.8926,
+      "step": 7712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.717857480058515e-07,
+      "loss": 0.641,
+      "step": 7713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.696161773071974e-07,
+      "loss": 1.0934,
+      "step": 7714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.67450068097708e-07,
+      "loss": 0.8602,
+      "step": 7715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.652874206855508e-07,
+      "loss": 0.7921,
+      "step": 7716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.631282353784158e-07,
+      "loss": 1.0255,
+      "step": 7717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.609725124834792e-07,
+      "loss": 1.0334,
+      "step": 7718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.588202523074432e-07,
+      "loss": 1.4534,
+      "step": 7719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.566714551565073e-07,
+      "loss": 0.9905,
+      "step": 7720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.545261213363851e-07,
+      "loss": 1.2002,
+      "step": 7721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.523842511522882e-07,
+      "loss": 0.9437,
+      "step": 7722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.502458449089449e-07,
+      "loss": 0.9871,
+      "step": 7723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.481109029105842e-07,
+      "loss": 1.0372,
+      "step": 7724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.459794254609491e-07,
+      "loss": 1.0457,
+      "step": 7725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.438514128632805e-07,
+      "loss": 1.1929,
+      "step": 7726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.417268654203363e-07,
+      "loss": 0.7889,
+      "step": 7727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.396057834343716e-07,
+      "loss": 1.0398,
+      "step": 7728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.374881672071537e-07,
+      "loss": 0.9082,
+      "step": 7729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.35374017039958e-07,
+      "loss": 0.9165,
+      "step": 7730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.332633332335664e-07,
+      "loss": 0.9777,
+      "step": 7731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.311561160882662e-07,
+      "loss": 1.1147,
+      "step": 7732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.290523659038455e-07,
+      "loss": 0.6321,
+      "step": 7733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.269520829796149e-07,
+      "loss": 0.8235,
+      "step": 7734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.248552676143771e-07,
+      "loss": 1.0059,
+      "step": 7735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.227619201064405e-07,
+      "loss": 0.6089,
+      "step": 7736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.206720407536365e-07,
+      "loss": 0.7516,
+      "step": 7737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.185856298532882e-07,
+      "loss": 0.6827,
+      "step": 7738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.165026877022278e-07,
+      "loss": 0.8588,
+      "step": 7739
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.144232145968016e-07,
+      "loss": 1.0126,
+      "step": 7740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.123472108328504e-07,
+      "loss": 0.59,
+      "step": 7741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.102746767057327e-07,
+      "loss": 0.7118,
+      "step": 7742
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.082056125103069e-07,
+      "loss": 0.7303,
+      "step": 7743
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.061400185409349e-07,
+      "loss": 0.7377,
+      "step": 7744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.040778950914955e-07,
+      "loss": 0.9531,
+      "step": 7745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.02019242455365e-07,
+      "loss": 0.927,
+      "step": 7746
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.999640609254286e-07,
+      "loss": 0.7915,
+      "step": 7747
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.979123507940748e-07,
+      "loss": 0.7524,
+      "step": 7748
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.958641123532061e-07,
+      "loss": 0.8701,
+      "step": 7749
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.938193458942198e-07,
+      "loss": 0.9625,
+      "step": 7750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.91778051708028e-07,
+      "loss": 0.8261,
+      "step": 7751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.897402300850508e-07,
+      "loss": 1.1643,
+      "step": 7752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.877058813152009e-07,
+      "loss": 0.7933,
+      "step": 7753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.856750056879106e-07,
+      "loss": 0.7457,
+      "step": 7754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.836476034921129e-07,
+      "loss": 0.8721,
+      "step": 7755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.816236750162463e-07,
+      "loss": 1.0027,
+      "step": 7756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.796032205482582e-07,
+      "loss": 0.9464,
+      "step": 7757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.775862403755911e-07,
+      "loss": 0.8762,
+      "step": 7758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.755727347852096e-07,
+      "loss": 0.8353,
+      "step": 7759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.73562704063571e-07,
+      "loss": 1.0579,
+      "step": 7760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.715561484966436e-07,
+      "loss": 1.0592,
+      "step": 7761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.695530683699046e-07,
+      "loss": 0.9091,
+      "step": 7762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.675534639683261e-07,
+      "loss": 0.7209,
+      "step": 7763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.655573355763971e-07,
+      "loss": 0.9791,
+      "step": 7764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.635646834781016e-07,
+      "loss": 0.7138,
+      "step": 7765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.615755079569434e-07,
+      "loss": 1.0161,
+      "step": 7766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.595898092959156e-07,
+      "loss": 0.892,
+      "step": 7767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.576075877775283e-07,
+      "loss": 1.0197,
+      "step": 7768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.556288436837864e-07,
+      "loss": 0.8254,
+      "step": 7769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.536535772962148e-07,
+      "loss": 0.8846,
+      "step": 7770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.516817888958304e-07,
+      "loss": 0.8503,
+      "step": 7771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.497134787631586e-07,
+      "loss": 0.6519,
+      "step": 7772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.477486471782367e-07,
+      "loss": 1.0339,
+      "step": 7773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.457872944205966e-07,
+      "loss": 1.0905,
+      "step": 7774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.438294207692818e-07,
+      "loss": 0.6791,
+      "step": 7775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.418750265028416e-07,
+      "loss": 1.1423,
+      "step": 7776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.399241118993232e-07,
+      "loss": 0.9543,
+      "step": 7777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.379766772362849e-07,
+      "loss": 1.046,
+      "step": 7778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.360327227907941e-07,
+      "loss": 1.0319,
+      "step": 7779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.340922488394156e-07,
+      "loss": 0.8766,
+      "step": 7780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.321552556582149e-07,
+      "loss": 0.6868,
+      "step": 7781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.302217435227769e-07,
+      "loss": 1.1259,
+      "step": 7782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.28291712708176e-07,
+      "loss": 1.0559,
+      "step": 7783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.26365163489001e-07,
+      "loss": 1.0448,
+      "step": 7784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.244420961393465e-07,
+      "loss": 0.8389,
+      "step": 7785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.225225109327992e-07,
+      "loss": 0.7197,
+      "step": 7786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 5.206064081424655e-07,
+      "loss": 1.0299,
+      "step": 7787
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.186937880409471e-07,
+      "loss": 0.8096,
+      "step": 7788
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.167846509003538e-07,
+      "loss": 0.8741,
+      "step": 7789
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.148789969922991e-07,
+      "loss": 0.917,
+      "step": 7790
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.129768265878992e-07,
+      "loss": 0.9246,
+      "step": 7791
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.110781399577769e-07,
+      "loss": 0.8118,
+      "step": 7792
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.091829373720603e-07,
+      "loss": 0.8492,
+      "step": 7793
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.072912191003809e-07,
+      "loss": 0.9655,
+      "step": 7794
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.054029854118708e-07,
+      "loss": 1.0162,
+      "step": 7795
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.035182365751706e-07,
+      "loss": 0.7131,
+      "step": 7796
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 5.016369728584241e-07,
+      "loss": 1.2522,
+      "step": 7797
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.997591945292812e-07,
+      "loss": 1.2714,
+      "step": 7798
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.97884901854892e-07,
+      "loss": 0.9156,
+      "step": 7799
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.960140951019127e-07,
+      "loss": 0.8026,
+      "step": 7800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.941467745365025e-07,
+      "loss": 0.6932,
+      "step": 7801
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.922829404243267e-07,
+      "loss": 1.1956,
+      "step": 7802
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.904225930305512e-07,
+      "loss": 0.8798,
+      "step": 7803
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.885657326198501e-07,
+      "loss": 0.6622,
+      "step": 7804
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.867123594563983e-07,
+      "loss": 0.8563,
+      "step": 7805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.848624738038766e-07,
+      "loss": 1.1858,
+      "step": 7806
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.830160759254715e-07,
+      "loss": 0.7775,
+      "step": 7807
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.811731660838647e-07,
+      "loss": 0.8684,
+      "step": 7808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.793337445412493e-07,
+      "loss": 0.8534,
+      "step": 7809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.774978115593215e-07,
+      "loss": 1.0073,
+      "step": 7810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.756653673992806e-07,
+      "loss": 1.0173,
+      "step": 7811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7383641232182643e-07,
+      "loss": 0.6144,
+      "step": 7812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7201094658716747e-07,
+      "loss": 0.9683,
+      "step": 7813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.7018897045501e-07,
+      "loss": 0.9033,
+      "step": 7814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6837048418456876e-07,
+      "loss": 0.777,
+      "step": 7815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.6655548803456186e-07,
+      "loss": 0.8594,
+      "step": 7816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.647439822632049e-07,
+      "loss": 0.7684,
+      "step": 7817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.629359671282252e-07,
+      "loss": 0.9052,
+      "step": 7818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.611314428868474e-07,
+      "loss": 0.7667,
+      "step": 7819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5933040979579955e-07,
+      "loss": 0.7162,
+      "step": 7820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.575328681113239e-07,
+      "loss": 0.8065,
+      "step": 7821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5573881808914644e-07,
+      "loss": 0.8547,
+      "step": 7822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.5394825998451296e-07,
+      "loss": 0.8507,
+      "step": 7823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.521611940521642e-07,
+      "loss": 0.8607,
+      "step": 7824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.503776205463467e-07,
+      "loss": 0.7362,
+      "step": 7825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.485975397208131e-07,
+      "loss": 0.9594,
+      "step": 7826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4682095182881076e-07,
+      "loss": 0.8788,
+      "step": 7827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.450478571230987e-07,
+      "loss": 0.8125,
+      "step": 7828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4327825585593617e-07,
+      "loss": 0.9352,
+      "step": 7829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.4151214827908015e-07,
+      "loss": 0.9651,
+      "step": 7830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.397495346437991e-07,
+      "loss": 0.7993,
+      "step": 7831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3799041520086184e-07,
+      "loss": 0.764,
+      "step": 7832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.3623479020053203e-07,
+      "loss": 0.8159,
+      "step": 7833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.344826598925877e-07,
+      "loss": 0.9712,
+      "step": 7834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.327340245263073e-07,
+      "loss": 0.989,
+      "step": 7835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.309888843504639e-07,
+      "loss": 0.9767,
+      "step": 7836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2924723961334513e-07,
+      "loss": 0.551,
+      "step": 7837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.275090905627305e-07,
+      "loss": 0.9104,
+      "step": 7838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.257744374459055e-07,
+      "loss": 0.85,
+      "step": 7839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2404328050966445e-07,
+      "loss": 0.7437,
+      "step": 7840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.2231562000029644e-07,
+      "loss": 0.8918,
+      "step": 7841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.205914561635965e-07,
+      "loss": 1.2026,
+      "step": 7842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1887078924486287e-07,
+      "loss": 0.7375,
+      "step": 7843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.171536194888942e-07,
+      "loss": 0.6418,
+      "step": 7844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1543994713999224e-07,
+      "loss": 0.7059,
+      "step": 7845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.13729772441962e-07,
+      "loss": 0.8171,
+      "step": 7846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.120230956381116e-07,
+      "loss": 0.7355,
+      "step": 7847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.1031991697124683e-07,
+      "loss": 0.9961,
+      "step": 7848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.08620236683685e-07,
+      "loss": 0.8347,
+      "step": 7849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.069240550172354e-07,
+      "loss": 0.5237,
+      "step": 7850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.052313722132162e-07,
+      "loss": 1.1344,
+      "step": 7851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0354218851244574e-07,
+      "loss": 0.9479,
+      "step": 7852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0185650415524303e-07,
+      "loss": 0.7726,
+      "step": 7853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.0017431938143e-07,
+      "loss": 0.9631,
+      "step": 7854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9849563443033754e-07,
+      "loss": 1.0762,
+      "step": 7855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.968204495407829e-07,
+      "loss": 0.8562,
+      "step": 7856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.95148764951106e-07,
+      "loss": 1.0031,
+      "step": 7857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9348058089912763e-07,
+      "loss": 0.8961,
+      "step": 7858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.9181589762218584e-07,
+      "loss": 0.9696,
+      "step": 7859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.901547153571161e-07,
+      "loss": 0.9226,
+      "step": 7860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8849703434025166e-07,
+      "loss": 0.8255,
+      "step": 7861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8684285480743433e-07,
+      "loss": 0.9058,
+      "step": 7862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8519217699400643e-07,
+      "loss": 0.9702,
+      "step": 7863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.835450011348024e-07,
+      "loss": 1.1813,
+      "step": 7864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.819013274641764e-07,
+      "loss": 0.8636,
+      "step": 7865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.8026115621596635e-07,
+      "loss": 1.1499,
+      "step": 7866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.786244876235245e-07,
+      "loss": 0.9989,
+      "step": 7867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.76991321919698e-07,
+      "loss": 0.9884,
+      "step": 7868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.75361659336837e-07,
+      "loss": 0.9811,
+      "step": 7869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.737355001067949e-07,
+      "loss": 1.0613,
+      "step": 7870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.7211284446092564e-07,
+      "loss": 0.8405,
+      "step": 7871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.704936926300834e-07,
+      "loss": 0.9736,
+      "step": 7872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.688780448446311e-07,
+      "loss": 1.0304,
+      "step": 7873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.672659013344182e-07,
+      "loss": 1.0798,
+      "step": 7874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6565726232881404e-07,
+      "loss": 0.6689,
+      "step": 7875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6405212805667166e-07,
+      "loss": 0.9503,
+      "step": 7876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.624504987463584e-07,
+      "loss": 0.7797,
+      "step": 7877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.6085237462573917e-07,
+      "loss": 1.0787,
+      "step": 7878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.592577559221766e-07,
+      "loss": 0.8943,
+      "step": 7879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.576666428625419e-07,
+      "loss": 0.7729,
+      "step": 7880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.560790356731958e-07,
+      "loss": 0.7623,
+      "step": 7881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.544949345800158e-07,
+      "loss": 1.1835,
+      "step": 7882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.529143398083662e-07,
+      "loss": 0.9958,
+      "step": 7883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.513372515831198e-07,
+      "loss": 0.7007,
+      "step": 7884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4976367012865277e-07,
+      "loss": 0.8036,
+      "step": 7885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4819359566883316e-07,
+      "loss": 0.8556,
+      "step": 7886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4662702842704065e-07,
+      "loss": 0.8196,
+      "step": 7887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4506396862614977e-07,
+      "loss": 0.7747,
+      "step": 7888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4350441648853817e-07,
+      "loss": 0.8018,
+      "step": 7889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4194837223607834e-07,
+      "loss": 1.1055,
+      "step": 7890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.403958360901571e-07,
+      "loss": 0.9165,
+      "step": 7891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.388468082716478e-07,
+      "loss": 0.837,
+      "step": 7892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.373012890009353e-07,
+      "loss": 1.2682,
+      "step": 7893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.357592784978991e-07,
+      "loss": 1.007,
+      "step": 7894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3422077698192225e-07,
+      "loss": 0.7208,
+      "step": 7895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3268578467188507e-07,
+      "loss": 0.8019,
+      "step": 7896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.3115430178617404e-07,
+      "loss": 0.663,
+      "step": 7897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.296263285426704e-07,
+      "loss": 0.637,
+      "step": 7898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2810186515876417e-07,
+      "loss": 1.085,
+      "step": 7899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2658091185133744e-07,
+      "loss": 0.8319,
+      "step": 7900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.250634688367754e-07,
+      "loss": 0.8948,
+      "step": 7901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.235495363309693e-07,
+      "loss": 0.9807,
+      "step": 7902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.220391145493051e-07,
+      "loss": 0.9159,
+      "step": 7903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.2053220370666934e-07,
+      "loss": 0.8219,
+      "step": 7904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.190288040174516e-07,
+      "loss": 0.7972,
+      "step": 7905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.175289156955391e-07,
+      "loss": 0.6343,
+      "step": 7906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.160325389543251e-07,
+      "loss": 0.6441,
+      "step": 7907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.145396740066975e-07,
+      "loss": 0.8297,
+      "step": 7908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.1305032106504484e-07,
+      "loss": 0.7503,
+      "step": 7909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.115644803412615e-07,
+      "loss": 0.9777,
+      "step": 7910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.10082152046734e-07,
+      "loss": 0.7777,
+      "step": 7911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0860333639235464e-07,
+      "loss": 1.0529,
+      "step": 7912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0712803358851625e-07,
+      "loss": 1.3105,
+      "step": 7913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0565624384510936e-07,
+      "loss": 0.9271,
+      "step": 7914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0418796737152756e-07,
+      "loss": 1.0277,
+      "step": 7915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0272320437666203e-07,
+      "loss": 0.8837,
+      "step": 7916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.0126195506890717e-07,
+      "loss": 1.1765,
+      "step": 7917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.998042196561468e-07,
+      "loss": 0.8218,
+      "step": 7918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.983499983457844e-07,
+      "loss": 0.8187,
+      "step": 7919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.968992913447044e-07,
+      "loss": 0.8517,
+      "step": 7920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.954520988593029e-07,
+      "loss": 1.0289,
+      "step": 7921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9400842109547354e-07,
+      "loss": 1.0243,
+      "step": 7922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.9256825825860203e-07,
+      "loss": 0.6328,
+      "step": 7923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.911316105535883e-07,
+      "loss": 0.9291,
+      "step": 7924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8969847818482163e-07,
+      "loss": 0.8381,
+      "step": 7925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.882688613561918e-07,
+      "loss": 1.2634,
+      "step": 7926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.868427602710944e-07,
+      "loss": 0.7709,
+      "step": 7927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8542017513241705e-07,
+      "loss": 1.057,
+      "step": 7928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8400110614255627e-07,
+      "loss": 1.0052,
+      "step": 7929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8258555350340053e-07,
+      "loss": 0.8594,
+      "step": 7930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.811735174163388e-07,
+      "loss": 0.8929,
+      "step": 7931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7976499808226584e-07,
+      "loss": 0.8998,
+      "step": 7932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7835999570157144e-07,
+      "loss": 0.7915,
+      "step": 7933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.769585104741429e-07,
+      "loss": 1.1903,
+      "step": 7934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7556054259937345e-07,
+      "loss": 0.8741,
+      "step": 7935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7416609227614855e-07,
+      "loss": 0.9545,
+      "step": 7936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.727751597028566e-07,
+      "loss": 1.11,
+      "step": 7937
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.713877450773894e-07,
+      "loss": 0.9844,
+      "step": 7938
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.7000384859713625e-07,
+      "loss": 0.7731,
+      "step": 7939
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6862347045897573e-07,
+      "loss": 0.9089,
+      "step": 7940
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.672466108593036e-07,
+      "loss": 0.8051,
+      "step": 7941
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.65873269994002e-07,
+      "loss": 0.8573,
+      "step": 7942
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6450344805845626e-07,
+      "loss": 1.0757,
+      "step": 7943
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.631371452475523e-07,
+      "loss": 0.7623,
+      "step": 7944
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.617743617556734e-07,
+      "loss": 1.0338,
+      "step": 7945
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.6041509777670613e-07,
+      "loss": 0.8878,
+      "step": 7946
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.590593535040292e-07,
+      "loss": 0.9553,
+      "step": 7947
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.577071291305272e-07,
+      "loss": 0.8322,
+      "step": 7948
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.563584248485795e-07,
+      "loss": 0.8798,
+      "step": 7949
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.550132408500688e-07,
+      "loss": 0.7906,
+      "step": 7950
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5367157732637257e-07,
+      "loss": 0.9147,
+      "step": 7951
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.523334344683714e-07,
+      "loss": 0.9023,
+      "step": 7952
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.5099881246644086e-07,
+      "loss": 0.9186,
+      "step": 7953
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.4966771151045953e-07,
+      "loss": 0.8823,
+      "step": 7954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4834013178980095e-07,
+      "loss": 0.7361,
+      "step": 7955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4701607349334745e-07,
+      "loss": 1.2,
+      "step": 7956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.456955368094649e-07,
+      "loss": 0.6936,
+      "step": 7957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.4437852192603084e-07,
+      "loss": 1.0166,
+      "step": 7958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.430650290304148e-07,
+      "loss": 0.8666,
+      "step": 7959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.417550583094896e-07,
+      "loss": 1.2069,
+      "step": 7960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.404486099496256e-07,
+      "loss": 0.8444,
+      "step": 7961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.391456841366879e-07,
+      "loss": 1.0647,
+      "step": 7962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3784628105605056e-07,
+      "loss": 0.9885,
+      "step": 7963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.365504008925712e-07,
+      "loss": 1.0992,
+      "step": 7964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.352580438306218e-07,
+      "loss": 0.9124,
+      "step": 7965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.339692100540636e-07,
+      "loss": 1.0057,
+      "step": 7966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3268389974626104e-07,
+      "loss": 0.9044,
+      "step": 7967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3140211309007341e-07,
+      "loss": 0.8849,
+      "step": 7968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.3012385026786042e-07,
+      "loss": 0.9004,
+      "step": 7969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2884911146148212e-07,
+      "loss": 1.1027,
+      "step": 7970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2757789685229625e-07,
+      "loss": 1.047,
+      "step": 7971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2631020662115808e-07,
+      "loss": 0.8834,
+      "step": 7972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.250460409484234e-07,
+      "loss": 1.2079,
+      "step": 7973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2378540001394553e-07,
+      "loss": 0.9802,
+      "step": 7974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.225282839970727e-07,
+      "loss": 1.075,
+      "step": 7975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.2127469307665627e-07,
+      "loss": 0.7416,
+      "step": 7976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.20024627431048e-07,
+      "loss": 0.8329,
+      "step": 7977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1877808723809456e-07,
+      "loss": 0.8051,
+      "step": 7978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.175350726751374e-07,
+      "loss": 0.827,
+      "step": 7979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1629558391902393e-07,
+      "loss": 1.1111,
+      "step": 7980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1505962114609367e-07,
+      "loss": 0.8876,
+      "step": 7981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1382718453219208e-07,
+      "loss": 0.9753,
+      "step": 7982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1259827425265388e-07,
+      "loss": 0.649,
+      "step": 7983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.113728904823198e-07,
+      "loss": 0.9364,
+      "step": 7984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.1015103339551979e-07,
+      "loss": 0.8928,
+      "step": 7985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0893270316609537e-07,
+      "loss": 0.9985,
+      "step": 7986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0771789996737178e-07,
+      "loss": 1.0387,
+      "step": 7987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.06506623972183e-07,
+      "loss": 1.1791,
+      "step": 7988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0529887535285785e-07,
+      "loss": 0.5985,
+      "step": 7989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0409465428122276e-07,
+      "loss": 1.0876,
+      "step": 7990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0289396092859902e-07,
+      "loss": 0.8762,
+      "step": 7991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0169679546581664e-07,
+      "loss": 0.8387,
+      "step": 7992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0050315806318664e-07,
+      "loss": 0.9005,
+      "step": 7993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9931304889053703e-07,
+      "loss": 0.8839,
+      "step": 7994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.981264681171824e-07,
+      "loss": 0.9578,
+      "step": 7995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9694341591193488e-07,
+      "loss": 0.6602,
+      "step": 7996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.957638924431099e-07,
+      "loss": 1.1228,
+      "step": 7997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9458789787852038e-07,
+      "loss": 0.8245,
+      "step": 7998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9341543238547143e-07,
+      "loss": 0.8729,
+      "step": 7999
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9224649613077406e-07,
+      "loss": 0.9282,
+      "step": 8000
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 0.98046875,
+      "eval_runtime": 3.5874,
+      "eval_samples_per_second": 27.875,
+      "eval_steps_per_second": 4.739,
+      "step": 8000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 8329,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 8.99873092141056e+17,
+  "train_batch_size": 6,
+  "trial_name": null,
+  "trial_params": null
+}