diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,32166 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 5356,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2422360248447206e-07,
+      "loss": 1.3434,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.484472049689441e-07,
+      "loss": 1.335,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.726708074534162e-07,
+      "loss": 1.3158,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.968944099378882e-07,
+      "loss": 1.3159,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.211180124223603e-07,
+      "loss": 1.339,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.453416149068324e-07,
+      "loss": 1.3072,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 1.2607,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.937888198757765e-07,
+      "loss": 1.2563,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1180124223602485e-06,
+      "loss": 1.3102,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2422360248447205e-06,
+      "loss": 1.2574,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3664596273291927e-06,
+      "loss": 1.2447,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4906832298136647e-06,
+      "loss": 1.229,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6149068322981367e-06,
+      "loss": 1.1909,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 1.1963,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8633540372670808e-06,
+      "loss": 1.1198,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.987577639751553e-06,
+      "loss": 1.136,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.111801242236025e-06,
+      "loss": 1.1905,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.236024844720497e-06,
+      "loss": 0.1953,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3602484472049692e-06,
+      "loss": 1.0786,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.484472049689441e-06,
+      "loss": 1.1613,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6086956521739132e-06,
+      "loss": 1.1289,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7329192546583855e-06,
+      "loss": 1.1322,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.0391,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9813664596273295e-06,
+      "loss": 1.1445,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1055900621118013e-06,
+      "loss": 1.083,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2298136645962735e-06,
+      "loss": 1.0067,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3540372670807457e-06,
+      "loss": 0.9801,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 1.0252,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6024844720496897e-06,
+      "loss": 1.0167,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7267080745341615e-06,
+      "loss": 0.9683,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.850931677018634e-06,
+      "loss": 1.0336,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.975155279503106e-06,
+      "loss": 1.014,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.099378881987578e-06,
+      "loss": 0.9369,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.22360248447205e-06,
+      "loss": 0.9783,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 0.9596,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.472049689440994e-06,
+      "loss": 1.0897,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.596273291925466e-06,
+      "loss": 0.9813,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7204968944099384e-06,
+      "loss": 1.0142,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.84472049689441e-06,
+      "loss": 1.042,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.968944099378882e-06,
+      "loss": 0.2196,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.093167701863354e-06,
+      "loss": 0.2189,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2173913043478265e-06,
+      "loss": 0.98,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.341614906832298e-06,
+      "loss": 0.9393,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.465838509316771e-06,
+      "loss": 0.9529,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.590062111801242e-06,
+      "loss": 0.9681,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.9642,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.838509316770186e-06,
+      "loss": 1.0013,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.962732919254659e-06,
+      "loss": 0.9235,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 0.9556,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2111801242236025e-06,
+      "loss": 1.0019,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.335403726708075e-06,
+      "loss": 0.9384,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.459627329192547e-06,
+      "loss": 0.2243,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.58385093167702e-06,
+      "loss": 0.9759,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.7080745341614914e-06,
+      "loss": 0.9076,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.832298136645963e-06,
+      "loss": 0.9549,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 0.2287,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.080745341614908e-06,
+      "loss": 0.9072,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.2049689440993795e-06,
+      "loss": 0.9366,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.329192546583852e-06,
+      "loss": 0.8662,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.453416149068323e-06,
+      "loss": 0.9584,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.577639751552796e-06,
+      "loss": 0.9441,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.701863354037268e-06,
+      "loss": 0.8872,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 0.9531,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.950310559006212e-06,
+      "loss": 0.8931,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.074534161490684e-06,
+      "loss": 0.9428,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.198757763975156e-06,
+      "loss": 0.9416,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.322981366459629e-06,
+      "loss": 0.2243,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.4472049689441e-06,
+      "loss": 0.9164,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.2369,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 0.9661,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.819875776397516e-06,
+      "loss": 0.9499,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.944099378881988e-06,
+      "loss": 0.9676,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.068322981366461e-06,
+      "loss": 0.9298,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.192546583850932e-06,
+      "loss": 0.9059,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.316770186335405e-06,
+      "loss": 0.9617,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.440993788819877e-06,
+      "loss": 0.9696,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 0.9293,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.68944099378882e-06,
+      "loss": 0.9723,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.813664596273292e-06,
+      "loss": 0.9531,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.937888198757764e-06,
+      "loss": 0.8877,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0062111801242236e-05,
+      "loss": 0.2393,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0186335403726708e-05,
+      "loss": 0.8878,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0310559006211181e-05,
+      "loss": 0.9411,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0434782608695653e-05,
+      "loss": 0.9137,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0559006211180125e-05,
+      "loss": 0.9429,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0683229813664597e-05,
+      "loss": 0.9038,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.080745341614907e-05,
+      "loss": 0.9222,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0931677018633542e-05,
+      "loss": 0.9573,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1055900621118014e-05,
+      "loss": 0.9196,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1180124223602484e-05,
+      "loss": 0.9493,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1304347826086957e-05,
+      "loss": 0.909,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.9291,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.15527950310559e-05,
+      "loss": 0.936,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1677018633540373e-05,
+      "loss": 0.9649,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1801242236024846e-05,
+      "loss": 0.2146,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1925465838509318e-05,
+      "loss": 0.988,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.204968944099379e-05,
+      "loss": 0.9097,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2173913043478263e-05,
+      "loss": 0.9353,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2298136645962735e-05,
+      "loss": 0.9403,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2422360248447205e-05,
+      "loss": 0.8946,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2546583850931677e-05,
+      "loss": 0.9397,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.267080745341615e-05,
+      "loss": 0.9569,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2795031055900622e-05,
+      "loss": 0.9454,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2919254658385094e-05,
+      "loss": 0.8804,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3043478260869566e-05,
+      "loss": 0.945,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.316770186335404e-05,
+      "loss": 0.9699,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3291925465838511e-05,
+      "loss": 0.9184,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3416149068322983e-05,
+      "loss": 0.9523,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3540372670807453e-05,
+      "loss": 0.8962,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3664596273291926e-05,
+      "loss": 0.9179,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3788819875776398e-05,
+      "loss": 0.8905,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.391304347826087e-05,
+      "loss": 0.8915,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4037267080745342e-05,
+      "loss": 0.8799,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4161490683229815e-05,
+      "loss": 0.8755,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.9529,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4409937888198759e-05,
+      "loss": 0.8505,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4534161490683232e-05,
+      "loss": 0.8738,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4658385093167704e-05,
+      "loss": 0.8705,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4782608695652174e-05,
+      "loss": 0.9168,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4906832298136646e-05,
+      "loss": 0.9366,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5031055900621118e-05,
+      "loss": 0.9378,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5155279503105591e-05,
+      "loss": 0.9584,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5279503105590063e-05,
+      "loss": 0.8655,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5403726708074537e-05,
+      "loss": 0.9228,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5527950310559007e-05,
+      "loss": 0.8944,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.565217391304348e-05,
+      "loss": 0.8714,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5776397515527954e-05,
+      "loss": 0.9325,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5900621118012424e-05,
+      "loss": 0.9302,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6024844720496894e-05,
+      "loss": 0.8791,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6149068322981367e-05,
+      "loss": 0.9326,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.627329192546584e-05,
+      "loss": 0.9383,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.639751552795031e-05,
+      "loss": 0.9085,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6521739130434785e-05,
+      "loss": 0.923,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6645962732919258e-05,
+      "loss": 0.9701,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6770186335403728e-05,
+      "loss": 0.8804,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.68944099378882e-05,
+      "loss": 0.9123,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7018633540372672e-05,
+      "loss": 0.8562,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.9401,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7267080745341615e-05,
+      "loss": 0.8969,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 0.8842,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.751552795031056e-05,
+      "loss": 0.8784,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7639751552795032e-05,
+      "loss": 0.945,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7763975155279506e-05,
+      "loss": 0.9071,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7888198757763976e-05,
+      "loss": 0.9947,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.801242236024845e-05,
+      "loss": 0.9267,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8136645962732923e-05,
+      "loss": 0.8818,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 0.9596,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8385093167701863e-05,
+      "loss": 0.9526,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8509316770186337e-05,
+      "loss": 0.8689,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.863354037267081e-05,
+      "loss": 0.9067,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.875776397515528e-05,
+      "loss": 0.9604,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8881987577639754e-05,
+      "loss": 0.9075,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9006211180124224e-05,
+      "loss": 0.8881,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9130434782608697e-05,
+      "loss": 0.818,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.925465838509317e-05,
+      "loss": 0.2453,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.937888198757764e-05,
+      "loss": 0.9011,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.950310559006211e-05,
+      "loss": 0.8905,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9627329192546585e-05,
+      "loss": 0.9242,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9751552795031058e-05,
+      "loss": 0.8766,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9875776397515528e-05,
+      "loss": 0.8214,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.864,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998171484557e-05,
+      "loss": 0.9119,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999992685938895e-05,
+      "loss": 0.9548,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999983543365024e-05,
+      "loss": 0.8671,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999970743766278e-05,
+      "loss": 0.9287,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995428714735e-05,
+      "loss": 0.8732,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993417351425e-05,
+      "loss": 0.9531,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999910402874337e-05,
+      "loss": 0.9181,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99998829752363e-05,
+      "loss": 0.8956,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999851890610174e-05,
+      "loss": 0.8718,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999817149007327e-05,
+      "loss": 0.8793,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999977875044046e-05,
+      "loss": 0.246,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999736694923617e-05,
+      "loss": 0.8811,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999969098247218e-05,
+      "loss": 0.8337,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999641613102864e-05,
+      "loss": 0.9056,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999588586833724e-05,
+      "loss": 0.88,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999531903684153e-05,
+      "loss": 0.9286,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999471563674878e-05,
+      "loss": 0.8956,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999940756682797e-05,
+      "loss": 0.94,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999339913166827e-05,
+      "loss": 0.8689,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999268602716193e-05,
+      "loss": 0.8956,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999193635502146e-05,
+      "loss": 0.8666,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999115011552104e-05,
+      "loss": 0.882,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999032730894817e-05,
+      "loss": 0.9514,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998946793560376e-05,
+      "loss": 0.8892,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999885719958021e-05,
+      "loss": 0.8926,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998763948987082e-05,
+      "loss": 0.8767,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999866704181509e-05,
+      "loss": 0.862,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998566478099682e-05,
+      "loss": 0.9109,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998462257877634e-05,
+      "loss": 0.9045,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999835438118705e-05,
+      "loss": 0.9295,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999824284806739e-05,
+      "loss": 0.922,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999812765855944e-05,
+      "loss": 0.9193,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998008812705323e-05,
+      "loss": 0.9147,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99978863105485e-05,
+      "loss": 0.8744,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997760152133776e-05,
+      "loss": 0.9261,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997630337507283e-05,
+      "loss": 0.8888,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99974968667165e-05,
+      "loss": 0.8643,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999735973981023e-05,
+      "loss": 0.9747,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999721895683862e-05,
+      "loss": 0.882,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997074517853163e-05,
+      "loss": 0.9302,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999692642290668e-05,
+      "loss": 0.9076,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996774672053322e-05,
+      "loss": 0.8811,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996619265348595e-05,
+      "loss": 0.9415,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999646020284932e-05,
+      "loss": 0.9399,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996297484613674e-05,
+      "loss": 0.9056,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996131110701163e-05,
+      "loss": 0.88,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995961081172634e-05,
+      "loss": 0.912,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999578739609026e-05,
+      "loss": 0.8259,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995610055517558e-05,
+      "loss": 0.9338,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999542905951939e-05,
+      "loss": 0.8927,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995244408161936e-05,
+      "loss": 0.8924,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995056101512735e-05,
+      "loss": 0.8886,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994864139640648e-05,
+      "loss": 0.8559,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999466852261587e-05,
+      "loss": 0.8985,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999446925050994e-05,
+      "loss": 0.985,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999426632339574e-05,
+      "loss": 0.9242,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994059741347474e-05,
+      "loss": 0.2679,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993849504440696e-05,
+      "loss": 0.8893,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999363561275228e-05,
+      "loss": 0.9067,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999341806636046e-05,
+      "loss": 0.9351,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993196865344782e-05,
+      "loss": 0.9192,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992972009786144e-05,
+      "loss": 0.9421,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999274349976678e-05,
+      "loss": 0.9529,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999251133537025e-05,
+      "loss": 0.9592,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999227551668146e-05,
+      "loss": 0.9195,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992036043786656e-05,
+      "loss": 0.9076,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991792916773404e-05,
+      "loss": 0.2323,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991546135730625e-05,
+      "loss": 0.9096,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991295700748557e-05,
+      "loss": 0.883,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991041611918794e-05,
+      "loss": 0.898,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999078386933426e-05,
+      "loss": 0.9214,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99905224730892e-05,
+      "loss": 0.2295,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990257423279213e-05,
+      "loss": 0.9172,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998998872000123e-05,
+      "loss": 0.2522,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989716363353522e-05,
+      "loss": 0.9252,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998944035343568e-05,
+      "loss": 0.9321,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989160690348646e-05,
+      "loss": 0.9381,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988877374194695e-05,
+      "loss": 0.9292,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988590405077433e-05,
+      "loss": 0.9191,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998829978310181e-05,
+      "loss": 0.8532,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988005508374107e-05,
+      "loss": 0.966,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998770758100194e-05,
+      "loss": 0.8835,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998740600109426e-05,
+      "loss": 0.8824,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987100768761357e-05,
+      "loss": 0.9257,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998679188411486e-05,
+      "loss": 0.8906,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998647934726772e-05,
+      "loss": 0.9226,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986163158334236e-05,
+      "loss": 0.9581,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985843317430044e-05,
+      "loss": 0.869,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998551982467211e-05,
+      "loss": 0.9026,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998519268017873e-05,
+      "loss": 0.8907,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998486188406955e-05,
+      "loss": 0.8849,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984527436465533e-05,
+      "loss": 0.9047,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984189337488996e-05,
+      "loss": 0.8383,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998384758726358e-05,
+      "loss": 0.9051,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983502185914264e-05,
+      "loss": 0.8891,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983153133567362e-05,
+      "loss": 0.9413,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982800430350523e-05,
+      "loss": 0.9231,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982444076392737e-05,
+      "loss": 0.8704,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998208407182432e-05,
+      "loss": 0.9179,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998172041677692e-05,
+      "loss": 0.9271,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981353111383538e-05,
+      "loss": 0.9214,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998098215577849e-05,
+      "loss": 0.8967,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980607550097445e-05,
+      "loss": 0.9691,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980229294477386e-05,
+      "loss": 0.8438,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979847389056656e-05,
+      "loss": 0.9157,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979461833974905e-05,
+      "loss": 0.9109,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979072629373134e-05,
+      "loss": 0.9068,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978679775393687e-05,
+      "loss": 0.915,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978283272180222e-05,
+      "loss": 0.9247,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977883119877748e-05,
+      "loss": 0.827,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977479318632594e-05,
+      "loss": 0.8368,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997707186859244e-05,
+      "loss": 0.8847,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976660769906283e-05,
+      "loss": 0.8999,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997624602272447e-05,
+      "loss": 0.8937,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975827627198675e-05,
+      "loss": 0.8535,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99754055834819e-05,
+      "loss": 0.8741,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997497989172849e-05,
+      "loss": 0.8781,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997455055209413e-05,
+      "loss": 0.8972,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974117564735825e-05,
+      "loss": 0.8394,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973680929811915e-05,
+      "loss": 0.9383,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973240647482085e-05,
+      "loss": 0.8653,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997279671790735e-05,
+      "loss": 0.9197,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972349141250048e-05,
+      "loss": 0.9117,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971897917673865e-05,
+      "loss": 0.2587,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997144304734381e-05,
+      "loss": 0.9022,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997098453042624e-05,
+      "loss": 0.8893,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970522367088824e-05,
+      "loss": 0.8827,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970056557500586e-05,
+      "loss": 0.8832,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996958710183187e-05,
+      "loss": 0.8704,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969114000254354e-05,
+      "loss": 0.8851,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968637252941062e-05,
+      "loss": 0.9136,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996815686006633e-05,
+      "loss": 0.9026,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967672821805855e-05,
+      "loss": 0.9122,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967185138336635e-05,
+      "loss": 0.2361,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966693809837027e-05,
+      "loss": 0.8695,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996619883648671e-05,
+      "loss": 0.8898,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996570021846669e-05,
+      "loss": 0.8864,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965197955959323e-05,
+      "loss": 0.2261,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964692049148284e-05,
+      "loss": 0.9215,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964182498218585e-05,
+      "loss": 0.9155,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963669303356568e-05,
+      "loss": 0.9043,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963152464749916e-05,
+      "loss": 0.9335,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962631982587634e-05,
+      "loss": 0.8852,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962107857060058e-05,
+      "loss": 0.8362,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961580088358876e-05,
+      "loss": 0.9214,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961048676677083e-05,
+      "loss": 0.9108,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960513622209023e-05,
+      "loss": 0.899,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995997492515037e-05,
+      "loss": 0.9692,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995943258569812e-05,
+      "loss": 0.9546,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995888660405061e-05,
+      "loss": 0.8549,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958336980407514e-05,
+      "loss": 0.9143,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957783714969822e-05,
+      "loss": 0.9392,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995722680793987e-05,
+      "loss": 0.8411,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995666625952132e-05,
+      "loss": 0.9439,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956102069919167e-05,
+      "loss": 0.8938,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955534239339732e-05,
+      "loss": 0.9011,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995496276799068e-05,
+      "loss": 0.841,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954387656080995e-05,
+      "loss": 0.9343,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953808903820998e-05,
+      "loss": 0.8879,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953226511422337e-05,
+      "loss": 0.8265,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952640479098002e-05,
+      "loss": 0.9487,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99520508070623e-05,
+      "loss": 0.8873,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951457495530882e-05,
+      "loss": 0.8581,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995086054472072e-05,
+      "loss": 0.898,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950259954850123e-05,
+      "loss": 0.8852,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949655726138727e-05,
+      "loss": 0.8692,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949047858807498e-05,
+      "loss": 0.8924,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994843635307874e-05,
+      "loss": 0.9176,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994782120917608e-05,
+      "loss": 0.8506,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947202427324476e-05,
+      "loss": 0.9427,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994658000775022e-05,
+      "loss": 0.9184,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945953950680933e-05,
+      "loss": 0.8609,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994532425634557e-05,
+      "loss": 0.2693,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944690924974407e-05,
+      "loss": 0.8654,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944053956799055e-05,
+      "loss": 0.9218,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943413352052458e-05,
+      "loss": 0.8914,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942769110968885e-05,
+      "loss": 0.8652,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994212123378394e-05,
+      "loss": 0.8595,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994146972073455e-05,
+      "loss": 0.838,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940814572058977e-05,
+      "loss": 0.9146,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994015578799681e-05,
+      "loss": 0.8277,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993949336878897e-05,
+      "loss": 0.8914,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938827314677704e-05,
+      "loss": 0.9401,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938157625906594e-05,
+      "loss": 0.8784,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993748430272054e-05,
+      "loss": 0.9219,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936807345365786e-05,
+      "loss": 0.934,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993612675408989e-05,
+      "loss": 0.8809,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935442529141754e-05,
+      "loss": 0.9443,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99347546707716e-05,
+      "loss": 0.8321,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993406317923097e-05,
+      "loss": 0.8834,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993336805477276e-05,
+      "loss": 0.2192,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932669297651167e-05,
+      "loss": 0.866,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993196690812173e-05,
+      "loss": 0.2504,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931260886441325e-05,
+      "loss": 0.8549,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930551232868135e-05,
+      "loss": 0.8854,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929837947661686e-05,
+      "loss": 0.8788,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992912103108283e-05,
+      "loss": 0.8549,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928400483393745e-05,
+      "loss": 0.9232,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927676304857935e-05,
+      "loss": 0.8858,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926948495740244e-05,
+      "loss": 0.9106,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992621705630682e-05,
+      "loss": 0.8737,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925481986825163e-05,
+      "loss": 0.9477,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924743287564084e-05,
+      "loss": 0.8979,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992400095879373e-05,
+      "loss": 0.8799,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923255000785575e-05,
+      "loss": 0.8147,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922505413812415e-05,
+      "loss": 0.8551,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992175219814838e-05,
+      "loss": 0.9281,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992099535406892e-05,
+      "loss": 0.8904,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920234881850815e-05,
+      "loss": 0.922,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919470781772175e-05,
+      "loss": 0.8915,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991870305411243e-05,
+      "loss": 0.8565,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991793169915234e-05,
+      "loss": 0.8727,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917156717173997e-05,
+      "loss": 0.8901,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916378108460815e-05,
+      "loss": 0.9166,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915595873297524e-05,
+      "loss": 0.8757,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914810011970197e-05,
+      "loss": 0.2618,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914020524766226e-05,
+      "loss": 0.8866,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913227411974327e-05,
+      "loss": 0.8682,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912430673884547e-05,
+      "loss": 0.8607,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911630310788253e-05,
+      "loss": 0.9308,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991082632297814e-05,
+      "loss": 0.8829,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910018710748226e-05,
+      "loss": 0.8202,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990920747439386e-05,
+      "loss": 0.8643,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908392614211717e-05,
+      "loss": 0.8349,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907574130499792e-05,
+      "loss": 0.9597,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906752023557402e-05,
+      "loss": 0.8995,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99059262936852e-05,
+      "loss": 0.8779,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905096941185155e-05,
+      "loss": 0.8644,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904263966360568e-05,
+      "loss": 0.915,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903427369516054e-05,
+      "loss": 0.8678,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902587150957564e-05,
+      "loss": 0.9019,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990174331099237e-05,
+      "loss": 0.8672,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900895849929055e-05,
+      "loss": 0.8886,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900044768077553e-05,
+      "loss": 0.2537,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98991900657491e-05,
+      "loss": 0.8993,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898331743256267e-05,
+      "loss": 0.9881,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897469800912937e-05,
+      "loss": 0.8901,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896604239034334e-05,
+      "loss": 0.8611,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989573505793699e-05,
+      "loss": 0.87,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894862257938776e-05,
+      "loss": 0.8892,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893985839358868e-05,
+      "loss": 0.9255,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989310580251778e-05,
+      "loss": 0.9303,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892222147737345e-05,
+      "loss": 0.8271,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891334875340714e-05,
+      "loss": 0.881,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989044398565237e-05,
+      "loss": 0.8798,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889549478998108e-05,
+      "loss": 0.8412,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888651355705056e-05,
+      "loss": 0.8892,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887749616101666e-05,
+      "loss": 0.9339,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886844260517696e-05,
+      "loss": 0.8843,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885935289284243e-05,
+      "loss": 0.8511,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885022702733723e-05,
+      "loss": 0.8952,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884106501199866e-05,
+      "loss": 0.8936,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988318668501773e-05,
+      "loss": 0.8955,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98822632545237e-05,
+      "loss": 0.822,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881336210055478e-05,
+      "loss": 0.2657,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988040555195208e-05,
+      "loss": 0.8494,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879471280553855e-05,
+      "loss": 0.9638,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878533396202466e-05,
+      "loss": 0.8998,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877591899240903e-05,
+      "loss": 0.9019,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876646790013476e-05,
+      "loss": 0.8863,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987569806886581e-05,
+      "loss": 0.8527,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874745736144857e-05,
+      "loss": 0.9116,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987378979219889e-05,
+      "loss": 0.8811,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872830237377496e-05,
+      "loss": 0.9176,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871867072031592e-05,
+      "loss": 0.8584,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987090029651341e-05,
+      "loss": 0.8542,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869929911176496e-05,
+      "loss": 0.8846,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868955916375732e-05,
+      "loss": 0.9238,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986797831246731e-05,
+      "loss": 0.9309,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866997099808735e-05,
+      "loss": 0.911,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866012278758847e-05,
+      "loss": 0.9173,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865023849677794e-05,
+      "loss": 0.8643,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986403181292705e-05,
+      "loss": 0.824,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863036168869402e-05,
+      "loss": 0.8693,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862036917868967e-05,
+      "loss": 0.8971,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986103406029117e-05,
+      "loss": 0.2517,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860027596502757e-05,
+      "loss": 0.8723,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98590175268718e-05,
+      "loss": 0.895,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858003851767682e-05,
+      "loss": 0.8576,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856986571561107e-05,
+      "loss": 0.9256,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855965686624096e-05,
+      "loss": 0.8228,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985494119732999e-05,
+      "loss": 0.8708,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853913104053453e-05,
+      "loss": 0.8876,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852881407170454e-05,
+      "loss": 0.872,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851846107058294e-05,
+      "loss": 0.9241,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850807204095583e-05,
+      "loss": 0.8968,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849764698662254e-05,
+      "loss": 0.9293,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848718591139548e-05,
+      "loss": 0.8799,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847668881910036e-05,
+      "loss": 0.8923,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846615571357596e-05,
+      "loss": 0.8639,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845558659867428e-05,
+      "loss": 0.8638,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844498147826047e-05,
+      "loss": 0.9118,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843434035621293e-05,
+      "loss": 0.8523,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842366323642302e-05,
+      "loss": 0.9023,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841295012279547e-05,
+      "loss": 0.8745,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984022010192481e-05,
+      "loss": 0.9178,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983914159297119e-05,
+      "loss": 0.8492,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838059485813098e-05,
+      "loss": 0.9309,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836973780846267e-05,
+      "loss": 0.874,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835884478467743e-05,
+      "loss": 0.8679,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983479157907588e-05,
+      "loss": 0.2892,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983369508307036e-05,
+      "loss": 0.8782,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832594990852176e-05,
+      "loss": 0.9028,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831491302823637e-05,
+      "loss": 0.8853,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983038401938836e-05,
+      "loss": 0.269,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829273140951283e-05,
+      "loss": 0.8363,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982815866791866e-05,
+      "loss": 0.8894,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827040600698054e-05,
+      "loss": 0.8655,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982591893969835e-05,
+      "loss": 0.8633,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824793685329737e-05,
+      "loss": 0.9445,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982366483800373e-05,
+      "loss": 0.8915,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982253239813315e-05,
+      "loss": 0.895,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982139636613213e-05,
+      "loss": 0.8551,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982025674241613e-05,
+      "loss": 0.8606,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819113527401904e-05,
+      "loss": 0.849,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817966721507533e-05,
+      "loss": 0.8536,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816816325152404e-05,
+      "loss": 0.2716,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815662338757224e-05,
+      "loss": 0.8588,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981450476274401e-05,
+      "loss": 0.8889,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813343597536096e-05,
+      "loss": 0.878,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812178843558113e-05,
+      "loss": 0.8628,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981101050123602e-05,
+      "loss": 0.9364,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809838570997085e-05,
+      "loss": 0.911,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980866305326988e-05,
+      "loss": 0.8518,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807483948484306e-05,
+      "loss": 0.9034,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980630125707156e-05,
+      "loss": 0.8456,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805114979464153e-05,
+      "loss": 0.8508,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803925116095915e-05,
+      "loss": 0.2426,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802731667401976e-05,
+      "loss": 0.8978,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801534633818793e-05,
+      "loss": 0.9275,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980033401578412e-05,
+      "loss": 0.9013,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799129813737025e-05,
+      "loss": 0.8874,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797922028117895e-05,
+      "loss": 0.9253,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979671065936841e-05,
+      "loss": 0.9544,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795495707931584e-05,
+      "loss": 0.8977,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979427717425172e-05,
+      "loss": 0.8526,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793055058774443e-05,
+      "loss": 0.2352,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979182936194668e-05,
+      "loss": 0.8777,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790600084216678e-05,
+      "loss": 0.9125,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789367226033985e-05,
+      "loss": 0.8706,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978813078784946e-05,
+      "loss": 0.9156,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786890770115272e-05,
+      "loss": 0.783,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785647173284904e-05,
+      "loss": 0.8884,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784399997813137e-05,
+      "loss": 0.9195,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978314924415607e-05,
+      "loss": 0.9306,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781894912771108e-05,
+      "loss": 0.876,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780637004116966e-05,
+      "loss": 0.8839,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977937551865366e-05,
+      "loss": 0.9109,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778110456842518e-05,
+      "loss": 0.8469,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776841819146185e-05,
+      "loss": 0.9024,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775569606028597e-05,
+      "loss": 0.7855,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774293817955013e-05,
+      "loss": 0.8357,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773014455391987e-05,
+      "loss": 0.9146,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771731518807393e-05,
+      "loss": 0.8587,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770445008670398e-05,
+      "loss": 0.2619,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976915492545149e-05,
+      "loss": 0.8967,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767861269622446e-05,
+      "loss": 0.8954,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976656404165637e-05,
+      "loss": 0.8793,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976526324202765e-05,
+      "loss": 0.8594,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976395887121201e-05,
+      "loss": 0.8164,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762650929686446e-05,
+      "loss": 0.8957,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976133941792929e-05,
+      "loss": 0.8887,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760024336420154e-05,
+      "loss": 0.2357,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758705685639975e-05,
+      "loss": 0.9665,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975738346607098e-05,
+      "loss": 0.8498,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756057678196717e-05,
+      "loss": 0.8602,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975472832250203e-05,
+      "loss": 0.9067,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753395399473067e-05,
+      "loss": 0.8789,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975205890959728e-05,
+      "loss": 0.8708,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750718853363424e-05,
+      "loss": 0.9179,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974937523126157e-05,
+      "loss": 0.8749,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974802804378308e-05,
+      "loss": 0.8677,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746677291420627e-05,
+      "loss": 0.93,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974532297466818e-05,
+      "loss": 0.8456,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743965094021023e-05,
+      "loss": 0.8225,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742603649975738e-05,
+      "loss": 0.8472,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741238643030206e-05,
+      "loss": 0.9183,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973987007368361e-05,
+      "loss": 0.9076,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738497942436446e-05,
+      "loss": 0.9318,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973712224979051e-05,
+      "loss": 0.9194,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735742996248887e-05,
+      "loss": 0.8918,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973436018231598e-05,
+      "loss": 0.9072,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973297380849749e-05,
+      "loss": 0.8704,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731583875300418e-05,
+      "loss": 0.8929,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973019038323306e-05,
+      "loss": 0.8325,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728793332805026e-05,
+      "loss": 0.8909,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727392724527224e-05,
+      "loss": 0.8071,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725988558911855e-05,
+      "loss": 0.8525,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972458083647243e-05,
+      "loss": 0.2505,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723169557723758e-05,
+      "loss": 0.8271,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721754723181944e-05,
+      "loss": 0.8581,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720336333364406e-05,
+      "loss": 0.2619,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718914388789844e-05,
+      "loss": 0.8377,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971748888997827e-05,
+      "loss": 0.276,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716059837450992e-05,
+      "loss": 0.9475,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714627231730626e-05,
+      "loss": 0.8731,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9713191073341074e-05,
+      "loss": 0.8709,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971175136280755e-05,
+      "loss": 0.8528,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971030810065655e-05,
+      "loss": 0.8682,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708861287415887e-05,
+      "loss": 0.8738,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707410923614664e-05,
+      "loss": 0.8619,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705957009783284e-05,
+      "loss": 0.8584,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704499546453445e-05,
+      "loss": 0.8931,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703038534158148e-05,
+      "loss": 0.8515,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970157397343169e-05,
+      "loss": 0.8615,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700105864809663e-05,
+      "loss": 0.8385,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698634208828964e-05,
+      "loss": 0.88,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697159006027777e-05,
+      "loss": 0.8656,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969568025694559e-05,
+      "loss": 0.8578,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694197962123187e-05,
+      "loss": 0.8263,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692712122102644e-05,
+      "loss": 0.9102,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691222737427344e-05,
+      "loss": 0.8818,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968972980864195e-05,
+      "loss": 0.9247,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968823333629244e-05,
+      "loss": 0.8821,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968673332092608e-05,
+      "loss": 0.8666,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685229763091415e-05,
+      "loss": 0.8566,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683722663338318e-05,
+      "loss": 0.7868,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682212022217927e-05,
+      "loss": 0.8672,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96806978402827e-05,
+      "loss": 0.7871,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967918011808637e-05,
+      "loss": 0.8777,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677658856183974e-05,
+      "loss": 0.878,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676134055131842e-05,
+      "loss": 0.8225,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96746057154876e-05,
+      "loss": 0.8437,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673073837810167e-05,
+      "loss": 0.2739,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671538422659754e-05,
+      "loss": 0.88,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966999947059787e-05,
+      "loss": 0.8758,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668456982187306e-05,
+      "loss": 0.8835,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666910957992166e-05,
+      "loss": 0.9146,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665361398577828e-05,
+      "loss": 0.8817,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663808304510975e-05,
+      "loss": 0.8901,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662251676359575e-05,
+      "loss": 0.8422,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660691514692894e-05,
+      "loss": 0.8842,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659127820081484e-05,
+      "loss": 0.8545,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96575605930972e-05,
+      "loss": 0.9035,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655989834313175e-05,
+      "loss": 0.86,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654415544303847e-05,
+      "loss": 0.8969,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965283772364493e-05,
+      "loss": 0.8702,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651256372913447e-05,
+      "loss": 0.8565,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649671492687698e-05,
+      "loss": 0.8507,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9648083083547277e-05,
+      "loss": 0.8286,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646491146073076e-05,
+      "loss": 0.8959,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644895680847263e-05,
+      "loss": 0.8632,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964329668845331e-05,
+      "loss": 0.8282,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641694169475974e-05,
+      "loss": 0.8408,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640088124501303e-05,
+      "loss": 0.8197,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638478554116623e-05,
+      "loss": 0.8917,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963686545891057e-05,
+      "loss": 0.925,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635248839473055e-05,
+      "loss": 0.8626,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633628696395278e-05,
+      "loss": 0.8814,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963200503026973e-05,
+      "loss": 0.9038,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963037784169019e-05,
+      "loss": 0.2253,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628747131251732e-05,
+      "loss": 0.8576,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627112899550704e-05,
+      "loss": 0.8703,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625475147184757e-05,
+      "loss": 0.2782,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962383387475282e-05,
+      "loss": 0.9004,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622189082855106e-05,
+      "loss": 0.8654,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620540772093122e-05,
+      "loss": 0.8197,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618888943069668e-05,
+      "loss": 0.863,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617233596388816e-05,
+      "loss": 0.8317,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961557473265593e-05,
+      "loss": 0.8453,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613912352477672e-05,
+      "loss": 0.8895,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612246456461967e-05,
+      "loss": 0.8987,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610577045218043e-05,
+      "loss": 0.8541,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608904119356408e-05,
+      "loss": 0.9333,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960722767948886e-05,
+      "loss": 0.2547,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605547726228477e-05,
+      "loss": 0.8753,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960386426018962e-05,
+      "loss": 0.8991,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602177281987945e-05,
+      "loss": 0.9108,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600486792240373e-05,
+      "loss": 0.8231,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959879279156513e-05,
+      "loss": 0.8813,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597095280581716e-05,
+      "loss": 0.8328,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595394259910917e-05,
+      "loss": 0.8846,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593689730174796e-05,
+      "loss": 0.894,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959198169199671e-05,
+      "loss": 0.8418,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590270146001296e-05,
+      "loss": 0.8135,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588555092814463e-05,
+      "loss": 0.8416,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958683653306342e-05,
+      "loss": 0.925,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585114467376644e-05,
+      "loss": 0.8752,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583388896383902e-05,
+      "loss": 0.878,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581659820716244e-05,
+      "loss": 0.8726,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579927241005992e-05,
+      "loss": 0.8728,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578191157886755e-05,
+      "loss": 0.856,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576451571993432e-05,
+      "loss": 0.8996,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957470848396219e-05,
+      "loss": 0.8673,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957296189443048e-05,
+      "loss": 0.8623,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957121180403704e-05,
+      "loss": 0.8648,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956945821342188e-05,
+      "loss": 0.7973,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567701123226294e-05,
+      "loss": 0.8555,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565940534092852e-05,
+      "loss": 0.8332,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564176446665412e-05,
+      "loss": 0.8659,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562408861589106e-05,
+      "loss": 0.8243,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560637779510345e-05,
+      "loss": 0.8836,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558863201076818e-05,
+      "loss": 0.9042,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955708512693749e-05,
+      "loss": 0.8438,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9555303557742617e-05,
+      "loss": 0.8976,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553518494143718e-05,
+      "loss": 0.8735,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551729936793595e-05,
+      "loss": 0.8777,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549937886346337e-05,
+      "loss": 0.8452,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548142343457293e-05,
+      "loss": 0.8669,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9546343308783103e-05,
+      "loss": 0.8771,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954454078298168e-05,
+      "loss": 0.8587,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9542734766712213e-05,
+      "loss": 0.2756,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540925260635165e-05,
+      "loss": 0.9208,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539112265412282e-05,
+      "loss": 0.8341,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953729578170658e-05,
+      "loss": 0.8488,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535475810182353e-05,
+      "loss": 0.8854,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953365235150517e-05,
+      "loss": 0.8758,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531825406341874e-05,
+      "loss": 0.8358,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529994975360588e-05,
+      "loss": 0.2357,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528161059230704e-05,
+      "loss": 0.8648,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952632365862289e-05,
+      "loss": 0.8189,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524482774209093e-05,
+      "loss": 0.8943,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522638406662523e-05,
+      "loss": 0.862,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952079055665768e-05,
+      "loss": 0.8751,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518939224870317e-05,
+      "loss": 0.8696,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517084411977482e-05,
+      "loss": 0.84,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515226118657485e-05,
+      "loss": 0.2526,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513364345589903e-05,
+      "loss": 0.8361,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511499093455596e-05,
+      "loss": 0.8397,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509630362936693e-05,
+      "loss": 0.9019,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507758154716594e-05,
+      "loss": 0.8758,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505882469479974e-05,
+      "loss": 0.8858,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950400330791277e-05,
+      "loss": 0.8909,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95021206707022e-05,
+      "loss": 0.7799,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500234558536757e-05,
+      "loss": 0.8672,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949834497210619e-05,
+      "loss": 0.833,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496451912101532e-05,
+      "loss": 0.8897,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494555379215073e-05,
+      "loss": 0.9074,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492655374140386e-05,
+      "loss": 0.8149,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490751897572316e-05,
+      "loss": 0.8628,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488844950206957e-05,
+      "loss": 0.8568,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486934532741692e-05,
+      "loss": 0.9033,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485020645875167e-05,
+      "loss": 0.9011,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483103290307293e-05,
+      "loss": 0.8291,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948118246673926e-05,
+      "loss": 0.916,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947925817587351e-05,
+      "loss": 0.8217,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947733041841377e-05,
+      "loss": 0.9055,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947539919506502e-05,
+      "loss": 0.9027,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473464506533523e-05,
+      "loss": 0.866,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947152635352679e-05,
+      "loss": 0.2276,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946958473675362e-05,
+      "loss": 0.8588,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467639656924056e-05,
+      "loss": 0.9344,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946569111474943e-05,
+      "loss": 0.8753,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946373911094233e-05,
+      "loss": 0.8809,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461783646216605e-05,
+      "loss": 0.8783,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945982472128737e-05,
+      "loss": 0.2406,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457862336871022e-05,
+      "loss": 0.8595,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455896493685205e-05,
+      "loss": 0.8132,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453927192448834e-05,
+      "loss": 0.8483,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451954433882087e-05,
+      "loss": 0.796,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944997821870641e-05,
+      "loss": 0.8682,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447998547644506e-05,
+      "loss": 0.9118,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446015421420356e-05,
+      "loss": 0.916,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944402884075919e-05,
+      "loss": 0.8781,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442038806387507e-05,
+      "loss": 0.2697,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944004531903307e-05,
+      "loss": 0.8629,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94380483794249e-05,
+      "loss": 0.8698,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436047988293286e-05,
+      "loss": 0.8673,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434044146369778e-05,
+      "loss": 0.9107,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432036854387187e-05,
+      "loss": 0.8862,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430026113079586e-05,
+      "loss": 0.8608,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.942801192318231e-05,
+      "loss": 0.8,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425994285431953e-05,
+      "loss": 0.853,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423973200566372e-05,
+      "loss": 0.8803,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421948669324685e-05,
+      "loss": 0.8486,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419920692447264e-05,
+      "loss": 0.8889,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941788927067575e-05,
+      "loss": 0.8661,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415854404753046e-05,
+      "loss": 0.911,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94138160954233e-05,
+      "loss": 0.851,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411774343431933e-05,
+      "loss": 0.8839,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940972914952562e-05,
+      "loss": 0.9103,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940768051445229e-05,
+      "loss": 0.8984,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940562843896114e-05,
+      "loss": 0.8057,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403572923802615e-05,
+      "loss": 0.8867,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401513969728434e-05,
+      "loss": 0.8317,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939945157749155e-05,
+      "loss": 0.8638,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397385747846193e-05,
+      "loss": 0.8276,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395316481547847e-05,
+      "loss": 0.8917,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939324377935324e-05,
+      "loss": 0.9058,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939116764202037e-05,
+      "loss": 0.8497,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938908807030849e-05,
+      "loss": 0.9129,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93870050649781e-05,
+      "loss": 0.8831,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938491862679096e-05,
+      "loss": 0.8598,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9382828756510096e-05,
+      "loss": 0.8465,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380735454899778e-05,
+      "loss": 0.8923,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378638722725528e-05,
+      "loss": 0.8378,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376538560754124e-05,
+      "loss": 0.8494,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937443496975361e-05,
+      "loss": 0.8507,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372327950493276e-05,
+      "loss": 0.2766,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937021750374366e-05,
+      "loss": 0.8968,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936810363027656e-05,
+      "loss": 0.8793,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936598633086503e-05,
+      "loss": 0.9331,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363865606283368e-05,
+      "loss": 0.2538,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361741457307132e-05,
+      "loss": 0.846,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935961388471313e-05,
+      "loss": 0.8979,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935748288927942e-05,
+      "loss": 0.8993,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355348471785316e-05,
+      "loss": 0.8353,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353210633011378e-05,
+      "loss": 0.8804,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351069373739424e-05,
+      "loss": 0.8667,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348924694752516e-05,
+      "loss": 0.8438,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346776596834977e-05,
+      "loss": 0.287,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934462508077236e-05,
+      "loss": 0.8853,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342470147351493e-05,
+      "loss": 0.8188,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340311797360437e-05,
+      "loss": 0.8907,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338150031588502e-05,
+      "loss": 0.8572,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933598485082626e-05,
+      "loss": 0.888,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333816255865528e-05,
+      "loss": 0.8922,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331644247499355e-05,
+      "loss": 0.7865,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932946882652206e-05,
+      "loss": 0.7907,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327289993729198e-05,
+      "loss": 0.8412,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932510774991758e-05,
+      "loss": 0.8787,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322922095885252e-05,
+      "loss": 0.8582,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932073303243152e-05,
+      "loss": 0.8415,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318540560356927e-05,
+      "loss": 0.8578,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931634468046327e-05,
+      "loss": 0.8694,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931414539355359e-05,
+      "loss": 0.8535,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311942700432166e-05,
+      "loss": 0.8782,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309736601904537e-05,
+      "loss": 0.8662,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307527098777477e-05,
+      "loss": 0.903,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305314191859015e-05,
+      "loss": 0.8336,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303097881958406e-05,
+      "loss": 0.8574,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930087816988617e-05,
+      "loss": 0.8857,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298655056454057e-05,
+      "loss": 0.9876,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929642854247507e-05,
+      "loss": 0.8435,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929419862876345e-05,
+      "loss": 0.8566,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291965316134687e-05,
+      "loss": 0.8906,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289728605405506e-05,
+      "loss": 0.907,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928748849739388e-05,
+      "loss": 0.285,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285244992919024e-05,
+      "loss": 0.8613,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92829980928014e-05,
+      "loss": 0.8514,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280747797862697e-05,
+      "loss": 0.8832,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278494108925857e-05,
+      "loss": 0.8836,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276237026815066e-05,
+      "loss": 0.2668,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927397655235574e-05,
+      "loss": 0.8276,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271712686374548e-05,
+      "loss": 0.8103,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269445429699386e-05,
+      "loss": 0.8419,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267174783159402e-05,
+      "loss": 0.8913,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264900747584976e-05,
+      "loss": 0.8676,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262623323807733e-05,
+      "loss": 0.7847,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260342512660526e-05,
+      "loss": 0.8607,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925805831497746e-05,
+      "loss": 0.9057,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255770731593877e-05,
+      "loss": 0.8533,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253479763346347e-05,
+      "loss": 0.8431,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251185411072684e-05,
+      "loss": 0.8951,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248887675611943e-05,
+      "loss": 0.2903,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246586557804414e-05,
+      "loss": 0.8825,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244282058491622e-05,
+      "loss": 0.8608,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241974178516326e-05,
+      "loss": 0.2589,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239662918722527e-05,
+      "loss": 0.8632,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237348279955462e-05,
+      "loss": 0.8629,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92350302630616e-05,
+      "loss": 0.2739,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232708868888642e-05,
+      "loss": 0.8367,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230384098285537e-05,
+      "loss": 0.8381,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228055952102455e-05,
+      "loss": 0.8349,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225724431190813e-05,
+      "loss": 0.8518,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922338953640325e-05,
+      "loss": 0.8457,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221051268593643e-05,
+      "loss": 0.8682,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218709628617107e-05,
+      "loss": 0.8976,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216364617329985e-05,
+      "loss": 0.2494,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214016235589858e-05,
+      "loss": 0.8953,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211664484255532e-05,
+      "loss": 0.8749,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209309364187058e-05,
+      "loss": 0.8954,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206950876245698e-05,
+      "loss": 0.8655,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204589021293967e-05,
+      "loss": 0.8504,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92022238001956e-05,
+      "loss": 0.8734,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199855213815567e-05,
+      "loss": 0.8504,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197483263020068e-05,
+      "loss": 0.7908,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919510794867653e-05,
+      "loss": 0.8232,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9192729271653616e-05,
+      "loss": 0.8762,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190347232821212e-05,
+      "loss": 0.8343,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187961833050442e-05,
+      "loss": 0.8277,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918557307321365e-05,
+      "loss": 0.8673,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918318095418441e-05,
+      "loss": 0.8675,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918078547683753e-05,
+      "loss": 0.8331,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917838664204905e-05,
+      "loss": 0.8837,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175984450696223e-05,
+      "loss": 0.8203,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173578903657542e-05,
+      "loss": 0.8507,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171170001812717e-05,
+      "loss": 0.8573,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91687577460427e-05,
+      "loss": 0.8135,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166342137229656e-05,
+      "loss": 0.9119,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916392317625698e-05,
+      "loss": 0.8571,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161500864009293e-05,
+      "loss": 0.8469,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915907520137244e-05,
+      "loss": 0.8649,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91566461892335e-05,
+      "loss": 0.8655,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154213828480767e-05,
+      "loss": 0.8638,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915177812000376e-05,
+      "loss": 0.8923,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914933906469323e-05,
+      "loss": 0.8904,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146896663441138e-05,
+      "loss": 0.9075,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914445091714069e-05,
+      "loss": 0.8308,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142001826686298e-05,
+      "loss": 0.7935,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91395493929736e-05,
+      "loss": 0.89,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137093616899456e-05,
+      "loss": 0.8239,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913463449936196e-05,
+      "loss": 0.8892,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913217204126041e-05,
+      "loss": 0.8376,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912970624349534e-05,
+      "loss": 0.8592,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127237106968498e-05,
+      "loss": 0.8318,
+      "step": 857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124764632582852e-05,
+      "loss": 0.8675,
+      "step": 858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91222888212426e-05,
+      "loss": 0.8705,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119809673853145e-05,
+      "loss": 0.8258,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9117327191321125e-05,
+      "loss": 0.8896,
+      "step": 861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114841374554392e-05,
+      "loss": 0.8657,
+      "step": 862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911235222446202e-05,
+      "loss": 0.8351,
+      "step": 863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109859741954288e-05,
+      "loss": 0.8723,
+      "step": 864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910736392794271e-05,
+      "loss": 0.8995,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910486478334002e-05,
+      "loss": 0.833,
+      "step": 866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102362309060152e-05,
+      "loss": 0.8947,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099856506018275e-05,
+      "loss": 0.8823,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097347375130766e-05,
+      "loss": 0.9091,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909483491731522e-05,
+      "loss": 0.8508,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909231913349046e-05,
+      "loss": 0.9052,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089800024576503e-05,
+      "loss": 0.8408,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087277591494606e-05,
+      "loss": 0.8984,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084751835167225e-05,
+      "loss": 0.9329,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082222756518036e-05,
+      "loss": 0.8395,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079690356471934e-05,
+      "loss": 0.8678,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077154635955025e-05,
+      "loss": 0.9086,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907461559589463e-05,
+      "loss": 0.8749,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907207323721928e-05,
+      "loss": 0.8495,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069527560858727e-05,
+      "loss": 0.8096,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066978567743933e-05,
+      "loss": 0.8258,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9064426258807073e-05,
+      "loss": 0.9267,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906187063498153e-05,
+      "loss": 0.8524,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9059311697201908e-05,
+      "loss": 0.9044,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056749446404016e-05,
+      "loss": 0.8854,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054183883524876e-05,
+      "loss": 0.9099,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051615009502727e-05,
+      "loss": 0.8706,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904904282527701e-05,
+      "loss": 0.9211,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046467331788383e-05,
+      "loss": 0.2615,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043888529978713e-05,
+      "loss": 0.8911,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041306420791067e-05,
+      "loss": 0.8251,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038721005169746e-05,
+      "loss": 0.8558,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903613228406023e-05,
+      "loss": 0.7696,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033540258409226e-05,
+      "loss": 0.8905,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030944929164655e-05,
+      "loss": 0.8467,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028346297275623e-05,
+      "loss": 0.8603,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025744363692468e-05,
+      "loss": 0.887,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023139129366716e-05,
+      "loss": 0.8688,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902053059525112e-05,
+      "loss": 0.8843,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017918762299622e-05,
+      "loss": 0.8335,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015303631467378e-05,
+      "loss": 0.8253,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012685203710754e-05,
+      "loss": 0.9016,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901006347998731e-05,
+      "loss": 0.8136,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007438461255824e-05,
+      "loss": 0.8754,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900481014847627e-05,
+      "loss": 0.9058,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002178542609833e-05,
+      "loss": 0.2854,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999543644618898e-05,
+      "loss": 0.8997,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996905455467052e-05,
+      "loss": 0.9152,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994263976119096e-05,
+      "loss": 0.3062,
+      "step": 909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899161920754102e-05,
+      "loss": 0.8477,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898897115070003e-05,
+      "loss": 0.9282,
+      "step": 911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8986319806564525e-05,
+      "loss": 0.9184,
+      "step": 912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983665176104113e-05,
+      "loss": 0.9089,
+      "step": 913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981007260289595e-05,
+      "loss": 0.827,
+      "step": 914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978346060092986e-05,
+      "loss": 0.8661,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8975681576487486e-05,
+      "loss": 0.8577,
+      "step": 916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973013810447516e-05,
+      "loss": 0.8439,
+      "step": 917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970342762948676e-05,
+      "loss": 0.7808,
+      "step": 918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967668434967787e-05,
+      "loss": 0.8888,
+      "step": 919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896499082748285e-05,
+      "loss": 0.8267,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962309941473073e-05,
+      "loss": 0.8019,
+      "step": 921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959625777918873e-05,
+      "loss": 0.8607,
+      "step": 922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895693833780185e-05,
+      "loss": 0.8838,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954247622104813e-05,
+      "loss": 0.883,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951553631811763e-05,
+      "loss": 0.7843,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8948856367907904e-05,
+      "loss": 0.9032,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946155831379628e-05,
+      "loss": 0.8429,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894345202321453e-05,
+      "loss": 0.8167,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8940744944401407e-05,
+      "loss": 0.8292,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938034595930244e-05,
+      "loss": 0.812,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935320978792222e-05,
+      "loss": 0.8546,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932604093979718e-05,
+      "loss": 0.8195,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929883942486306e-05,
+      "loss": 0.8396,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927160525306755e-05,
+      "loss": 0.2418,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892443384343703e-05,
+      "loss": 0.8442,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892170389787428e-05,
+      "loss": 0.8251,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891897068961686e-05,
+      "loss": 0.8821,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891623421966431e-05,
+      "loss": 0.9133,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913494489017363e-05,
+      "loss": 0.8037,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910751498677954e-05,
+      "loss": 0.2415,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908005249649198e-05,
+      "loss": 0.8442,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890525574293541e-05,
+      "loss": 0.8793,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890250297954209e-05,
+      "loss": 0.8869,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889974696047593e-05,
+      "loss": 0.8416,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896987686744823e-05,
+      "loss": 0.911,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889422515935784e-05,
+      "loss": 0.8167,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891459379325238e-05,
+      "loss": 0.916,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8888690347658483e-05,
+      "loss": 0.8938,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885918065370212e-05,
+      "loss": 0.8499,
+      "step": 949
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883142533474256e-05,
+      "loss": 0.8635,
+      "step": 950
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888036375298564e-05,
+      "loss": 0.8828,
+      "step": 951
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877581724920576e-05,
+      "loss": 0.9159,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887479645029645e-05,
+      "loss": 0.8647,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887200793013185e-05,
+      "loss": 0.8796,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869216165446547e-05,
+      "loss": 0.8827,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88664211572615e-05,
+      "loss": 0.8383,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886362290659885e-05,
+      "loss": 0.8442,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860821414481923e-05,
+      "loss": 0.8429,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885801668193524e-05,
+      "loss": 0.8665,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855208709984498e-05,
+      "loss": 0.8669,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852397499656575e-05,
+      "loss": 0.8585,
+      "step": 961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849583051979546e-05,
+      "loss": 0.7952,
+      "step": 962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884676536798266e-05,
+      "loss": 0.822,
+      "step": 963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843944448696356e-05,
+      "loss": 0.8772,
+      "step": 964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884112029515225e-05,
+      "loss": 0.9199,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838292908383144e-05,
+      "loss": 0.886,
+      "step": 966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835462289423022e-05,
+      "loss": 0.7994,
+      "step": 967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883262843930705e-05,
+      "loss": 0.8466,
+      "step": 968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882979135907158e-05,
+      "loss": 0.8462,
+      "step": 969
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826951049754136e-05,
+      "loss": 0.7862,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882410751239343e-05,
+      "loss": 0.8541,
+      "step": 971
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882126074802935e-05,
+      "loss": 0.8391,
+      "step": 972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881841075770297e-05,
+      "loss": 0.2635,
+      "step": 973
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815557542456537e-05,
+      "loss": 0.8455,
+      "step": 974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812701103333483e-05,
+      "loss": 0.258,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809841441378416e-05,
+      "loss": 0.2444,
+      "step": 976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8806978557637122e-05,
+      "loss": 0.9096,
+      "step": 977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880411245315657e-05,
+      "loss": 0.8268,
+      "step": 978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801243128984896e-05,
+      "loss": 0.9113,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879837058617143e-05,
+      "loss": 0.8277,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795494825766664e-05,
+      "loss": 0.8339,
+      "step": 981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879261584882227e-05,
+      "loss": 0.8042,
+      "step": 982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789733656391104e-05,
+      "loss": 0.8461,
+      "step": 983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786848249527194e-05,
+      "loss": 0.8858,
+      "step": 984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783959629285736e-05,
+      "loss": 0.8751,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781067796723112e-05,
+      "loss": 0.8073,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877817275289687e-05,
+      "loss": 0.8418,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775274498865745e-05,
+      "loss": 0.8605,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772373035689628e-05,
+      "loss": 0.8499,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769468364429596e-05,
+      "loss": 0.7993,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8766560486147897e-05,
+      "loss": 0.8186,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763649401907953e-05,
+      "loss": 0.7918,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760735112774353e-05,
+      "loss": 0.8993,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875781761981286e-05,
+      "loss": 0.8664,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875489692409042e-05,
+      "loss": 0.9231,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875197302667513e-05,
+      "loss": 0.864,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874904592863627e-05,
+      "loss": 0.7937,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746115631044293e-05,
+      "loss": 0.8377,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743182134970815e-05,
+      "loss": 0.8852,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874024544148863e-05,
+      "loss": 0.8599,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737305551671686e-05,
+      "loss": 0.8251,
+      "step": 1001
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734362466595117e-05,
+      "loss": 0.8494,
+      "step": 1002
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731416187335213e-05,
+      "loss": 0.8572,
+      "step": 1003
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872846671496944e-05,
+      "loss": 0.8401,
+      "step": 1004
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725514050576433e-05,
+      "loss": 0.8798,
+      "step": 1005
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722558195235986e-05,
+      "loss": 0.8137,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719599150029068e-05,
+      "loss": 0.833,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716636916037806e-05,
+      "loss": 0.8426,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87136714943455e-05,
+      "loss": 0.8698,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710702886036618e-05,
+      "loss": 0.8439,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707731092196784e-05,
+      "loss": 0.824,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704756113912793e-05,
+      "loss": 0.8778,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701777952272605e-05,
+      "loss": 0.8474,
+      "step": 1013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8698796608365345e-05,
+      "loss": 0.8127,
+      "step": 1014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695812083281297e-05,
+      "loss": 0.8774,
+      "step": 1015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869282437811191e-05,
+      "loss": 0.8323,
+      "step": 1016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86898334939498e-05,
+      "loss": 0.8924,
+      "step": 1017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8686839431888736e-05,
+      "loss": 0.8767,
+      "step": 1018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683842193023664e-05,
+      "loss": 0.8664,
+      "step": 1019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680841778450683e-05,
+      "loss": 0.8565,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867783818926705e-05,
+      "loss": 0.8848,
+      "step": 1021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8674831426571187e-05,
+      "loss": 0.8429,
+      "step": 1022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8671821491462675e-05,
+      "loss": 0.8639,
+      "step": 1023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668808385042264e-05,
+      "loss": 0.8581,
+      "step": 1024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665792108411847e-05,
+      "loss": 0.8874,
+      "step": 1025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662772662674492e-05,
+      "loss": 0.8832,
+      "step": 1026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8659750048934415e-05,
+      "loss": 0.8512,
+      "step": 1027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8656724268297e-05,
+      "loss": 0.8827,
+      "step": 1028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653695321868783e-05,
+      "loss": 0.8515,
+      "step": 1029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650663210757457e-05,
+      "loss": 0.7963,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647627936071876e-05,
+      "loss": 0.8279,
+      "step": 1031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8644589498922048e-05,
+      "loss": 0.8409,
+      "step": 1032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864154790041914e-05,
+      "loss": 0.865,
+      "step": 1033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638503141675476e-05,
+      "loss": 0.8695,
+      "step": 1034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635455223804528e-05,
+      "loss": 0.8578,
+      "step": 1035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8632404147920934e-05,
+      "loss": 0.8923,
+      "step": 1036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862934991514048e-05,
+      "loss": 0.9226,
+      "step": 1037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862629252658011e-05,
+      "loss": 0.7926,
+      "step": 1038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623231983357916e-05,
+      "loss": 0.8981,
+      "step": 1039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862016828659315e-05,
+      "loss": 0.8863,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8617101437406223e-05,
+      "loss": 0.8466,
+      "step": 1041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861403143691868e-05,
+      "loss": 0.8563,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610958286253233e-05,
+      "loss": 0.8346,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8607881986533747e-05,
+      "loss": 0.8137,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860480253888523e-05,
+      "loss": 0.8331,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860171994443385e-05,
+      "loss": 0.8911,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598634204306917e-05,
+      "loss": 0.8786,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595545319632892e-05,
+      "loss": 0.8959,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592453291541402e-05,
+      "loss": 0.8326,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589358121163198e-05,
+      "loss": 0.8252,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85862598096302e-05,
+      "loss": 0.8098,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583158358075474e-05,
+      "loss": 0.8548,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858005376763322e-05,
+      "loss": 0.2826,
+      "step": 1053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576946039438804e-05,
+      "loss": 0.9212,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857383517462873e-05,
+      "loss": 0.9013,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857072117434065e-05,
+      "loss": 0.8736,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567604039713366e-05,
+      "loss": 0.2689,
+      "step": 1057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564483771886822e-05,
+      "loss": 0.8624,
+      "step": 1058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561360372002107e-05,
+      "loss": 0.8422,
+      "step": 1059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558233841201462e-05,
+      "loss": 0.8723,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855510418062827e-05,
+      "loss": 0.8195,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855197139142705e-05,
+      "loss": 0.8233,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854883547474348e-05,
+      "loss": 0.8445,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545696431724375e-05,
+      "loss": 0.8886,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542554263517684e-05,
+      "loss": 0.855,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853940897127252e-05,
+      "loss": 0.8352,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536260556139117e-05,
+      "loss": 0.864,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853310901926886e-05,
+      "loss": 0.806,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852995436181428e-05,
+      "loss": 0.866,
+      "step": 1069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526796584929047e-05,
+      "loss": 0.8853,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852363568976796e-05,
+      "loss": 0.8139,
+      "step": 1071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520471677486977e-05,
+      "loss": 0.8584,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517304549243184e-05,
+      "loss": 0.8695,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851413430619481e-05,
+      "loss": 0.8174,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8510960949501227e-05,
+      "loss": 0.8221,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507784480322934e-05,
+      "loss": 0.8842,
+      "step": 1076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504604899821575e-05,
+      "loss": 0.8349,
+      "step": 1077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501422209159942e-05,
+      "loss": 0.8878,
+      "step": 1078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498236409501945e-05,
+      "loss": 0.7919,
+      "step": 1079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495047502012643e-05,
+      "loss": 0.8699,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491855487858233e-05,
+      "loss": 0.8417,
+      "step": 1081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488660368206044e-05,
+      "loss": 0.8739,
+      "step": 1082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848546214422454e-05,
+      "loss": 0.8815,
+      "step": 1083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8482260817083317e-05,
+      "loss": 0.8126,
+      "step": 1084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8479056387953115e-05,
+      "loss": 0.8365,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475848858005803e-05,
+      "loss": 0.8019,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8472638228414385e-05,
+      "loss": 0.8553,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8469424500353e-05,
+      "loss": 0.909,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466207674996912e-05,
+      "loss": 0.8061,
+      "step": 1089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462987753522526e-05,
+      "loss": 0.8699,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459764737107384e-05,
+      "loss": 0.9263,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8456538626930148e-05,
+      "loss": 0.8591,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845330942417061e-05,
+      "loss": 0.2815,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845007713000971e-05,
+      "loss": 0.838,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446841745629507e-05,
+      "loss": 0.8388,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8443603272213183e-05,
+      "loss": 0.8837,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8440361710945067e-05,
+      "loss": 0.8657,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8437117063010598e-05,
+      "loss": 0.8292,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433869329596364e-05,
+      "loss": 0.8345,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430618511890066e-05,
+      "loss": 0.8616,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427364611080536e-05,
+      "loss": 0.8205,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842410762835774e-05,
+      "loss": 0.8514,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420847564912764e-05,
+      "loss": 0.8441,
+      "step": 1103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417584421937825e-05,
+      "loss": 0.8801,
+      "step": 1104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414318200626266e-05,
+      "loss": 0.8776,
+      "step": 1105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411048902172548e-05,
+      "loss": 0.8542,
+      "step": 1106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407776527772268e-05,
+      "loss": 0.8253,
+      "step": 1107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404501078622143e-05,
+      "loss": 0.881,
+      "step": 1108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401222555920015e-05,
+      "loss": 0.8573,
+      "step": 1109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397940960864846e-05,
+      "loss": 0.8182,
+      "step": 1110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394656294656733e-05,
+      "loss": 0.8563,
+      "step": 1111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391368558496883e-05,
+      "loss": 0.8195,
+      "step": 1112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388077753587634e-05,
+      "loss": 0.8366,
+      "step": 1113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838478388113244e-05,
+      "loss": 0.8215,
+      "step": 1114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381486942335885e-05,
+      "loss": 0.8638,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8378186938403667e-05,
+      "loss": 0.8335,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374883870542607e-05,
+      "loss": 0.8443,
+      "step": 1117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837157773996065e-05,
+      "loss": 0.8495,
+      "step": 1118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368268547866854e-05,
+      "loss": 0.9059,
+      "step": 1119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83649562954714e-05,
+      "loss": 0.8153,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8361640983985598e-05,
+      "loss": 0.8849,
+      "step": 1121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835832261462186e-05,
+      "loss": 0.8828,
+      "step": 1122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355001188593724e-05,
+      "loss": 0.8456,
+      "step": 1123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351676707115847e-05,
+      "loss": 0.8228,
+      "step": 1124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348349171404004e-05,
+      "loss": 0.8095,
+      "step": 1125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345018582675083e-05,
+      "loss": 0.8228,
+      "step": 1126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834168494214709e-05,
+      "loss": 0.7907,
+      "step": 1127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833834825103915e-05,
+      "loss": 0.28,
+      "step": 1128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8335008510571503e-05,
+      "loss": 0.8367,
+      "step": 1129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331665721965494e-05,
+      "loss": 0.8878,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83283198864436e-05,
+      "loss": 0.8898,
+      "step": 1131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324971005229397e-05,
+      "loss": 0.86,
+      "step": 1132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321619079547584e-05,
+      "loss": 0.2553,
+      "step": 1133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318264110623972e-05,
+      "loss": 0.7844,
+      "step": 1134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8314906099685484e-05,
+      "loss": 0.8437,
+      "step": 1135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831154504796015e-05,
+      "loss": 0.8276,
+      "step": 1136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830818095667712e-05,
+      "loss": 0.8167,
+      "step": 1137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304813827066656e-05,
+      "loss": 0.8641,
+      "step": 1138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301443660360125e-05,
+      "loss": 0.8808,
+      "step": 1139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298070457790003e-05,
+      "loss": 0.8231,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294694220589883e-05,
+      "loss": 0.8152,
+      "step": 1141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829131494999447e-05,
+      "loss": 0.272,
+      "step": 1142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828793264723957e-05,
+      "loss": 0.8368,
+      "step": 1143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284547313562103e-05,
+      "loss": 0.8148,
+      "step": 1144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8281158950200095e-05,
+      "loss": 0.8523,
+      "step": 1145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8277767558392677e-05,
+      "loss": 0.8549,
+      "step": 1146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82743731393801e-05,
+      "loss": 0.8504,
+      "step": 1147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827097569440371e-05,
+      "loss": 0.8625,
+      "step": 1148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267575224705955e-05,
+      "loss": 0.8299,
+      "step": 1149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826417173153041e-05,
+      "loss": 0.9033,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8260765216121738e-05,
+      "loss": 0.8124,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8257355679725708e-05,
+      "loss": 0.872,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82539431235892e-05,
+      "loss": 0.8415,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82505275489602e-05,
+      "loss": 0.8655,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247108957087794e-05,
+      "loss": 0.8521,
+      "step": 1155
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243687349222168e-05,
+      "loss": 0.8997,
+      "step": 1156
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824026272661462e-05,
+      "loss": 0.7476,
+      "step": 1157
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236835090517533e-05,
+      "loss": 0.8754,
+      "step": 1158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233404442184417e-05,
+      "loss": 0.8434,
+      "step": 1159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8229970782869867e-05,
+      "loss": 0.2846,
+      "step": 1160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822653411382958e-05,
+      "loss": 0.2893,
+      "step": 1161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223094436320358e-05,
+      "loss": 0.8161,
+      "step": 1162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219651751600104e-05,
+      "loss": 0.8922,
+      "step": 1163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8216206060927815e-05,
+      "loss": 0.8587,
+      "step": 1164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212757365563593e-05,
+      "loss": 0.8018,
+      "step": 1165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8209305666768634e-05,
+      "loss": 0.262,
+      "step": 1166
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820585096580524e-05,
+      "loss": 0.813,
+      "step": 1167
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202393263936796e-05,
+      "loss": 0.8704,
+      "step": 1168
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198932562427807e-05,
+      "loss": 0.844,
+      "step": 1169
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195468862543853e-05,
+      "loss": 0.8754,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192002165551624e-05,
+      "loss": 0.8784,
+      "step": 1171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81885324727189e-05,
+      "loss": 0.8298,
+      "step": 1172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818505978531456e-05,
+      "loss": 0.816,
+      "step": 1173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181584104608574e-05,
+      "loss": 0.2706,
+      "step": 1174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178105431872008e-05,
+      "loss": 0.7861,
+      "step": 1175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174623768377027e-05,
+      "loss": 0.8718,
+      "step": 1176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817113911539689e-05,
+      "loss": 0.8134,
+      "step": 1177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167651474205935e-05,
+      "loss": 0.8835,
+      "step": 1178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816416084607961e-05,
+      "loss": 0.8434,
+      "step": 1179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160667232294446e-05,
+      "loss": 0.8354,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157170634128073e-05,
+      "loss": 0.8469,
+      "step": 1181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153671052859202e-05,
+      "loss": 0.8206,
+      "step": 1182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815016848976764e-05,
+      "loss": 0.8371,
+      "step": 1183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146662946134294e-05,
+      "loss": 0.8982,
+      "step": 1184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814315442324114e-05,
+      "loss": 0.8285,
+      "step": 1185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139642922371263e-05,
+      "loss": 0.8502,
+      "step": 1186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136128444808833e-05,
+      "loss": 0.8319,
+      "step": 1187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132610991839095e-05,
+      "loss": 0.8579,
+      "step": 1188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81290905647484e-05,
+      "loss": 0.8514,
+      "step": 1189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812556716482418e-05,
+      "loss": 0.8496,
+      "step": 1190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812204079335494e-05,
+      "loss": 0.9273,
+      "step": 1191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8118511451630306e-05,
+      "loss": 0.8653,
+      "step": 1192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114979140940957e-05,
+      "loss": 0.8799,
+      "step": 1193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111443862578667e-05,
+      "loss": 0.8186,
+      "step": 1194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107905617836308e-05,
+      "loss": 0.887,
+      "step": 1195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810436440800782e-05,
+      "loss": 0.865,
+      "step": 1196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100820234388236e-05,
+      "loss": 0.7957,
+      "step": 1197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809727309827367e-05,
+      "loss": 0.8647,
+      "step": 1198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8093723000961324e-05,
+      "loss": 0.8545,
+      "step": 1199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.8677,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086613927937492e-05,
+      "loss": 0.9183,
+      "step": 1201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083054954825815e-05,
+      "loss": 0.8757,
+      "step": 1202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807949302571598e-05,
+      "loss": 0.8724,
+      "step": 1203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807592814191059e-05,
+      "loss": 0.9145,
+      "step": 1204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8072360304713327e-05,
+      "loss": 0.8318,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068789515428973e-05,
+      "loss": 0.8353,
+      "step": 1206
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806521577536337e-05,
+      "loss": 0.8527,
+      "step": 1207
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806163908582345e-05,
+      "loss": 0.8447,
+      "step": 1208
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058059448117214e-05,
+      "loss": 0.8739,
+      "step": 1209
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054476863553748e-05,
+      "loss": 0.917,
+      "step": 1210
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050891333443216e-05,
+      "loss": 0.8444,
+      "step": 1211
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8047302859096856e-05,
+      "loss": 0.7929,
+      "step": 1212
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043711441826986e-05,
+      "loss": 0.2754,
+      "step": 1213
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040117082946996e-05,
+      "loss": 0.8727,
+      "step": 1214
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036519783771358e-05,
+      "loss": 0.8755,
+      "step": 1215
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803291954561561e-05,
+      "loss": 0.9381,
+      "step": 1216
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802931636979637e-05,
+      "loss": 0.8403,
+      "step": 1217
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802571025763134e-05,
+      "loss": 0.8532,
+      "step": 1218
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022101210439275e-05,
+      "loss": 0.2606,
+      "step": 1219
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018489229540023e-05,
+      "loss": 0.8314,
+      "step": 1220
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801487431625449e-05,
+      "loss": 0.8144,
+      "step": 1221
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8011256471904665e-05,
+      "loss": 0.8588,
+      "step": 1222
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007635697813605e-05,
+      "loss": 0.8257,
+      "step": 1223
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8004011995305437e-05,
+      "loss": 0.8037,
+      "step": 1224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800038536570536e-05,
+      "loss": 0.8519,
+      "step": 1225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996755810339643e-05,
+      "loss": 0.8671,
+      "step": 1226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993123330535626e-05,
+      "loss": 0.8647,
+      "step": 1227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989487927621724e-05,
+      "loss": 0.8614,
+      "step": 1228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79858496029274e-05,
+      "loss": 0.8234,
+      "step": 1229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798220835778322e-05,
+      "loss": 0.919,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7978564193520783e-05,
+      "loss": 0.7391,
+      "step": 1231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797491711147278e-05,
+      "loss": 0.863,
+      "step": 1232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971267112972957e-05,
+      "loss": 0.8555,
+      "step": 1233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796761419935613e-05,
+      "loss": 0.8432,
+      "step": 1234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7963958371958182e-05,
+      "loss": 0.8517,
+      "step": 1235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7960299632116056e-05,
+      "loss": 0.8892,
+      "step": 1236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956637981167765e-05,
+      "loss": 0.8772,
+      "step": 1237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7952973420452395e-05,
+      "loss": 0.8196,
+      "step": 1238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949305951310077e-05,
+      "loss": 0.8432,
+      "step": 1239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794563557508202e-05,
+      "loss": 0.8803,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941962293110494e-05,
+      "loss": 0.2759,
+      "step": 1241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7938286106738825e-05,
+      "loss": 0.8609,
+      "step": 1242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793460701731141e-05,
+      "loss": 0.8226,
+      "step": 1243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930925026173697e-05,
+      "loss": 0.862,
+      "step": 1244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792724013467221e-05,
+      "loss": 0.8474,
+      "step": 1245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792355234415452e-05,
+      "loss": 0.8814,
+      "step": 1246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791986165596926e-05,
+      "loss": 0.8428,
+      "step": 1247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7916168071466137e-05,
+      "loss": 0.8547,
+      "step": 1248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912471591995893e-05,
+      "loss": 0.9156,
+      "step": 1249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908772218910356e-05,
+      "loss": 0.8359,
+      "step": 1250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905069953562385e-05,
+      "loss": 0.8561,
+      "step": 1251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790136479730592e-05,
+      "loss": 0.8329,
+      "step": 1252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789765675149594e-05,
+      "loss": 0.8361,
+      "step": 1253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893945817488494e-05,
+      "loss": 0.8648,
+      "step": 1254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789023199664068e-05,
+      "loss": 0.8375,
+      "step": 1255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7886515290310656e-05,
+      "loss": 0.8891,
+      "step": 1256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788279569985763e-05,
+      "loss": 0.8698,
+      "step": 1257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787907322664187e-05,
+      "loss": 0.8481,
+      "step": 1258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787534787202469e-05,
+      "loss": 0.8348,
+      "step": 1259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871619637368473e-05,
+      "loss": 0.7722,
+      "step": 1260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786788852403664e-05,
+      "loss": 0.8533,
+      "step": 1261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7864154533393674e-05,
+      "loss": 0.862,
+      "step": 1262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7860417666805104e-05,
+      "loss": 0.889,
+      "step": 1263
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785667792563751e-05,
+      "loss": 0.2942,
+      "step": 1264
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7852935311258542e-05,
+      "loss": 0.8387,
+      "step": 1265
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784918982503687e-05,
+      "loss": 0.82,
+      "step": 1266
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845441468342236e-05,
+      "loss": 0.2552,
+      "step": 1267
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7841690242545423e-05,
+      "loss": 0.7933,
+      "step": 1268
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837936149018272e-05,
+      "loss": 0.8835,
+      "step": 1269
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834179189133657e-05,
+      "loss": 0.885,
+      "step": 1270
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783041936426552e-05,
+      "loss": 0.8501,
+      "step": 1271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782665667578883e-05,
+      "loss": 0.8819,
+      "step": 1272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782289112507962e-05,
+      "loss": 0.9085,
+      "step": 1273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819122713514968e-05,
+      "loss": 0.8835,
+      "step": 1274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7815351442472984e-05,
+      "loss": 0.8735,
+      "step": 1275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811577313332843e-05,
+      "loss": 0.8121,
+      "step": 1276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807800327474748e-05,
+      "loss": 0.8125,
+      "step": 1277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780402048627996e-05,
+      "loss": 0.8422,
+      "step": 1278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800237791130772e-05,
+      "loss": 0.8865,
+      "step": 1279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796452243410533e-05,
+      "loss": 0.8802,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792663844503628e-05,
+      "loss": 0.8485,
+      "step": 1281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788872595795486e-05,
+      "loss": 0.8612,
+      "step": 1282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778507849867258e-05,
+      "loss": 0.7774,
+      "step": 1283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7781281554522414e-05,
+      "loss": 0.856,
+      "step": 1284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777481764733558e-05,
+      "loss": 0.8362,
+      "step": 1285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773679130695596e-05,
+      "loss": 0.8481,
+      "step": 1286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769873653799162e-05,
+      "loss": 0.7439,
+      "step": 1287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766065335435937e-05,
+      "loss": 0.8328,
+      "step": 1288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776225417699863e-05,
+      "loss": 0.8989,
+      "step": 1289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758440179880998e-05,
+      "loss": 0.88,
+      "step": 1290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754623345477825e-05,
+      "loss": 0.814,
+      "step": 1291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750803675184946e-05,
+      "loss": 0.8363,
+      "step": 1292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7746981170399223e-05,
+      "loss": 0.8643,
+      "step": 1293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743155832518557e-05,
+      "loss": 0.7868,
+      "step": 1294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739327662941888e-05,
+      "loss": 0.8777,
+      "step": 1295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735496663069187e-05,
+      "loss": 0.8408,
+      "step": 1296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731662834301464e-05,
+      "loss": 0.8677,
+      "step": 1297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772782617804076e-05,
+      "loss": 0.7729,
+      "step": 1298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723986695690155e-05,
+      "loss": 0.8285,
+      "step": 1299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772014438865376e-05,
+      "loss": 0.8583,
+      "step": 1300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716299258336717e-05,
+      "loss": 0.8126,
+      "step": 1301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712451306145198e-05,
+      "loss": 0.8234,
+      "step": 1302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770860053348642e-05,
+      "loss": 0.829,
+      "step": 1303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7704746941768615e-05,
+      "loss": 0.9123,
+      "step": 1304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700890532401052e-05,
+      "loss": 0.8725,
+      "step": 1305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769703130679404e-05,
+      "loss": 0.8904,
+      "step": 1306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693169266358906e-05,
+      "loss": 0.7968,
+      "step": 1307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7689304412508008e-05,
+      "loss": 0.8874,
+      "step": 1308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7685436746654736e-05,
+      "loss": 0.2709,
+      "step": 1309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768156627021351e-05,
+      "loss": 0.7886,
+      "step": 1310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7677692984599768e-05,
+      "loss": 0.8837,
+      "step": 1311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7673816891229992e-05,
+      "loss": 0.7972,
+      "step": 1312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669937991521673e-05,
+      "loss": 0.8349,
+      "step": 1313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7666056286893346e-05,
+      "loss": 0.8581,
+      "step": 1314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766217177876455e-05,
+      "loss": 0.8228,
+      "step": 1315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658284468555875e-05,
+      "loss": 0.8147,
+      "step": 1316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765439435768891e-05,
+      "loss": 0.8362,
+      "step": 1317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765050144758629e-05,
+      "loss": 0.8554,
+      "step": 1318
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646605739671662e-05,
+      "loss": 0.8882,
+      "step": 1319
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642707235369692e-05,
+      "loss": 0.8345,
+      "step": 1320
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638805936106083e-05,
+      "loss": 0.8818,
+      "step": 1321
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763490184330755e-05,
+      "loss": 0.8238,
+      "step": 1322
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7630994958401826e-05,
+      "loss": 0.7998,
+      "step": 1323
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7627085282817682e-05,
+      "loss": 0.8772,
+      "step": 1324
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7623172817984892e-05,
+      "loss": 0.8744,
+      "step": 1325
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619257565334253e-05,
+      "loss": 0.2782,
+      "step": 1326
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761533952629759e-05,
+      "loss": 0.8304,
+      "step": 1327
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761141870230774e-05,
+      "loss": 0.8392,
+      "step": 1328
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7607495094798562e-05,
+      "loss": 0.8597,
+      "step": 1329
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603568705204932e-05,
+      "loss": 0.8571,
+      "step": 1330
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759963953496274e-05,
+      "loss": 0.8387,
+      "step": 1331
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75957075855089e-05,
+      "loss": 0.8276,
+      "step": 1332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759177285828133e-05,
+      "loss": 0.8353,
+      "step": 1333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758783535471898e-05,
+      "loss": 0.8599,
+      "step": 1334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7583895076261804e-05,
+      "loss": 0.8496,
+      "step": 1335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579952024350774e-05,
+      "loss": 0.8979,
+      "step": 1336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7576006200427876e-05,
+      "loss": 0.8621,
+      "step": 1337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757205760593611e-05,
+      "loss": 0.8466,
+      "step": 1338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568106242319492e-05,
+      "loss": 0.8342,
+      "step": 1339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564152111023047e-05,
+      "loss": 0.817,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756019521349281e-05,
+      "loss": 0.8693,
+      "step": 1341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755623555117583e-05,
+      "loss": 0.8377,
+      "step": 1342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755227312552017e-05,
+      "loss": 0.8344,
+      "step": 1343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7548307937974906e-05,
+      "loss": 0.8346,
+      "step": 1344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754433998999011e-05,
+      "loss": 0.8843,
+      "step": 1345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7540369283016877e-05,
+      "loss": 0.8507,
+      "step": 1346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753639581850731e-05,
+      "loss": 0.8226,
+      "step": 1347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532419597914513e-05,
+      "loss": 0.8238,
+      "step": 1348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528440622692608e-05,
+      "loss": 0.7708,
+      "step": 1349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524458894296708e-05,
+      "loss": 0.8622,
+      "step": 1350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752047441418295e-05,
+      "loss": 0.8467,
+      "step": 1351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516487183808476e-05,
+      "loss": 0.8179,
+      "step": 1352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7512497204631417e-05,
+      "loss": 0.8705,
+      "step": 1353
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750850447811093e-05,
+      "loss": 0.8147,
+      "step": 1354
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7504509005707162e-05,
+      "loss": 0.8391,
+      "step": 1355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500510788881273e-05,
+      "loss": 0.8133,
+      "step": 1356
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7496509829095424e-05,
+      "loss": 0.8178,
+      "step": 1357
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749250612781277e-05,
+      "loss": 0.844,
+      "step": 1358
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7488499686497484e-05,
+      "loss": 0.8022,
+      "step": 1359
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484490506614737e-05,
+      "loss": 0.8639,
+      "step": 1360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7480478589630693e-05,
+      "loss": 0.812,
+      "step": 1361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7476463937012522e-05,
+      "loss": 0.764,
+      "step": 1362
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7472446550228396e-05,
+      "loss": 0.855,
+      "step": 1363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746842643074749e-05,
+      "loss": 0.8827,
+      "step": 1364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746440358003997e-05,
+      "loss": 0.8233,
+      "step": 1365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7460377999577e-05,
+      "loss": 0.8527,
+      "step": 1366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745634969083076e-05,
+      "loss": 0.8195,
+      "step": 1367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452318655274407e-05,
+      "loss": 0.8288,
+      "step": 1368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74482848943821e-05,
+      "loss": 0.8696,
+      "step": 1369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7444248409629002e-05,
+      "loss": 0.8681,
+      "step": 1370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7440209202491267e-05,
+      "loss": 0.8367,
+      "step": 1371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7436167274446048e-05,
+      "loss": 0.8119,
+      "step": 1372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743212262697149e-05,
+      "loss": 0.8799,
+      "step": 1373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742807526154673e-05,
+      "loss": 0.8885,
+      "step": 1374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7424025179651903e-05,
+      "loss": 0.8775,
+      "step": 1375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741997238276814e-05,
+      "loss": 0.821,
+      "step": 1376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415916872377553e-05,
+      "loss": 0.8498,
+      "step": 1377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411858649963268e-05,
+      "loss": 0.7992,
+      "step": 1378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740779771700938e-05,
+      "loss": 0.8612,
+      "step": 1379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7403734075000983e-05,
+      "loss": 0.859,
+      "step": 1380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399667725424165e-05,
+      "loss": 0.8261,
+      "step": 1381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739559866976601e-05,
+      "loss": 0.7988,
+      "step": 1382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391526909514572e-05,
+      "loss": 0.8736,
+      "step": 1383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738745244615892e-05,
+      "loss": 0.285,
+      "step": 1384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738337528118909e-05,
+      "loss": 0.874,
+      "step": 1385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737929541609611e-05,
+      "loss": 0.8903,
+      "step": 1386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7375212852372008e-05,
+      "loss": 0.2447,
+      "step": 1387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371127591509783e-05,
+      "loss": 0.906,
+      "step": 1388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367039635003435e-05,
+      "loss": 0.8471,
+      "step": 1389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362948984347932e-05,
+      "loss": 0.8309,
+      "step": 1390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735885564103925e-05,
+      "loss": 0.8248,
+      "step": 1391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7354759606574327e-05,
+      "loss": 0.8317,
+      "step": 1392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350660882451102e-05,
+      "loss": 0.2801,
+      "step": 1393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346559470168482e-05,
+      "loss": 0.3013,
+      "step": 1394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734245537122638e-05,
+      "loss": 0.7937,
+      "step": 1395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338348587125666e-05,
+      "loss": 0.7768,
+      "step": 1396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733423911936821e-05,
+      "loss": 0.8973,
+      "step": 1397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733012696945685e-05,
+      "loss": 0.8615,
+      "step": 1398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7326012138895422e-05,
+      "loss": 0.8404,
+      "step": 1399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7321894629188726e-05,
+      "loss": 0.8676,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317774441842547e-05,
+      "loss": 0.8009,
+      "step": 1401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313651578363654e-05,
+      "loss": 0.8132,
+      "step": 1402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730952604025979e-05,
+      "loss": 0.8548,
+      "step": 1403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305397829039674e-05,
+      "loss": 0.8335,
+      "step": 1404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7301266946213008e-05,
+      "loss": 0.8026,
+      "step": 1405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729713339329047e-05,
+      "loss": 0.8876,
+      "step": 1406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729299717178371e-05,
+      "loss": 0.8263,
+      "step": 1407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728885828320536e-05,
+      "loss": 0.8553,
+      "step": 1408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728471672906902e-05,
+      "loss": 0.8663,
+      "step": 1409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7280572510889273e-05,
+      "loss": 0.8769,
+      "step": 1410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7276425630181672e-05,
+      "loss": 0.8215,
+      "step": 1411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7272276088462744e-05,
+      "loss": 0.9093,
+      "step": 1412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268123887249988e-05,
+      "loss": 0.8395,
+      "step": 1413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7263969028061874e-05,
+      "loss": 0.8486,
+      "step": 1414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7259811512417854e-05,
+      "loss": 0.8304,
+      "step": 1415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725565134183834e-05,
+      "loss": 0.873,
+      "step": 1416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7251488517844722e-05,
+      "loss": 0.8081,
+      "step": 1417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247323041959352e-05,
+      "loss": 0.3288,
+      "step": 1418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243154915705563e-05,
+      "loss": 0.8377,
+      "step": 1419
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238984140607647e-05,
+      "loss": 0.8372,
+      "step": 1420
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234810718190874e-05,
+      "loss": 0.8509,
+      "step": 1421
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7230634649981476e-05,
+      "loss": 0.2943,
+      "step": 1422
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722645593750665e-05,
+      "loss": 0.8382,
+      "step": 1423
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722227458229457e-05,
+      "loss": 0.8234,
+      "step": 1424
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218090585874363e-05,
+      "loss": 0.839,
+      "step": 1425
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721390394977614e-05,
+      "loss": 0.839,
+      "step": 1426
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7209714675530958e-05,
+      "loss": 0.7934,
+      "step": 1427
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720552276467085e-05,
+      "loss": 0.8831,
+      "step": 1428
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201328218728814e-05,
+      "loss": 0.8085,
+      "step": 1429
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197131039238805e-05,
+      "loss": 0.8905,
+      "step": 1430
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7192931227735746e-05,
+      "loss": 0.8721,
+      "step": 1431
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188728785755518e-05,
+      "loss": 0.8175,
+      "step": 1432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184523714834968e-05,
+      "loss": 0.9075,
+      "step": 1433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180316016511907e-05,
+      "loss": 0.8667,
+      "step": 1434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7176105692325104e-05,
+      "loss": 0.8601,
+      "step": 1435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171892743814277e-05,
+      "loss": 0.7616,
+      "step": 1436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716767717252013e-05,
+      "loss": 0.8513,
+      "step": 1437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7163458979984297e-05,
+      "loss": 0.8801,
+      "step": 1438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715923816774939e-05,
+      "loss": 0.8767,
+      "step": 1439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715501473735897e-05,
+      "loss": 0.9261,
+      "step": 1440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150788690357564e-05,
+      "loss": 0.8523,
+      "step": 1441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714656002829065e-05,
+      "loss": 0.7789,
+      "step": 1442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142328752704658e-05,
+      "loss": 0.8822,
+      "step": 1443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713809486514698e-05,
+      "loss": 0.7681,
+      "step": 1444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713385836716596e-05,
+      "loss": 0.8131,
+      "step": 1445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7129619260310902e-05,
+      "loss": 0.8459,
+      "step": 1446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125377546132063e-05,
+      "loss": 0.759,
+      "step": 1447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121133226180644e-05,
+      "loss": 0.8305,
+      "step": 1448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116886302008812e-05,
+      "loss": 0.8345,
+      "step": 1449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112636775169674e-05,
+      "loss": 0.8177,
+      "step": 1450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71083846472173e-05,
+      "loss": 0.8893,
+      "step": 1451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7104129919706704e-05,
+      "loss": 0.8131,
+      "step": 1452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099872594193855e-05,
+      "loss": 0.7916,
+      "step": 1453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709561267223567e-05,
+      "loss": 0.8124,
+      "step": 1454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709135015539001e-05,
+      "loss": 0.8295,
+      "step": 1455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70870850452157e-05,
+      "loss": 0.8554,
+      "step": 1456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082817343272497e-05,
+      "loss": 0.8043,
+      "step": 1457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7078547051121113e-05,
+      "loss": 0.8886,
+      "step": 1458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074274170323214e-05,
+      "loss": 0.86,
+      "step": 1459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7069998702441396e-05,
+      "loss": 0.7753,
+      "step": 1460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065720649039216e-05,
+      "loss": 0.8356,
+      "step": 1461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7061440011681176e-05,
+      "loss": 0.8385,
+      "step": 1462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057156791932708e-05,
+      "loss": 0.8367,
+      "step": 1463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70528709913602e-05,
+      "loss": 0.8422,
+      "step": 1464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048582611530993e-05,
+      "loss": 0.8076,
+      "step": 1465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7044291654013353e-05,
+      "loss": 0.879,
+      "step": 1466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039998120376498e-05,
+      "loss": 0.8676,
+      "step": 1467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7035702012190585e-05,
+      "loss": 0.8508,
+      "step": 1468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031403331026712e-05,
+      "loss": 0.8059,
+      "step": 1469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702710207845693e-05,
+      "loss": 0.2865,
+      "step": 1470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7022798256054205e-05,
+      "loss": 0.8059,
+      "step": 1471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701849186539247e-05,
+      "loss": 0.8345,
+      "step": 1472
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7014182908046585e-05,
+      "loss": 0.8544,
+      "step": 1473
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700987138559234e-05,
+      "loss": 0.85,
+      "step": 1474
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700555729960648e-05,
+      "loss": 0.8277,
+      "step": 1475
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700124065166668e-05,
+      "loss": 0.8018,
+      "step": 1476
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699692144335155e-05,
+      "loss": 0.8569,
+      "step": 1477
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992599676240633e-05,
+      "loss": 0.7901,
+      "step": 1478
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988275351914417e-05,
+      "loss": 0.8791,
+      "step": 1479
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698394847195432e-05,
+      "loss": 0.7573,
+      "step": 1480
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979619037942696e-05,
+      "loss": 0.8707,
+      "step": 1481
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975287051462833e-05,
+      "loss": 0.8355,
+      "step": 1482
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697095251409895e-05,
+      "loss": 0.8636,
+      "step": 1483
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6966615427436202e-05,
+      "loss": 0.8481,
+      "step": 1484
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962275793060672e-05,
+      "loss": 0.8536,
+      "step": 1485
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957933612559383e-05,
+      "loss": 0.8645,
+      "step": 1486
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695358888752028e-05,
+      "loss": 0.8724,
+      "step": 1487
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6949241619532243e-05,
+      "loss": 0.8781,
+      "step": 1488
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944891810185083e-05,
+      "loss": 0.8521,
+      "step": 1489
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940539461069535e-05,
+      "loss": 0.8982,
+      "step": 1490
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693618457377727e-05,
+      "loss": 0.8127,
+      "step": 1491
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931827149900884e-05,
+      "loss": 0.8518,
+      "step": 1492
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927467191033895e-05,
+      "loss": 0.8405,
+      "step": 1493
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692310469877076e-05,
+      "loss": 0.8735,
+      "step": 1494
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6918739674706853e-05,
+      "loss": 0.9006,
+      "step": 1495
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691437212043848e-05,
+      "loss": 0.7965,
+      "step": 1496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6910002037562863e-05,
+      "loss": 0.7991,
+      "step": 1497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690562942767816e-05,
+      "loss": 0.8272,
+      "step": 1498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901254292383442e-05,
+      "loss": 0.8061,
+      "step": 1499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6896876633278716e-05,
+      "loss": 0.8249,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892496451964902e-05,
+      "loss": 0.8834,
+      "step": 1501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888113750043847e-05,
+      "loss": 0.8337,
+      "step": 1502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688372852911832e-05,
+      "loss": 0.2903,
+      "step": 1503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6879340790792008e-05,
+      "loss": 0.8814,
+      "step": 1504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874950536669518e-05,
+      "loss": 0.8196,
+      "step": 1505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687055776835638e-05,
+      "loss": 0.8689,
+      "step": 1506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6866162487459052e-05,
+      "loss": 0.2961,
+      "step": 1507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6861764695584893e-05,
+      "loss": 0.8896,
+      "step": 1508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685736439434219e-05,
+      "loss": 0.8557,
+      "step": 1509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6852961585340146e-05,
+      "loss": 0.2942,
+      "step": 1510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6848556270188883e-05,
+      "loss": 0.7954,
+      "step": 1511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684414845049944e-05,
+      "loss": 0.8565,
+      "step": 1512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6839738127883772e-05,
+      "loss": 0.8556,
+      "step": 1513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683532530395474e-05,
+      "loss": 0.8352,
+      "step": 1514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6830909980326134e-05,
+      "loss": 0.8496,
+      "step": 1515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682649215861265e-05,
+      "loss": 0.2717,
+      "step": 1516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68220718404299e-05,
+      "loss": 0.883,
+      "step": 1517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6817649027394405e-05,
+      "loss": 0.8571,
+      "step": 1518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813223721123603e-05,
+      "loss": 0.8722,
+      "step": 1519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680879592323584e-05,
+      "loss": 0.8596,
+      "step": 1520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680436563535038e-05,
+      "loss": 0.8347,
+      "step": 1521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679993285908739e-05,
+      "loss": 0.8413,
+      "step": 1522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6795497596067955e-05,
+      "loss": 0.8016,
+      "step": 1523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6791059847914055e-05,
+      "loss": 0.829,
+      "step": 1524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786619616248594e-05,
+      "loss": 0.7936,
+      "step": 1525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782176902695376e-05,
+      "loss": 0.8652,
+      "step": 1526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677773170887912e-05,
+      "loss": 0.8377,
+      "step": 1527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6773284036425443e-05,
+      "loss": 0.8255,
+      "step": 1528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6768833886960872e-05,
+      "loss": 0.7807,
+      "step": 1529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676438126211284e-05,
+      "loss": 0.8352,
+      "step": 1530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675992616350969e-05,
+      "loss": 0.8429,
+      "step": 1531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675546859278066e-05,
+      "loss": 0.8475,
+      "step": 1532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6751008551555906e-05,
+      "loss": 0.8719,
+      "step": 1533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6746546041466466e-05,
+      "loss": 0.783,
+      "step": 1534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742081064144306e-05,
+      "loss": 0.8648,
+      "step": 1535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737613621222274e-05,
+      "loss": 0.802,
+      "step": 1536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6733143714334128e-05,
+      "loss": 0.8299,
+      "step": 1537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6728671345114533e-05,
+      "loss": 0.8599,
+      "step": 1538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672419651519904e-05,
+      "loss": 0.8361,
+      "step": 1539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6719719226224117e-05,
+      "loss": 0.8288,
+      "step": 1540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715239479827114e-05,
+      "loss": 0.8816,
+      "step": 1541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6710757277646296e-05,
+      "loss": 0.8743,
+      "step": 1542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670627262132081e-05,
+      "loss": 0.8472,
+      "step": 1543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701785512490714e-05,
+      "loss": 0.8766,
+      "step": 1544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6697295952796957e-05,
+      "loss": 0.8727,
+      "step": 1545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6692803943881385e-05,
+      "loss": 0.8213,
+      "step": 1546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688309487386738e-05,
+      "loss": 0.8363,
+      "step": 1547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668381258495665e-05,
+      "loss": 0.8218,
+      "step": 1548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667931323823566e-05,
+      "loss": 0.8727,
+      "step": 1549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674811448869185e-05,
+      "loss": 0.8455,
+      "step": 1550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667030721850355e-05,
+      "loss": 0.8139,
+      "step": 1551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666580054878596e-05,
+      "loss": 0.8813,
+      "step": 1552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661291441364523e-05,
+      "loss": 0.8412,
+      "step": 1553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665677989788823e-05,
+      "loss": 0.871,
+      "step": 1554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652265920006968e-05,
+      "loss": 0.8109,
+      "step": 1555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664774950937151e-05,
+      "loss": 0.9158,
+      "step": 1556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643230667633526e-05,
+      "loss": 0.8135,
+      "step": 1557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638709396445567e-05,
+      "loss": 0.7788,
+      "step": 1558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6634185697461077e-05,
+      "loss": 0.8453,
+      "step": 1559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629659572334388e-05,
+      "loss": 0.8218,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625131022720716e-05,
+      "loss": 0.8412,
+      "step": 1561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620600050276164e-05,
+      "loss": 0.8145,
+      "step": 1562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6616066656657724e-05,
+      "loss": 0.8273,
+      "step": 1563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6611530843523277e-05,
+      "loss": 0.7821,
+      "step": 1564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606992612531574e-05,
+      "loss": 0.804,
+      "step": 1565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660245196534227e-05,
+      "loss": 0.8502,
+      "step": 1566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597908903615888e-05,
+      "loss": 0.8883,
+      "step": 1567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6593363429013838e-05,
+      "loss": 0.2677,
+      "step": 1568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658881554319842e-05,
+      "loss": 0.8224,
+      "step": 1569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6584265247832805e-05,
+      "loss": 0.9118,
+      "step": 1570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6579712544581052e-05,
+      "loss": 0.8162,
+      "step": 1571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6575157435108102e-05,
+      "loss": 0.8418,
+      "step": 1572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657059992107977e-05,
+      "loss": 0.8216,
+      "step": 1573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656604000416275e-05,
+      "loss": 0.8349,
+      "step": 1574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6561477686024616e-05,
+      "loss": 0.8262,
+      "step": 1575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6556912968333832e-05,
+      "loss": 0.862,
+      "step": 1576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655234585275972e-05,
+      "loss": 0.7824,
+      "step": 1577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654777634097249e-05,
+      "loss": 0.8947,
+      "step": 1578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6543204434643233e-05,
+      "loss": 0.8268,
+      "step": 1579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6538630135443902e-05,
+      "loss": 0.8348,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534053445047336e-05,
+      "loss": 0.8141,
+      "step": 1581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652947436512724e-05,
+      "loss": 0.818,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524892897358203e-05,
+      "loss": 0.8977,
+      "step": 1583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652030904341568e-05,
+      "loss": 0.8189,
+      "step": 1584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515722804975998e-05,
+      "loss": 0.8652,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511134183716366e-05,
+      "loss": 0.8282,
+      "step": 1586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506543181314846e-05,
+      "loss": 0.8375,
+      "step": 1587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6501949799450393e-05,
+      "loss": 0.8992,
+      "step": 1588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497354039802817e-05,
+      "loss": 0.8463,
+      "step": 1589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64927559040528e-05,
+      "loss": 0.8585,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6488155393881887e-05,
+      "loss": 0.8452,
+      "step": 1591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648355251097251e-05,
+      "loss": 0.8437,
+      "step": 1592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647894725700796e-05,
+      "loss": 0.8847,
+      "step": 1593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647433963367238e-05,
+      "loss": 0.8267,
+      "step": 1594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6469729642650804e-05,
+      "loss": 0.8381,
+      "step": 1595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646511728562911e-05,
+      "loss": 0.8374,
+      "step": 1596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460502564294056e-05,
+      "loss": 0.8434,
+      "step": 1597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6455885480333264e-05,
+      "loss": 0.8218,
+      "step": 1598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645126603543521e-05,
+      "loss": 0.8276,
+      "step": 1599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446644231289237e-05,
+      "loss": 0.8977,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644202006958556e-05,
+      "loss": 0.8141,
+      "step": 1601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6437393552015245e-05,
+      "loss": 0.8404,
+      "step": 1602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432764680270227e-05,
+      "loss": 0.8926,
+      "step": 1603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6428133456043297e-05,
+      "loss": 0.8482,
+      "step": 1604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6423499881028103e-05,
+      "loss": 0.8204,
+      "step": 1605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418863956919167e-05,
+      "loss": 0.8516,
+      "step": 1606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6414225685411852e-05,
+      "loss": 0.8198,
+      "step": 1607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409585068202397e-05,
+      "loss": 0.9057,
+      "step": 1608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6404942106987882e-05,
+      "loss": 0.9098,
+      "step": 1609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6400296803466257e-05,
+      "loss": 0.8554,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395649159336323e-05,
+      "loss": 0.8115,
+      "step": 1611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639099917629773e-05,
+      "loss": 0.3011,
+      "step": 1612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6386346856051007e-05,
+      "loss": 0.8654,
+      "step": 1613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638169220029751e-05,
+      "loss": 0.2875,
+      "step": 1614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6377035210739465e-05,
+      "loss": 0.8655,
+      "step": 1615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372375889079948e-05,
+      "loss": 0.8913,
+      "step": 1616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6367714237022886e-05,
+      "loss": 0.8463,
+      "step": 1617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6363050256273053e-05,
+      "loss": 0.8858,
+      "step": 1618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6358383948536093e-05,
+      "loss": 0.8141,
+      "step": 1619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353715315518485e-05,
+      "loss": 0.843,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349044358927558e-05,
+      "loss": 0.8348,
+      "step": 1621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6344371080471504e-05,
+      "loss": 0.848,
+      "step": 1622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6339695481859346e-05,
+      "loss": 0.8204,
+      "step": 1623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6335017564800968e-05,
+      "loss": 0.845,
+      "step": 1624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6330337331007103e-05,
+      "loss": 0.7965,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6325654782189322e-05,
+      "loss": 0.8524,
+      "step": 1626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632096992006005e-05,
+      "loss": 0.8101,
+      "step": 1627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6316282746332553e-05,
+      "loss": 0.8454,
+      "step": 1628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6311593262720945e-05,
+      "loss": 0.8929,
+      "step": 1629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6306901470940192e-05,
+      "loss": 0.797,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6302207372706085e-05,
+      "loss": 0.8922,
+      "step": 1631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629751096973528e-05,
+      "loss": 0.8199,
+      "step": 1632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629281226374526e-05,
+      "loss": 0.8346,
+      "step": 1633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628811125645436e-05,
+      "loss": 0.2741,
+      "step": 1634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283407949581752e-05,
+      "loss": 0.8211,
+      "step": 1635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6278702344847448e-05,
+      "loss": 0.7933,
+      "step": 1636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273994443972303e-05,
+      "loss": 0.2634,
+      "step": 1637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626928424867801e-05,
+      "loss": 0.911,
+      "step": 1638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6264571760687107e-05,
+      "loss": 0.8941,
+      "step": 1639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259856981722963e-05,
+      "loss": 0.8483,
+      "step": 1640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625513991350978e-05,
+      "loss": 0.8397,
+      "step": 1641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6250420557772613e-05,
+      "loss": 0.8461,
+      "step": 1642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245698916237342e-05,
+      "loss": 0.8406,
+      "step": 1643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6240974990630688e-05,
+      "loss": 0.8666,
+      "step": 1644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236248782680205e-05,
+      "loss": 0.8596,
+      "step": 1645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6231520294114278e-05,
+      "loss": 0.8313,
+      "step": 1646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6226789526662134e-05,
+      "loss": 0.7815,
+      "step": 1647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622205648205382e-05,
+      "loss": 0.8664,
+      "step": 1648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621732116202024e-05,
+      "loss": 0.8565,
+      "step": 1649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6212583568293105e-05,
+      "loss": 0.8616,
+      "step": 1650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620784370260497e-05,
+      "loss": 0.7864,
+      "step": 1651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6203101566689223e-05,
+      "loss": 0.2841,
+      "step": 1652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619835716228007e-05,
+      "loss": 0.8075,
+      "step": 1653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6193610491112557e-05,
+      "loss": 0.8063,
+      "step": 1654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618886155492256e-05,
+      "loss": 0.8429,
+      "step": 1655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6184110355446775e-05,
+      "loss": 0.8637,
+      "step": 1656
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6179356894422732e-05,
+      "loss": 0.8494,
+      "step": 1657
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174601173588786e-05,
+      "loss": 0.8148,
+      "step": 1658
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169843194684123e-05,
+      "loss": 0.8324,
+      "step": 1659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616508295944874e-05,
+      "loss": 0.2788,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6160320469623485e-05,
+      "loss": 0.8344,
+      "step": 1661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615555572695e-05,
+      "loss": 0.8298,
+      "step": 1662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6150788733170776e-05,
+      "loss": 0.8656,
+      "step": 1663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146019490029115e-05,
+      "loss": 0.295,
+      "step": 1664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614124799926914e-05,
+      "loss": 0.266,
+      "step": 1665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613647426263581e-05,
+      "loss": 0.2849,
+      "step": 1666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613169828187488e-05,
+      "loss": 0.8538,
+      "step": 1667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6126920058732957e-05,
+      "loss": 0.8407,
+      "step": 1668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6122139594957442e-05,
+      "loss": 0.7881,
+      "step": 1669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6117356892296564e-05,
+      "loss": 0.8712,
+      "step": 1670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6112571952499377e-05,
+      "loss": 0.7918,
+      "step": 1671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610778477731575e-05,
+      "loss": 0.8035,
+      "step": 1672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6102995368496362e-05,
+      "loss": 0.9064,
+      "step": 1673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6098203727792724e-05,
+      "loss": 0.8029,
+      "step": 1674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609340985695714e-05,
+      "loss": 0.8508,
+      "step": 1675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608861375774275e-05,
+      "loss": 0.7863,
+      "step": 1676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6083815431903507e-05,
+      "loss": 0.2729,
+      "step": 1677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607901488119417e-05,
+      "loss": 0.8171,
+      "step": 1678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6074212107370312e-05,
+      "loss": 0.8795,
+      "step": 1679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069407112188326e-05,
+      "loss": 0.8139,
+      "step": 1680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064599897405413e-05,
+      "loss": 0.8782,
+      "step": 1681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059790464779586e-05,
+      "loss": 0.7993,
+      "step": 1682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605497881606967e-05,
+      "loss": 0.826,
+      "step": 1683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050164953035294e-05,
+      "loss": 0.8514,
+      "step": 1684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6045348877436912e-05,
+      "loss": 0.8192,
+      "step": 1685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604053059103577e-05,
+      "loss": 0.8485,
+      "step": 1686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6035710095593936e-05,
+      "loss": 0.838,
+      "step": 1687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6030887392874278e-05,
+      "loss": 0.7965,
+      "step": 1688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602606248464047e-05,
+      "loss": 0.7844,
+      "step": 1689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021235372657e-05,
+      "loss": 0.7793,
+      "step": 1690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601640605868916e-05,
+      "loss": 0.2981,
+      "step": 1691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6011574544503038e-05,
+      "loss": 0.8068,
+      "step": 1692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6006740831865538e-05,
+      "loss": 0.8622,
+      "step": 1693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6001904922544363e-05,
+      "loss": 0.8303,
+      "step": 1694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599706681830802e-05,
+      "loss": 0.8084,
+      "step": 1695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5992226520925815e-05,
+      "loss": 0.8032,
+      "step": 1696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5987384032167867e-05,
+      "loss": 0.8442,
+      "step": 1697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5982539353805084e-05,
+      "loss": 0.7942,
+      "step": 1698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977692487609183e-05,
+      "loss": 0.8268,
+      "step": 1699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5972843435352673e-05,
+      "loss": 0.7885,
+      "step": 1700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596799219880887e-05,
+      "loss": 0.8738,
+      "step": 1701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596313877975189e-05,
+      "loss": 0.7841,
+      "step": 1702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595828317995664e-05,
+      "loss": 0.8492,
+      "step": 1703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953425401198823e-05,
+      "loss": 0.8674,
+      "step": 1704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594856544525495e-05,
+      "loss": 0.8303,
+      "step": 1705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943703313902322e-05,
+      "loss": 0.8277,
+      "step": 1706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938839008919034e-05,
+      "loss": 0.8442,
+      "step": 1707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593397253208398e-05,
+      "loss": 0.8883,
+      "step": 1708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5929103885176838e-05,
+      "loss": 0.8066,
+      "step": 1709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924233069978094e-05,
+      "loss": 0.8358,
+      "step": 1710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5919360088269015e-05,
+      "loss": 0.7956,
+      "step": 1711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5914484941831673e-05,
+      "loss": 0.8418,
+      "step": 1712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5909607632448917e-05,
+      "loss": 0.8097,
+      "step": 1713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59047281619044e-05,
+      "loss": 0.8547,
+      "step": 1714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589984653198255e-05,
+      "loss": 0.8362,
+      "step": 1715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894962744468606e-05,
+      "loss": 0.8262,
+      "step": 1716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5890076801148576e-05,
+      "loss": 0.7922,
+      "step": 1717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885188703809266e-05,
+      "loss": 0.8182,
+      "step": 1718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880298454238273e-05,
+      "loss": 0.2534,
+      "step": 1719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5875406054223973e-05,
+      "loss": 0.8377,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587051150555553e-05,
+      "loss": 0.838,
+      "step": 1721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5865614810022897e-05,
+      "loss": 0.8454,
+      "step": 1722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586071596941681e-05,
+      "loss": 0.8101,
+      "step": 1723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855814985528794e-05,
+      "loss": 0.8366,
+      "step": 1724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5850911860151147e-05,
+      "loss": 0.8394,
+      "step": 1725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5846006595076966e-05,
+      "loss": 0.2814,
+      "step": 1726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5841099192100114e-05,
+      "loss": 0.8474,
+      "step": 1727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836189653015242e-05,
+      "loss": 0.8732,
+      "step": 1728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5831277979617797e-05,
+      "loss": 0.8108,
+      "step": 1729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5826364173703973e-05,
+      "loss": 0.8231,
+      "step": 1730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582144823707078e-05,
+      "loss": 0.8605,
+      "step": 1731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816530171515988e-05,
+      "loss": 0.8254,
+      "step": 1732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5811609978838146e-05,
+      "loss": 0.8566,
+      "step": 1733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5806687660836583e-05,
+      "loss": 0.8775,
+      "step": 1734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5801763219311406e-05,
+      "loss": 0.8383,
+      "step": 1735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796836656063504e-05,
+      "loss": 0.2941,
+      "step": 1736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791907972894528e-05,
+      "loss": 0.8757,
+      "step": 1737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578697717160692e-05,
+      "loss": 0.2679,
+      "step": 1738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5782044254003886e-05,
+      "loss": 0.8152,
+      "step": 1739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5777109221889408e-05,
+      "loss": 0.8228,
+      "step": 1740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772172077068243e-05,
+      "loss": 0.8955,
+      "step": 1741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5767232821345922e-05,
+      "loss": 0.8809,
+      "step": 1742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762291456528744e-05,
+      "loss": 0.8376,
+      "step": 1743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757347984423786e-05,
+      "loss": 0.8121,
+      "step": 1744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5752402406838883e-05,
+      "loss": 0.7987,
+      "step": 1745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747454725582655e-05,
+      "loss": 0.8435,
+      "step": 1746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574250494246448e-05,
+      "loss": 0.796,
+      "step": 1747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573755305929451e-05,
+      "loss": 0.8509,
+      "step": 1748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573259907788367e-05,
+      "loss": 0.7281,
+      "step": 1749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5727643000043635e-05,
+      "loss": 0.8526,
+      "step": 1750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722684827586867e-05,
+      "loss": 0.8245,
+      "step": 1751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571772456232658e-05,
+      "loss": 0.8247,
+      "step": 1752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571276220607676e-05,
+      "loss": 0.8142,
+      "step": 1753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5707797760652155e-05,
+      "loss": 0.7548,
+      "step": 1754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702831227868285e-05,
+      "loss": 0.8048,
+      "step": 1755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697862609541416e-05,
+      "loss": 0.8303,
+      "step": 1756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5692891907488595e-05,
+      "loss": 0.8024,
+      "step": 1757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5687919123527614e-05,
+      "loss": 0.8269,
+      "step": 1758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682944259477046e-05,
+      "loss": 0.8303,
+      "step": 1759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567796731715621e-05,
+      "loss": 0.8392,
+      "step": 1760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567298829838519e-05,
+      "loss": 0.7577,
+      "step": 1761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566800720498482e-05,
+      "loss": 0.8534,
+      "step": 1762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5663024038776716e-05,
+      "loss": 0.8432,
+      "step": 1763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658038801583225e-05,
+      "loss": 0.8417,
+      "step": 1764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653051495227472e-05,
+      "loss": 0.8378,
+      "step": 1765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5648062121533324e-05,
+      "loss": 0.8431,
+      "step": 1766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564307068232541e-05,
+      "loss": 0.846,
+      "step": 1767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563807717942912e-05,
+      "loss": 0.8002,
+      "step": 1768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5633081614670596e-05,
+      "loss": 0.8937,
+      "step": 1769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562808398987672e-05,
+      "loss": 0.8394,
+      "step": 1770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562308430687515e-05,
+      "loss": 0.8376,
+      "step": 1771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561808256749428e-05,
+      "loss": 0.8455,
+      "step": 1772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5613078773563262e-05,
+      "loss": 0.8103,
+      "step": 1773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608072926912e-05,
+      "loss": 0.8231,
+      "step": 1774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603065029371146e-05,
+      "loss": 0.7923,
+      "step": 1775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5598055082772103e-05,
+      "loss": 0.8416,
+      "step": 1776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593043088947027e-05,
+      "loss": 0.8562,
+      "step": 1777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558802904972882e-05,
+      "loss": 0.8999,
+      "step": 1778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5583012966951128e-05,
+      "loss": 0.8432,
+      "step": 1779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557799484244835e-05,
+      "loss": 0.7811,
+      "step": 1780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557297467805563e-05,
+      "loss": 0.2838,
+      "step": 1781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5567952475608854e-05,
+      "loss": 0.8488,
+      "step": 1782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5562928236944666e-05,
+      "loss": 0.8634,
+      "step": 1783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5557901963900436e-05,
+      "loss": 0.8318,
+      "step": 1784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555287365831429e-05,
+      "loss": 0.835,
+      "step": 1785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5547843322025096e-05,
+      "loss": 0.7893,
+      "step": 1786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542810956872463e-05,
+      "loss": 0.8311,
+      "step": 1787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5537776564696747e-05,
+      "loss": 0.8811,
+      "step": 1788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.553274014733903e-05,
+      "loss": 0.8961,
+      "step": 1789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552770170664116e-05,
+      "loss": 0.8805,
+      "step": 1790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55226612444457e-05,
+      "loss": 0.8336,
+      "step": 1791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5517618762595964e-05,
+      "loss": 0.7943,
+      "step": 1792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5512574262936e-05,
+      "loss": 0.8224,
+      "step": 1793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.550752774731061e-05,
+      "loss": 0.8358,
+      "step": 1794
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550247921756531e-05,
+      "loss": 0.8363,
+      "step": 1795
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5497428675546358e-05,
+      "loss": 0.8532,
+      "step": 1796
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5492376123100764e-05,
+      "loss": 0.8246,
+      "step": 1797
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548732156207626e-05,
+      "loss": 0.8056,
+      "step": 1798
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5482264994321308e-05,
+      "loss": 0.833,
+      "step": 1799
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5477206421685115e-05,
+      "loss": 0.8231,
+      "step": 1800
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472145846017616e-05,
+      "loss": 0.8156,
+      "step": 1801
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546708326916948e-05,
+      "loss": 0.8202,
+      "step": 1802
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546201869299211e-05,
+      "loss": 0.8814,
+      "step": 1803
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5456952119337628e-05,
+      "loss": 0.8178,
+      "step": 1804
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5451883550058904e-05,
+      "loss": 0.8748,
+      "step": 1805
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544681298700952e-05,
+      "loss": 0.883,
+      "step": 1806
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5441740432043816e-05,
+      "loss": 0.8812,
+      "step": 1807
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5436665887016817e-05,
+      "loss": 0.8515,
+      "step": 1808
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5431589353784317e-05,
+      "loss": 0.784,
+      "step": 1809
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426510834202807e-05,
+      "loss": 0.8251,
+      "step": 1810
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5421430330129532e-05,
+      "loss": 0.8589,
+      "step": 1811
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5416347843422437e-05,
+      "loss": 0.8552,
+      "step": 1812
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5411263375940207e-05,
+      "loss": 0.8454,
+      "step": 1813
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5406176929542246e-05,
+      "loss": 0.8216,
+      "step": 1814
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5401088506088682e-05,
+      "loss": 0.8022,
+      "step": 1815
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539599810744037e-05,
+      "loss": 0.8192,
+      "step": 1816
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5390905735458886e-05,
+      "loss": 0.8469,
+      "step": 1817
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538581139200652e-05,
+      "loss": 0.8466,
+      "step": 1818
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5380715078946294e-05,
+      "loss": 0.8795,
+      "step": 1819
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375616798141947e-05,
+      "loss": 0.8044,
+      "step": 1820
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537051655145793e-05,
+      "loss": 0.8492,
+      "step": 1821
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365414340759422e-05,
+      "loss": 0.8103,
+      "step": 1822
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536031016791232e-05,
+      "loss": 0.8461,
+      "step": 1823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5355204034783234e-05,
+      "loss": 0.807,
+      "step": 1824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535009594323949e-05,
+      "loss": 0.8189,
+      "step": 1825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534498589514913e-05,
+      "loss": 0.8303,
+      "step": 1826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5339873892380924e-05,
+      "loss": 0.8672,
+      "step": 1827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533475993680434e-05,
+      "loss": 0.8486,
+      "step": 1828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5329644030289573e-05,
+      "loss": 0.796,
+      "step": 1829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532452617470752e-05,
+      "loss": 0.838,
+      "step": 1830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5319406371929803e-05,
+      "loss": 0.8494,
+      "step": 1831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5314284623828744e-05,
+      "loss": 0.8192,
+      "step": 1832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530916093227738e-05,
+      "loss": 0.809,
+      "step": 1833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530403529914947e-05,
+      "loss": 0.8092,
+      "step": 1834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5298907726319466e-05,
+      "loss": 0.7787,
+      "step": 1835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529377821566254e-05,
+      "loss": 0.9023,
+      "step": 1836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5288646769054566e-05,
+      "loss": 0.7973,
+      "step": 1837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5283513388372137e-05,
+      "loss": 0.7599,
+      "step": 1838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278378075492542e-05,
+      "loss": 0.8614,
+      "step": 1839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273240832293785e-05,
+      "loss": 0.844,
+      "step": 1840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5268101660654565e-05,
+      "loss": 0.7967,
+      "step": 1841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5262960562454295e-05,
+      "loss": 0.8117,
+      "step": 1842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257817539573094e-05,
+      "loss": 0.8234,
+      "step": 1843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252672593891778e-05,
+      "loss": 0.7997,
+      "step": 1844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.524752572729187e-05,
+      "loss": 0.7763,
+      "step": 1845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5242376941655597e-05,
+      "loss": 0.7995,
+      "step": 1846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5237226238865881e-05,
+      "loss": 0.8025,
+      "step": 1847
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5232073620806357e-05,
+      "loss": 0.2818,
+      "step": 1848
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522691908936135e-05,
+      "loss": 0.8226,
+      "step": 1849
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5221762646415885e-05,
+      "loss": 0.7819,
+      "step": 1850
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5216604293855693e-05,
+      "loss": 0.9147,
+      "step": 1851
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211444033567198e-05,
+      "loss": 0.7541,
+      "step": 1852
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206281867437522e-05,
+      "loss": 0.8652,
+      "step": 1853
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201117797354488e-05,
+      "loss": 0.2561,
+      "step": 1854
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519595182520661e-05,
+      "loss": 0.8514,
+      "step": 1855
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5190783952883101e-05,
+      "loss": 0.8466,
+      "step": 1856
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518561418227387e-05,
+      "loss": 0.8715,
+      "step": 1857
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180442515269513e-05,
+      "loss": 0.8643,
+      "step": 1858
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5175268953761326e-05,
+      "loss": 0.8521,
+      "step": 1859
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170093499641301e-05,
+      "loss": 0.8702,
+      "step": 1860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164916154802112e-05,
+      "loss": 0.8086,
+      "step": 1861
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159736921137134e-05,
+      "loss": 0.8103,
+      "step": 1862
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515455580054043e-05,
+      "loss": 0.8295,
+      "step": 1863
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5149372794906742e-05,
+      "loss": 0.8409,
+      "step": 1864
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5144187906131523e-05,
+      "loss": 0.8098,
+      "step": 1865
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5139001136110896e-05,
+      "loss": 0.8274,
+      "step": 1866
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133812486741681e-05,
+      "loss": 0.8662,
+      "step": 1867
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128621959921384e-05,
+      "loss": 0.2937,
+      "step": 1868
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123429557548193e-05,
+      "loss": 0.8446,
+      "step": 1869
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5118235281520991e-05,
+      "loss": 0.8639,
+      "step": 1870
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5113039133739334e-05,
+      "loss": 0.2565,
+      "step": 1871
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107841116103477e-05,
+      "loss": 0.787,
+      "step": 1872
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102641230514344e-05,
+      "loss": 0.8549,
+      "step": 1873
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5097439478873557e-05,
+      "loss": 0.8685,
+      "step": 1874
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5092235863083407e-05,
+      "loss": 0.7976,
+      "step": 1875
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508703038504687e-05,
+      "loss": 0.8665,
+      "step": 1876
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5081823046667612e-05,
+      "loss": 0.8622,
+      "step": 1877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076613849849967e-05,
+      "loss": 0.8512,
+      "step": 1878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507140279649896e-05,
+      "loss": 0.8372,
+      "step": 1879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066189888520284e-05,
+      "loss": 0.2883,
+      "step": 1880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5060975127820316e-05,
+      "loss": 0.7438,
+      "step": 1881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055758516306118e-05,
+      "loss": 0.8518,
+      "step": 1882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050540055885411e-05,
+      "loss": 0.8462,
+      "step": 1883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045319748466607e-05,
+      "loss": 0.8096,
+      "step": 1884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040097595958785e-05,
+      "loss": 0.851,
+      "step": 1885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034873600271706e-05,
+      "loss": 0.8432,
+      "step": 1886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029647763315805e-05,
+      "loss": 0.7847,
+      "step": 1887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024420087002174e-05,
+      "loss": 0.8483,
+      "step": 1888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019190573242604e-05,
+      "loss": 0.8284,
+      "step": 1889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013959223949535e-05,
+      "loss": 0.8182,
+      "step": 1890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008726041036092e-05,
+      "loss": 0.8317,
+      "step": 1891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003491026416071e-05,
+      "loss": 0.8575,
+      "step": 1892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998254182003922e-05,
+      "loss": 0.88,
+      "step": 1893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993015509714781e-05,
+      "loss": 0.8415,
+      "step": 1894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987775011464444e-05,
+      "loss": 0.7786,
+      "step": 1895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4982532689169382e-05,
+      "loss": 0.2638,
+      "step": 1896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977288544746728e-05,
+      "loss": 0.8141,
+      "step": 1897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4972042580114276e-05,
+      "loss": 0.8019,
+      "step": 1898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966794797190499e-05,
+      "loss": 0.7855,
+      "step": 1899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961545197894522e-05,
+      "loss": 0.859,
+      "step": 1900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.495629378414614e-05,
+      "loss": 0.8563,
+      "step": 1901
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951040557865815e-05,
+      "loss": 0.8407,
+      "step": 1902
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945785520974666e-05,
+      "loss": 0.8103,
+      "step": 1903
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940528675394474e-05,
+      "loss": 0.8302,
+      "step": 1904
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935270023047686e-05,
+      "loss": 0.8371,
+      "step": 1905
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930009565857408e-05,
+      "loss": 0.898,
+      "step": 1906
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4924747305747403e-05,
+      "loss": 0.2776,
+      "step": 1907
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4919483244642097e-05,
+      "loss": 0.7827,
+      "step": 1908
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4914217384466576e-05,
+      "loss": 0.7968,
+      "step": 1909
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908949727146578e-05,
+      "loss": 0.829,
+      "step": 1910
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4903680274608503e-05,
+      "loss": 0.8217,
+      "step": 1911
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4898409028779404e-05,
+      "loss": 0.8592,
+      "step": 1912
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4893135991586993e-05,
+      "loss": 0.8649,
+      "step": 1913
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887861164959641e-05,
+      "loss": 0.8413,
+      "step": 1914
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882584550826357e-05,
+      "loss": 0.8523,
+      "step": 1915
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877306151116826e-05,
+      "loss": 0.7792,
+      "step": 1916
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872025967761372e-05,
+      "loss": 0.7554,
+      "step": 1917
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4866744002690974e-05,
+      "loss": 0.8521,
+      "step": 1918
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861460257837257e-05,
+      "loss": 0.8137,
+      "step": 1919
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856174735132513e-05,
+      "loss": 0.8302,
+      "step": 1920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485088743650967e-05,
+      "loss": 0.8607,
+      "step": 1921
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845598363902303e-05,
+      "loss": 0.7538,
+      "step": 1922
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840307519244653e-05,
+      "loss": 0.8068,
+      "step": 1923
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4835014904471587e-05,
+      "loss": 0.7821,
+      "step": 1924
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4829720521518642e-05,
+      "loss": 0.8838,
+      "step": 1925
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4824424372321984e-05,
+      "loss": 0.7819,
+      "step": 1926
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4819126458818431e-05,
+      "loss": 0.8524,
+      "step": 1927
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813826782945448e-05,
+      "loss": 0.8213,
+      "step": 1928
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4808525346641141e-05,
+      "loss": 0.7676,
+      "step": 1929
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480322215184426e-05,
+      "loss": 0.8205,
+      "step": 1930
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4797917200494207e-05,
+      "loss": 0.8357,
+      "step": 1931
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4792610494531012e-05,
+      "loss": 0.7781,
+      "step": 1932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787302035895356e-05,
+      "loss": 0.8183,
+      "step": 1933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478199182652856e-05,
+      "loss": 0.873,
+      "step": 1934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4776679868372585e-05,
+      "loss": 0.2778,
+      "step": 1935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771366163370027e-05,
+      "loss": 0.8303,
+      "step": 1936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4766050713464124e-05,
+      "loss": 0.8012,
+      "step": 1937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4760733520598756e-05,
+      "loss": 0.2728,
+      "step": 1938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755414586718433e-05,
+      "loss": 0.9067,
+      "step": 1939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750093913768309e-05,
+      "loss": 0.8249,
+      "step": 1940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744771503694168e-05,
+      "loss": 0.8532,
+      "step": 1941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4739447358442434e-05,
+      "loss": 0.8263,
+      "step": 1942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734121479960162e-05,
+      "loss": 0.8059,
+      "step": 1943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4728793870195044e-05,
+      "loss": 0.847,
+      "step": 1944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723464531095401e-05,
+      "loss": 0.8644,
+      "step": 1945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471813346461019e-05,
+      "loss": 0.881,
+      "step": 1946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712800672688997e-05,
+      "loss": 0.8984,
+      "step": 1947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470746615728204e-05,
+      "loss": 0.8173,
+      "step": 1948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4702129920340171e-05,
+      "loss": 0.8204,
+      "step": 1949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696791963814869e-05,
+      "loss": 0.81,
+      "step": 1950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4691452289658234e-05,
+      "loss": 0.8056,
+      "step": 1951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686110899823009e-05,
+      "loss": 0.8599,
+      "step": 1952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680767796262556e-05,
+      "loss": 0.7693,
+      "step": 1953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4675422980930862e-05,
+      "loss": 0.822,
+      "step": 1954
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.467007645578254e-05,
+      "loss": 0.8307,
+      "step": 1955
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4664728222772835e-05,
+      "loss": 0.8369,
+      "step": 1956
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659378283857612e-05,
+      "loss": 0.2533,
+      "step": 1957
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4654026640993357e-05,
+      "loss": 0.8345,
+      "step": 1958
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648673296137186e-05,
+      "loss": 0.8259,
+      "step": 1959
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643318251246832e-05,
+      "loss": 0.8089,
+      "step": 1960
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463796150828065e-05,
+      "loss": 0.8476,
+      "step": 1961
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4632603069197624e-05,
+      "loss": 0.8769,
+      "step": 1962
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627242935957342e-05,
+      "loss": 0.83,
+      "step": 1963
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4621881110520028e-05,
+      "loss": 0.2805,
+      "step": 1964
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4616517594846515e-05,
+      "loss": 0.7848,
+      "step": 1965
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611152390898262e-05,
+      "loss": 0.2775,
+      "step": 1966
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4605785500637335e-05,
+      "loss": 0.814,
+      "step": 1967
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600416926026424e-05,
+      "loss": 0.8575,
+      "step": 1968
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595046669028834e-05,
+      "loss": 0.8171,
+      "step": 1969
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589674731608485e-05,
+      "loss": 0.7915,
+      "step": 1970
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584301115729911e-05,
+      "loss": 0.2803,
+      "step": 1971
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4578925823358258e-05,
+      "loss": 0.8093,
+      "step": 1972
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457354885645929e-05,
+      "loss": 0.8428,
+      "step": 1973
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4568170216999377e-05,
+      "loss": 0.8368,
+      "step": 1974
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562789906945507e-05,
+      "loss": 0.8469,
+      "step": 1975
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4557407928265277e-05,
+      "loss": 0.8242,
+      "step": 1976
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552024282926888e-05,
+      "loss": 0.2721,
+      "step": 1977
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546638972899161e-05,
+      "loss": 0.8101,
+      "step": 1978
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4541252000151518e-05,
+      "loss": 0.9119,
+      "step": 1979
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4535863366653991e-05,
+      "loss": 0.8055,
+      "step": 1980
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530473074377223e-05,
+      "loss": 0.8292,
+      "step": 1981
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4525081125292455e-05,
+      "loss": 0.8072,
+      "step": 1982
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519687521371544e-05,
+      "loss": 0.7848,
+      "step": 1983
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514292264586949e-05,
+      "loss": 0.834,
+      "step": 1984
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4508895356911726e-05,
+      "loss": 0.8487,
+      "step": 1985
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503496800319548e-05,
+      "loss": 0.8065,
+      "step": 1986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449809659678468e-05,
+      "loss": 0.817,
+      "step": 1987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449269474828199e-05,
+      "loss": 0.8338,
+      "step": 1988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487291256786957e-05,
+      "loss": 0.9537,
+      "step": 1989
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481886124275649e-05,
+      "loss": 0.3008,
+      "step": 1990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4476479352724743e-05,
+      "loss": 0.8215,
+      "step": 1991
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471070944111511e-05,
+      "loss": 0.2657,
+      "step": 1992
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4465660900413828e-05,
+      "loss": 0.81,
+      "step": 1993
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460249223610157e-05,
+      "loss": 0.8108,
+      "step": 1994
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4454835915679569e-05,
+      "loss": 0.8744,
+      "step": 1995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4449420978601732e-05,
+      "loss": 0.8197,
+      "step": 1996
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4444004414356893e-05,
+      "loss": 0.836,
+      "step": 1997
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438586224925917e-05,
+      "loss": 0.8395,
+      "step": 1998
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433166412290248e-05,
+      "loss": 0.8786,
+      "step": 1999
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442774497843193e-05,
+      "loss": 0.2922,
+      "step": 2000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422321925333598e-05,
+      "loss": 0.7889,
+      "step": 2001
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4416897254978477e-05,
+      "loss": 0.263,
+      "step": 2002
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411470969350389e-05,
+      "loss": 0.8067,
+      "step": 2003
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4406043070433742e-05,
+      "loss": 0.8139,
+      "step": 2004
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4400613560213535e-05,
+      "loss": 0.8444,
+      "step": 2005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4395182440675357e-05,
+      "loss": 0.8113,
+      "step": 2006
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4389749713805383e-05,
+      "loss": 0.8089,
+      "step": 2007
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4384315381590384e-05,
+      "loss": 0.8277,
+      "step": 2008
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378879446017702e-05,
+      "loss": 0.7815,
+      "step": 2009
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4373441909075286e-05,
+      "loss": 0.767,
+      "step": 2010
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368002772751655e-05,
+      "loss": 0.8557,
+      "step": 2011
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4362562039035916e-05,
+      "loss": 0.773,
+      "step": 2012
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357119709917768e-05,
+      "loss": 0.8189,
+      "step": 2013
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4351675787387484e-05,
+      "loss": 0.8226,
+      "step": 2014
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346230273435925e-05,
+      "loss": 0.8306,
+      "step": 2015
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340783170054526e-05,
+      "loss": 0.8702,
+      "step": 2016
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4335334479235318e-05,
+      "loss": 0.8655,
+      "step": 2017
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4329884202970901e-05,
+      "loss": 0.8615,
+      "step": 2018
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324432343254456e-05,
+      "loss": 0.8157,
+      "step": 2019
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4318978902079746e-05,
+      "loss": 0.8755,
+      "step": 2020
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4313523881441112e-05,
+      "loss": 0.7821,
+      "step": 2021
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430806728333347e-05,
+      "loss": 0.8099,
+      "step": 2022
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430260910975232e-05,
+      "loss": 0.8863,
+      "step": 2023
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4297149362693723e-05,
+      "loss": 0.8156,
+      "step": 2024
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4291688044154335e-05,
+      "loss": 0.8649,
+      "step": 2025
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286225156131373e-05,
+      "loss": 0.8071,
+      "step": 2026
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280760700622631e-05,
+      "loss": 0.8357,
+      "step": 2027
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275294679626481e-05,
+      "loss": 0.8797,
+      "step": 2028
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269827095141862e-05,
+      "loss": 0.7764,
+      "step": 2029
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4264357949168286e-05,
+      "loss": 0.8472,
+      "step": 2030
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258887243705837e-05,
+      "loss": 0.7689,
+      "step": 2031
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4253414980755167e-05,
+      "loss": 0.8575,
+      "step": 2032
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42479411623175e-05,
+      "loss": 0.8206,
+      "step": 2033
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4242465790394631e-05,
+      "loss": 0.8546,
+      "step": 2034
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423698886698892e-05,
+      "loss": 0.803,
+      "step": 2035
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423151039410329e-05,
+      "loss": 0.8206,
+      "step": 2036
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4226030373741243e-05,
+      "loss": 0.8489,
+      "step": 2037
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422054880790683e-05,
+      "loss": 0.82,
+      "step": 2038
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215065698604689e-05,
+      "loss": 0.826,
+      "step": 2039
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209581047839999e-05,
+      "loss": 0.798,
+      "step": 2040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4204094857618518e-05,
+      "loss": 0.8084,
+      "step": 2041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198607129946563e-05,
+      "loss": 0.7523,
+      "step": 2042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419311786683101e-05,
+      "loss": 0.7804,
+      "step": 2043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4187627070279306e-05,
+      "loss": 0.8192,
+      "step": 2044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4182134742299445e-05,
+      "loss": 0.8438,
+      "step": 2045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176640884899992e-05,
+      "loss": 0.8684,
+      "step": 2046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171145500090067e-05,
+      "loss": 0.837,
+      "step": 2047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416564858987935e-05,
+      "loss": 0.837,
+      "step": 2048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160150156278079e-05,
+      "loss": 0.8332,
+      "step": 2049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154650201297043e-05,
+      "loss": 0.8391,
+      "step": 2050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4149148726947597e-05,
+      "loss": 0.8681,
+      "step": 2051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4143645735241647e-05,
+      "loss": 0.3115,
+      "step": 2052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4138141228191652e-05,
+      "loss": 0.7984,
+      "step": 2053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4132635207810628e-05,
+      "loss": 0.8366,
+      "step": 2054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4127127676112142e-05,
+      "loss": 0.2844,
+      "step": 2055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121618635110323e-05,
+      "loss": 0.8066,
+      "step": 2056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116108086819837e-05,
+      "loss": 0.8308,
+      "step": 2057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110596033255909e-05,
+      "loss": 0.8317,
+      "step": 2058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4105082476434316e-05,
+      "loss": 0.822,
+      "step": 2059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4099567418371378e-05,
+      "loss": 0.8265,
+      "step": 2060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4094050861083978e-05,
+      "loss": 0.8078,
+      "step": 2061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.408853280658953e-05,
+      "loss": 0.7683,
+      "step": 2062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083013256906009e-05,
+      "loss": 0.8335,
+      "step": 2063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077492214051926e-05,
+      "loss": 0.7856,
+      "step": 2064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4071969680046348e-05,
+      "loss": 0.8058,
+      "step": 2065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4066445656908882e-05,
+      "loss": 0.8564,
+      "step": 2066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4060920146659676e-05,
+      "loss": 0.8101,
+      "step": 2067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4055393151319433e-05,
+      "loss": 0.8037,
+      "step": 2068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049864672909386e-05,
+      "loss": 0.8473,
+      "step": 2069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404433471345132e-05,
+      "loss": 0.8156,
+      "step": 2070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4038803274967556e-05,
+      "loss": 0.8474,
+      "step": 2071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033270359480962e-05,
+      "loss": 0.8123,
+      "step": 2072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402773596901494e-05,
+      "loss": 0.8145,
+      "step": 2073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402220010559343e-05,
+      "loss": 0.8256,
+      "step": 2074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016662771240924e-05,
+      "loss": 0.8145,
+      "step": 2075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011123967982433e-05,
+      "loss": 0.7964,
+      "step": 2076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005583697843517e-05,
+      "loss": 0.8044,
+      "step": 2077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4000041962850273e-05,
+      "loss": 0.8644,
+      "step": 2078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3994498765029325e-05,
+      "loss": 0.7953,
+      "step": 2079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3988954106407843e-05,
+      "loss": 0.7987,
+      "step": 2080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983407989013524e-05,
+      "loss": 0.808,
+      "step": 2081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39778604148746e-05,
+      "loss": 0.804,
+      "step": 2082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3972311386019836e-05,
+      "loss": 0.8523,
+      "step": 2083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3966760904478527e-05,
+      "loss": 0.8217,
+      "step": 2084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961208972280504e-05,
+      "loss": 0.8133,
+      "step": 2085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955655591456125e-05,
+      "loss": 0.816,
+      "step": 2086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3950100764036279e-05,
+      "loss": 0.7894,
+      "step": 2087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3944544492052381e-05,
+      "loss": 0.8344,
+      "step": 2088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3938986777536379e-05,
+      "loss": 0.8438,
+      "step": 2089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3933427622520747e-05,
+      "loss": 0.8181,
+      "step": 2090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927867029038487e-05,
+      "loss": 0.8481,
+      "step": 2091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392230499912312e-05,
+      "loss": 0.8632,
+      "step": 2092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39167415348087e-05,
+      "loss": 0.8768,
+      "step": 2093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3911176638129802e-05,
+      "loss": 0.861,
+      "step": 2094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905610311121529e-05,
+      "loss": 0.8303,
+      "step": 2095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39000425558195e-05,
+      "loss": 0.7572,
+      "step": 2096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894473374259865e-05,
+      "loss": 0.8091,
+      "step": 2097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888902768479287e-05,
+      "loss": 0.7778,
+      "step": 2098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3883330740514951e-05,
+      "loss": 0.782,
+      "step": 2099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3877757292404576e-05,
+      "loss": 0.8358,
+      "step": 2100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872182426186376e-05,
+      "loss": 0.8009,
+      "step": 2101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3866606143899105e-05,
+      "loss": 0.8148,
+      "step": 2102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861028447582028e-05,
+      "loss": 0.7836,
+      "step": 2103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3855449339274915e-05,
+      "loss": 0.8623,
+      "step": 2104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3849868821018075e-05,
+      "loss": 0.848,
+      "step": 2105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844286894852315e-05,
+      "loss": 0.8814,
+      "step": 2106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838703562818963e-05,
+      "loss": 0.7829,
+      "step": 2107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3833118826959861e-05,
+      "loss": 0.893,
+      "step": 2108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3827532689317366e-05,
+      "loss": 0.8299,
+      "step": 2109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3821945151934347e-05,
+      "loss": 0.2856,
+      "step": 2110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816356216854177e-05,
+      "loss": 0.8165,
+      "step": 2111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810765886120754e-05,
+      "loss": 0.7696,
+      "step": 2112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3805174161778472e-05,
+      "loss": 0.284,
+      "step": 2113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799581045872247e-05,
+      "loss": 0.8328,
+      "step": 2114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793986540447502e-05,
+      "loss": 0.7801,
+      "step": 2115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3788390647550157e-05,
+      "loss": 0.7898,
+      "step": 2116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378279336922665e-05,
+      "loss": 0.8404,
+      "step": 2117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3777194707523927e-05,
+      "loss": 0.8053,
+      "step": 2118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771594664489433e-05,
+      "loss": 0.791,
+      "step": 2119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.376599324217112e-05,
+      "loss": 0.8298,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3760390442617447e-05,
+      "loss": 0.8566,
+      "step": 2121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375478626787737e-05,
+      "loss": 0.2896,
+      "step": 2122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374918072000036e-05,
+      "loss": 0.8256,
+      "step": 2123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3743573801036383e-05,
+      "loss": 0.8393,
+      "step": 2124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3737965513035905e-05,
+      "loss": 0.7927,
+      "step": 2125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732355858049891e-05,
+      "loss": 0.8116,
+      "step": 2126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372674483812981e-05,
+      "loss": 0.7276,
+      "step": 2127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3721132455327632e-05,
+      "loss": 0.8333,
+      "step": 2128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3715518711695824e-05,
+      "loss": 0.8406,
+      "step": 2129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709903609287342e-05,
+      "loss": 0.8385,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704287150155655e-05,
+      "loss": 0.8073,
+      "step": 2131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3698669336354717e-05,
+      "loss": 0.2433,
+      "step": 2132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3693050169938974e-05,
+      "loss": 0.8578,
+      "step": 2133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3687429652963382e-05,
+      "loss": 0.8291,
+      "step": 2134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3681807787483376e-05,
+      "loss": 0.8154,
+      "step": 2135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676184575554889e-05,
+      "loss": 0.852,
+      "step": 2136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670560019234346e-05,
+      "loss": 0.7979,
+      "step": 2137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366493412057867e-05,
+      "loss": 0.861,
+      "step": 2138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659306881645264e-05,
+      "loss": 0.2614,
+      "step": 2139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3653678304492028e-05,
+      "loss": 0.7947,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3648048391177355e-05,
+      "loss": 0.8172,
+      "step": 2141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642417143760116e-05,
+      "loss": 0.2612,
+      "step": 2142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3636784564299676e-05,
+      "loss": 0.8396,
+      "step": 2143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3631150654855889e-05,
+      "loss": 0.7845,
+      "step": 2144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362551541748909e-05,
+      "loss": 0.7454,
+      "step": 2145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3619878854260106e-05,
+      "loss": 0.888,
+      "step": 2146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614240967230243e-05,
+      "loss": 0.79,
+      "step": 2147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36086017584613e-05,
+      "loss": 0.8626,
+      "step": 2148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602961230015544e-05,
+      "loss": 0.8337,
+      "step": 2149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3597319383955737e-05,
+      "loss": 0.8359,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3591676222345122e-05,
+      "loss": 0.8305,
+      "step": 2151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358603174724742e-05,
+      "loss": 0.8361,
+      "step": 2152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580385960726829e-05,
+      "loss": 0.7841,
+      "step": 2153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3574738864848033e-05,
+      "loss": 0.8631,
+      "step": 2154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569090461676195e-05,
+      "loss": 0.8706,
+      "step": 2155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563440753276953e-05,
+      "loss": 0.8291,
+      "step": 2156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3557789741716415e-05,
+      "loss": 0.7693,
+      "step": 2157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3552137429061185e-05,
+      "loss": 0.8099,
+      "step": 2158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546483817378324e-05,
+      "loss": 0.8194,
+      "step": 2159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3540828908735376e-05,
+      "loss": 0.8364,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3535172705200366e-05,
+      "loss": 0.8269,
+      "step": 2161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529515208841771e-05,
+      "loss": 0.7696,
+      "step": 2162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3523856421728566e-05,
+      "loss": 0.8616,
+      "step": 2163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3518196345930181e-05,
+      "loss": 0.8397,
+      "step": 2164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512534983516528e-05,
+      "loss": 0.8602,
+      "step": 2165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506872336557982e-05,
+      "loss": 0.8599,
+      "step": 2166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3501208407125387e-05,
+      "loss": 0.8232,
+      "step": 2167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3495543197290068e-05,
+      "loss": 0.833,
+      "step": 2168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3489876709123804e-05,
+      "loss": 0.8013,
+      "step": 2169
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3484208944698849e-05,
+      "loss": 0.8568,
+      "step": 2170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478539906087918e-05,
+      "loss": 0.8218,
+      "step": 2171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3472869595364201e-05,
+      "loss": 0.7878,
+      "step": 2172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467198014601349e-05,
+      "loss": 0.7822,
+      "step": 2173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461525165873471e-05,
+      "loss": 0.8224,
+      "step": 2174
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3455851051255148e-05,
+      "loss": 0.8501,
+      "step": 2175
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450175672821421e-05,
+      "loss": 0.2899,
+      "step": 2176
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3444499032647796e-05,
+      "loss": 0.7671,
+      "step": 2177
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438821132810236e-05,
+      "loss": 0.8311,
+      "step": 2178
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3433141975385167e-05,
+      "loss": 0.775,
+      "step": 2179
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342746156244947e-05,
+      "loss": 0.8307,
+      "step": 2180
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421779896080495e-05,
+      "loss": 0.8334,
+      "step": 2181
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3416096978356044e-05,
+      "loss": 0.8768,
+      "step": 2182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3410412811354377e-05,
+      "loss": 0.8361,
+      "step": 2183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3404727397154209e-05,
+      "loss": 0.8007,
+      "step": 2184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3399040737834716e-05,
+      "loss": 0.8862,
+      "step": 2185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393352835475528e-05,
+      "loss": 0.8339,
+      "step": 2186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338766369215673e-05,
+      "loss": 0.8484,
+      "step": 2187
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381973309958852e-05,
+      "loss": 0.7915,
+      "step": 2188
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3376281690962889e-05,
+      "loss": 0.2736,
+      "step": 2189
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370588837250282e-05,
+      "loss": 0.8409,
+      "step": 2190
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3364894750902927e-05,
+      "loss": 0.8561,
+      "step": 2191
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3359199434003171e-05,
+      "loss": 0.759,
+      "step": 2192
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3353502888633806e-05,
+      "loss": 0.8307,
+      "step": 2193
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347805116878077e-05,
+      "loss": 0.8501,
+      "step": 2194
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3342106120819674e-05,
+      "loss": 0.8354,
+      "step": 2195
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3336405902542742e-05,
+      "loss": 0.8354,
+      "step": 2196
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3330704464131868e-05,
+      "loss": 0.2474,
+      "step": 2197
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3325001807672078e-05,
+      "loss": 0.9041,
+      "step": 2198
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331929793524886e-05,
+      "loss": 0.8321,
+      "step": 2199
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3313592848948137e-05,
+      "loss": 0.855,
+      "step": 2200
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330788655085627e-05,
+      "loss": 0.8264,
+      "step": 2201
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3302179043060077e-05,
+      "loss": 0.87,
+      "step": 2202
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296470327646811e-05,
+      "loss": 0.7966,
+      "step": 2203
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3290760406704161e-05,
+      "loss": 0.8043,
+      "step": 2204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285049282320265e-05,
+      "loss": 0.7937,
+      "step": 2205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279336956583703e-05,
+      "loss": 0.8212,
+      "step": 2206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273623431583489e-05,
+      "loss": 0.8553,
+      "step": 2207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326790870940907e-05,
+      "loss": 0.8149,
+      "step": 2208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3262192792150343e-05,
+      "loss": 0.8191,
+      "step": 2209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256475681897636e-05,
+      "loss": 0.8605,
+      "step": 2210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3250757380741717e-05,
+      "loss": 0.8166,
+      "step": 2211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3245037890773782e-05,
+      "loss": 0.7902,
+      "step": 2212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3239317214085467e-05,
+      "loss": 0.8314,
+      "step": 2213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233595352768842e-05,
+      "loss": 0.8145,
+      "step": 2214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227872308916405e-05,
+      "loss": 0.8621,
+      "step": 2215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32221480846211e-05,
+      "loss": 0.8639,
+      "step": 2216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3216422681976285e-05,
+      "loss": 0.8202,
+      "step": 2217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3210696103075762e-05,
+      "loss": 0.7936,
+      "step": 2218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3204968350013759e-05,
+      "loss": 0.8501,
+      "step": 2219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3199239424884927e-05,
+      "loss": 0.7906,
+      "step": 2220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3193509329784358e-05,
+      "loss": 0.8356,
+      "step": 2221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3187778066807565e-05,
+      "loss": 0.3033,
+      "step": 2222
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3182045638050485e-05,
+      "loss": 0.8245,
+      "step": 2223
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3176312045609488e-05,
+      "loss": 0.8851,
+      "step": 2224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170577291581364e-05,
+      "loss": 0.8458,
+      "step": 2225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316484137806333e-05,
+      "loss": 0.8242,
+      "step": 2226
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.315910430715303e-05,
+      "loss": 0.7573,
+      "step": 2227
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3153366080948529e-05,
+      "loss": 0.806,
+      "step": 2228
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3147626701548314e-05,
+      "loss": 0.8369,
+      "step": 2229
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141886171051287e-05,
+      "loss": 0.776,
+      "step": 2230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136144491556789e-05,
+      "loss": 0.8017,
+      "step": 2231
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3130401665164558e-05,
+      "loss": 0.8804,
+      "step": 2232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3124657693974772e-05,
+      "loss": 0.7894,
+      "step": 2233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118912580088012e-05,
+      "loss": 0.8577,
+      "step": 2234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113166325605295e-05,
+      "loss": 0.8231,
+      "step": 2235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3107418932628034e-05,
+      "loss": 0.8387,
+      "step": 2236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3101670403258071e-05,
+      "loss": 0.853,
+      "step": 2237
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309592073959766e-05,
+      "loss": 0.8347,
+      "step": 2238
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 0.2984,
+      "step": 2239
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084418017816598e-05,
+      "loss": 0.725,
+      "step": 2240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078664963902524e-05,
+      "loss": 0.8427,
+      "step": 2241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3072910784111161e-05,
+      "loss": 0.837,
+      "step": 2242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3067155480546834e-05,
+      "loss": 0.8278,
+      "step": 2243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3061399055314274e-05,
+      "loss": 0.8191,
+      "step": 2244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055641510518626e-05,
+      "loss": 0.7507,
+      "step": 2245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3049882848265439e-05,
+      "loss": 0.775,
+      "step": 2246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3044123070660673e-05,
+      "loss": 0.8006,
+      "step": 2247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.30383621798107e-05,
+      "loss": 0.8423,
+      "step": 2248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303260017782229e-05,
+      "loss": 0.7886,
+      "step": 2249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3026837066802629e-05,
+      "loss": 0.8521,
+      "step": 2250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021072848859304e-05,
+      "loss": 0.8507,
+      "step": 2251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3015307526100307e-05,
+      "loss": 0.7584,
+      "step": 2252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3009541100634035e-05,
+      "loss": 0.8647,
+      "step": 2253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3003773574569286e-05,
+      "loss": 0.8109,
+      "step": 2254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998004950015265e-05,
+      "loss": 0.8132,
+      "step": 2255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992235229081571e-05,
+      "loss": 0.8497,
+      "step": 2256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2986464413878214e-05,
+      "loss": 0.8312,
+      "step": 2257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980692506515598e-05,
+      "loss": 0.7995,
+      "step": 2258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2974919509104522e-05,
+      "loss": 0.8589,
+      "step": 2259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969145423756193e-05,
+      "loss": 0.7929,
+      "step": 2260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2963370252582213e-05,
+      "loss": 0.869,
+      "step": 2261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.295759399769458e-05,
+      "loss": 0.8307,
+      "step": 2262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2951816661205683e-05,
+      "loss": 0.8117,
+      "step": 2263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2946038245228324e-05,
+      "loss": 0.8569,
+      "step": 2264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940258751875672e-05,
+      "loss": 0.8357,
+      "step": 2265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934478183261314e-05,
+      "loss": 0.827,
+      "step": 2266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2928696541499221e-05,
+      "loss": 0.7962,
+      "step": 2267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922913828703756e-05,
+      "loss": 0.8296,
+      "step": 2268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917130046989679e-05,
+      "loss": 0.816,
+      "step": 2269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911345198472131e-05,
+      "loss": 0.7859,
+      "step": 2270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2905559285266653e-05,
+      "loss": 0.8173,
+      "step": 2271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2899772309489169e-05,
+      "loss": 0.8135,
+      "step": 2272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2893984273255997e-05,
+      "loss": 0.8312,
+      "step": 2273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888195178683836e-05,
+      "loss": 0.2541,
+      "step": 2274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882405027889777e-05,
+      "loss": 0.8546,
+      "step": 2275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2876613822991298e-05,
+      "loss": 0.8262,
+      "step": 2276
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870821566106257e-05,
+      "loss": 0.7935,
+      "step": 2277
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28650282593529e-05,
+      "loss": 0.8179,
+      "step": 2278
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859233904849864e-05,
+      "loss": 0.8336,
+      "step": 2279
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2853438504716156e-05,
+      "loss": 0.7953,
+      "step": 2280
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2847642061071172e-05,
+      "loss": 0.8638,
+      "step": 2281
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841844576034695e-05,
+      "loss": 0.8213,
+      "step": 2282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836046051726875e-05,
+      "loss": 0.7923,
+      "step": 2283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2830246490268257e-05,
+      "loss": 0.8238,
+      "step": 2284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2824445893779754e-05,
+      "loss": 0.8056,
+      "step": 2285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818644264382662e-05,
+      "loss": 0.8124,
+      "step": 2286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281284160419866e-05,
+      "loss": 0.7842,
+      "step": 2287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2807037915349794e-05,
+      "loss": 0.7776,
+      "step": 2288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280123319995849e-05,
+      "loss": 0.8559,
+      "step": 2289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795427460147555e-05,
+      "loss": 0.7759,
+      "step": 2290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789620698040165e-05,
+      "loss": 0.8577,
+      "step": 2291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278381291575987e-05,
+      "loss": 0.7694,
+      "step": 2292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778004115430588e-05,
+      "loss": 0.8301,
+      "step": 2293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772194299176626e-05,
+      "loss": 0.8504,
+      "step": 2294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276638346912264e-05,
+      "loss": 0.8002,
+      "step": 2295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276057162739368e-05,
+      "loss": 0.8834,
+      "step": 2296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2754758776115148e-05,
+      "loss": 0.8653,
+      "step": 2297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748944917412823e-05,
+      "loss": 0.8166,
+      "step": 2298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743130053412851e-05,
+      "loss": 0.8966,
+      "step": 2299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2737314186241748e-05,
+      "loss": 0.8059,
+      "step": 2300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.273149731802639e-05,
+      "loss": 0.8236,
+      "step": 2301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725679450894024e-05,
+      "loss": 0.8166,
+      "step": 2302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719860586972268e-05,
+      "loss": 0.8106,
+      "step": 2303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2714040728389091e-05,
+      "loss": 0.8,
+      "step": 2304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2708219877272838e-05,
+      "loss": 0.8142,
+      "step": 2305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702398035752212e-05,
+      "loss": 0.7973,
+      "step": 2306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2696575205956281e-05,
+      "loss": 0.8394,
+      "step": 2307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2690751390014464e-05,
+      "loss": 0.8569,
+      "step": 2308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684926590056552e-05,
+      "loss": 0.8163,
+      "step": 2309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679100808212694e-05,
+      "loss": 0.7969,
+      "step": 2310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673274046613395e-05,
+      "loss": 0.8331,
+      "step": 2311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266744630738952e-05,
+      "loss": 0.8015,
+      "step": 2312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2661617592672293e-05,
+      "loss": 0.3187,
+      "step": 2313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265578790459329e-05,
+      "loss": 0.2932,
+      "step": 2314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649957245284448e-05,
+      "loss": 0.7905,
+      "step": 2315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2644125616878058e-05,
+      "loss": 0.8093,
+      "step": 2316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2638293021506763e-05,
+      "loss": 0.7855,
+      "step": 2317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263245946130356e-05,
+      "loss": 0.8272,
+      "step": 2318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26266249384018e-05,
+      "loss": 0.8693,
+      "step": 2319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620789454935187e-05,
+      "loss": 0.7767,
+      "step": 2320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614953013037779e-05,
+      "loss": 0.7841,
+      "step": 2321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609115614843971e-05,
+      "loss": 0.8053,
+      "step": 2322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260327726248853e-05,
+      "loss": 0.8492,
+      "step": 2323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597437958106551e-05,
+      "loss": 0.7989,
+      "step": 2324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591597703833488e-05,
+      "loss": 0.8479,
+      "step": 2325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2585756501805139e-05,
+      "loss": 0.8723,
+      "step": 2326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2579914354157655e-05,
+      "loss": 0.8423,
+      "step": 2327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2574071263027521e-05,
+      "loss": 0.8499,
+      "step": 2328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2568227230551574e-05,
+      "loss": 0.2817,
+      "step": 2329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2562382258866997e-05,
+      "loss": 0.8095,
+      "step": 2330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255653635011131e-05,
+      "loss": 0.8192,
+      "step": 2331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550689506422388e-05,
+      "loss": 0.7991,
+      "step": 2332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2544841729938434e-05,
+      "loss": 0.8021,
+      "step": 2333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2538993022798e-05,
+      "loss": 0.8017,
+      "step": 2334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533143387139976e-05,
+      "loss": 0.7924,
+      "step": 2335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2527292825103587e-05,
+      "loss": 0.8008,
+      "step": 2336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252144133882841e-05,
+      "loss": 0.8421,
+      "step": 2337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515588930454347e-05,
+      "loss": 0.7915,
+      "step": 2338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250973560212164e-05,
+      "loss": 0.8519,
+      "step": 2339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2503881355970875e-05,
+      "loss": 0.7641,
+      "step": 2340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2498026194142961e-05,
+      "loss": 0.7516,
+      "step": 2341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492170118779152e-05,
+      "loss": 0.8327,
+      "step": 2342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486313132021035e-05,
+      "loss": 0.8288,
+      "step": 2343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480455236010527e-05,
+      "loss": 0.7646,
+      "step": 2344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474596432889876e-05,
+      "loss": 0.7923,
+      "step": 2345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2468736724801666e-05,
+      "loss": 0.8402,
+      "step": 2346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2462876113888811e-05,
+      "loss": 0.8318,
+      "step": 2347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2457014602294552e-05,
+      "loss": 0.7802,
+      "step": 2348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2451152192162467e-05,
+      "loss": 0.8178,
+      "step": 2349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445288885636455e-05,
+      "loss": 0.8408,
+      "step": 2350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2439424684860743e-05,
+      "loss": 0.8238,
+      "step": 2351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2433559591979892e-05,
+      "loss": 0.8176,
+      "step": 2352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427693609138777e-05,
+      "loss": 0.788,
+      "step": 2353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2421826738482614e-05,
+      "loss": 0.834,
+      "step": 2354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2415958982156933e-05,
+      "loss": 0.8371,
+      "step": 2355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410090342307586e-05,
+      "loss": 0.8418,
+      "step": 2356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240422082108076e-05,
+      "loss": 0.82,
+      "step": 2357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398350420622951e-05,
+      "loss": 0.8459,
+      "step": 2358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392479143080988e-05,
+      "loss": 0.826,
+      "step": 2359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386606990602012e-05,
+      "loss": 0.7771,
+      "step": 2360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380733965333489e-05,
+      "loss": 0.8098,
+      "step": 2361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374860069423203e-05,
+      "loss": 0.8688,
+      "step": 2362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236898530501925e-05,
+      "loss": 0.8636,
+      "step": 2363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2363109674270056e-05,
+      "loss": 0.8341,
+      "step": 2364
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2357233179324352e-05,
+      "loss": 0.8103,
+      "step": 2365
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351355822331198e-05,
+      "loss": 0.8854,
+      "step": 2366
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2345477605439958e-05,
+      "loss": 0.822,
+      "step": 2367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339598530800312e-05,
+      "loss": 0.7924,
+      "step": 2368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2333718600562256e-05,
+      "loss": 0.8372,
+      "step": 2369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327837816876098e-05,
+      "loss": 0.8199,
+      "step": 2370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2321956181892463e-05,
+      "loss": 0.82,
+      "step": 2371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.231607369776228e-05,
+      "loss": 0.811,
+      "step": 2372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310190366636789e-05,
+      "loss": 0.273,
+      "step": 2373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.230430619066755e-05,
+      "loss": 0.8715,
+      "step": 2374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2298421172006416e-05,
+      "loss": 0.8014,
+      "step": 2375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2292535312805563e-05,
+      "loss": 0.8057,
+      "step": 2376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286648615217465e-05,
+      "loss": 0.7911,
+      "step": 2377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280761081394905e-05,
+      "loss": 0.8712,
+      "step": 2378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274872713490973e-05,
+      "loss": 0.8341,
+      "step": 2379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2268983513659062e-05,
+      "loss": 0.8563,
+      "step": 2380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2263093484052875e-05,
+      "loss": 0.8188,
+      "step": 2381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2257202626826406e-05,
+      "loss": 0.8495,
+      "step": 2382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225131094413397e-05,
+      "loss": 0.7671,
+      "step": 2383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2245418438130166e-05,
+      "loss": 0.8471,
+      "step": 2384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.22395251109699e-05,
+      "loss": 0.7907,
+      "step": 2385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233630964808387e-05,
+      "loss": 0.8144,
+      "step": 2386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227736001801134e-05,
+      "loss": 0.8245,
+      "step": 2387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.222184022410394e-05,
+      "loss": 0.8307,
+      "step": 2388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2215943633872915e-05,
+      "loss": 0.8186,
+      "step": 2389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210046233264458e-05,
+      "loss": 0.8583,
+      "step": 2390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220414802443527e-05,
+      "loss": 0.841,
+      "step": 2391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2198249009542338e-05,
+      "loss": 0.8164,
+      "step": 2392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192349190742956e-05,
+      "loss": 0.8207,
+      "step": 2393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186448570194703e-05,
+      "loss": 0.8606,
+      "step": 2394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2180547150055456e-05,
+      "loss": 0.8241,
+      "step": 2395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217464493248338e-05,
+      "loss": 0.8603,
+      "step": 2396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216874191963693e-05,
+      "loss": 0.3057,
+      "step": 2397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2162838113674868e-05,
+      "loss": 0.2931,
+      "step": 2398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156933516756225e-05,
+      "loss": 0.8559,
+      "step": 2399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2151028131040331e-05,
+      "loss": 0.8149,
+      "step": 2400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214512195868681e-05,
+      "loss": 0.7948,
+      "step": 2401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213921500185556e-05,
+      "loss": 0.7513,
+      "step": 2402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2133307262706778e-05,
+      "loss": 0.8194,
+      "step": 2403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2127398743400943e-05,
+      "loss": 0.259,
+      "step": 2404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2121489446098813e-05,
+      "loss": 0.8081,
+      "step": 2405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211557937296144e-05,
+      "loss": 0.8534,
+      "step": 2406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109668526150156e-05,
+      "loss": 0.8592,
+      "step": 2407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2103756907826578e-05,
+      "loss": 0.8696,
+      "step": 2408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2097844520152597e-05,
+      "loss": 0.8586,
+      "step": 2409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091931365290399e-05,
+      "loss": 0.8605,
+      "step": 2410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.208601744540244e-05,
+      "loss": 0.7884,
+      "step": 2411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2080102762651458e-05,
+      "loss": 0.7858,
+      "step": 2412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207418731920047e-05,
+      "loss": 0.8583,
+      "step": 2413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2068271117212768e-05,
+      "loss": 0.7712,
+      "step": 2414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062354158851936e-05,
+      "loss": 0.7823,
+      "step": 2415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2056436446281816e-05,
+      "loss": 0.3024,
+      "step": 2416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050517981666536e-05,
+      "loss": 0.7792,
+      "step": 2417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20445987671705e-05,
+      "loss": 0.7726,
+      "step": 2418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2038678804958375e-05,
+      "loss": 0.8228,
+      "step": 2419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032758097195116e-05,
+      "loss": 0.8891,
+      "step": 2420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026836646045942e-05,
+      "loss": 0.829,
+      "step": 2421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2020914453676346e-05,
+      "loss": 0.8505,
+      "step": 2422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2014991522252094e-05,
+      "loss": 0.8403,
+      "step": 2423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2009067853939218e-05,
+      "loss": 0.8266,
+      "step": 2424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2003143450904022e-05,
+      "loss": 0.8447,
+      "step": 2425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997218315313077e-05,
+      "loss": 0.8379,
+      "step": 2426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991292449333228e-05,
+      "loss": 0.7885,
+      "step": 2427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1985365855131576e-05,
+      "loss": 0.8732,
+      "step": 2428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979438534875505e-05,
+      "loss": 0.8295,
+      "step": 2429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1973510490732645e-05,
+      "loss": 0.8229,
+      "step": 2430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19675817248709e-05,
+      "loss": 0.8798,
+      "step": 2431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1961652239458445e-05,
+      "loss": 0.7495,
+      "step": 2432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1955722036663704e-05,
+      "loss": 0.8104,
+      "step": 2433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1949791118655376e-05,
+      "loss": 0.8264,
+      "step": 2434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1943859487602411e-05,
+      "loss": 0.8013,
+      "step": 2435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193792714567403e-05,
+      "loss": 0.8402,
+      "step": 2436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1931994095039704e-05,
+      "loss": 0.846,
+      "step": 2437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1926060337869172e-05,
+      "loss": 0.7602,
+      "step": 2438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920125876332425e-05,
+      "loss": 0.8135,
+      "step": 2439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914190712599714e-05,
+      "loss": 0.7692,
+      "step": 2440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1908254848841547e-05,
+      "loss": 0.8277,
+      "step": 2441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902318287228687e-05,
+      "loss": 0.8743,
+      "step": 2442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896381029932154e-05,
+      "loss": 0.8667,
+      "step": 2443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1890443079123221e-05,
+      "loss": 0.8446,
+      "step": 2444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884504436973414e-05,
+      "loss": 0.7823,
+      "step": 2445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878565105654515e-05,
+      "loss": 0.7856,
+      "step": 2446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1872625087338557e-05,
+      "loss": 0.8142,
+      "step": 2447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1866684384197818e-05,
+      "loss": 0.826,
+      "step": 2448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1860742998404834e-05,
+      "loss": 0.8139,
+      "step": 2449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.185480093213239e-05,
+      "loss": 0.8082,
+      "step": 2450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1848858187553514e-05,
+      "loss": 0.7965,
+      "step": 2451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842914766841487e-05,
+      "loss": 0.824,
+      "step": 2452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1836970672169841e-05,
+      "loss": 0.8366,
+      "step": 2453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831025905712346e-05,
+      "loss": 0.7979,
+      "step": 2454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182508046964302e-05,
+      "loss": 0.8043,
+      "step": 2455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181913436613613e-05,
+      "loss": 0.7239,
+      "step": 2456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813187597366183e-05,
+      "loss": 0.8406,
+      "step": 2457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1807240165507931e-05,
+      "loss": 0.7917,
+      "step": 2458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180129207273637e-05,
+      "loss": 0.8458,
+      "step": 2459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795343321226733e-05,
+      "loss": 0.6785,
+      "step": 2460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17893939131545e-05,
+      "loss": 0.8655,
+      "step": 2461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783443850695381e-05,
+      "loss": 0.7562,
+      "step": 2462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1777493136025342e-05,
+      "loss": 0.8673,
+      "step": 2463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1771541771320573e-05,
+      "loss": 0.8403,
+      "step": 2464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765589758757506e-05,
+      "loss": 0.8138,
+      "step": 2465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175963710051281e-05,
+      "loss": 0.8631,
+      "step": 2466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1753683798763392e-05,
+      "loss": 0.8189,
+      "step": 2467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1747729855686392e-05,
+      "loss": 0.8333,
+      "step": 2468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741775273459185e-05,
+      "loss": 0.7951,
+      "step": 2469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173582005425938e-05,
+      "loss": 0.747,
+      "step": 2470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729864200264822e-05,
+      "loss": 0.8226,
+      "step": 2471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1723907713653582e-05,
+      "loss": 0.776,
+      "step": 2472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717950596603967e-05,
+      "loss": 0.8648,
+      "step": 2473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171199285129451e-05,
+      "loss": 0.8165,
+      "step": 2474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1706034479903983e-05,
+      "loss": 0.8111,
+      "step": 2475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700075484611374e-05,
+      "loss": 0.8545,
+      "step": 2476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169411586759591e-05,
+      "loss": 0.8154,
+      "step": 2477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688155631037043e-05,
+      "loss": 0.2314,
+      "step": 2478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1682194777114443e-05,
+      "loss": 0.8102,
+      "step": 2479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1676233308008022e-05,
+      "loss": 0.7809,
+      "step": 2480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670271225897902e-05,
+      "loss": 0.8351,
+      "step": 2481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664308532964435e-05,
+      "loss": 0.8009,
+      "step": 2482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1658345231388198e-05,
+      "loss": 0.8402,
+      "step": 2483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1652381323349986e-05,
+      "loss": 0.8168,
+      "step": 2484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646416811030821e-05,
+      "loss": 0.7892,
+      "step": 2485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640451696611941e-05,
+      "loss": 0.8107,
+      "step": 2486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1634485982274811e-05,
+      "loss": 0.2778,
+      "step": 2487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1628519670201109e-05,
+      "loss": 0.8222,
+      "step": 2488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1622552762572732e-05,
+      "loss": 0.8109,
+      "step": 2489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616585261571803e-05,
+      "loss": 0.8344,
+      "step": 2490
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1610617169380643e-05,
+      "loss": 0.8511,
+      "step": 2491
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1604648488181813e-05,
+      "loss": 0.7695,
+      "step": 2492
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1598679220158078e-05,
+      "loss": 0.824,
+      "step": 2493
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1592709367492408e-05,
+      "loss": 0.8719,
+      "step": 2494
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586738932368008e-05,
+      "loss": 0.8774,
+      "step": 2495
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1580767916968276e-05,
+      "loss": 0.8485,
+      "step": 2496
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1574796323476836e-05,
+      "loss": 0.8195,
+      "step": 2497
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1568824154077516e-05,
+      "loss": 0.9022,
+      "step": 2498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562851410954358e-05,
+      "loss": 0.757,
+      "step": 2499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1556878096291613e-05,
+      "loss": 0.7978,
+      "step": 2500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550904212273737e-05,
+      "loss": 0.7132,
+      "step": 2501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1544929761085401e-05,
+      "loss": 0.7996,
+      "step": 2502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1538954744911479e-05,
+      "loss": 0.8479,
+      "step": 2503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1532979165937055e-05,
+      "loss": 0.8199,
+      "step": 2504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527003026347414e-05,
+      "loss": 0.8142,
+      "step": 2505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152102632832805e-05,
+      "loss": 0.822,
+      "step": 2506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1515049074064659e-05,
+      "loss": 0.8284,
+      "step": 2507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1509071265743143e-05,
+      "loss": 0.8026,
+      "step": 2508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503092905549605e-05,
+      "loss": 0.8081,
+      "step": 2509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497113995670347e-05,
+      "loss": 0.8331,
+      "step": 2510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491134538291878e-05,
+      "loss": 0.8235,
+      "step": 2511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1485154535600902e-05,
+      "loss": 0.7747,
+      "step": 2512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479173989784323e-05,
+      "loss": 0.8007,
+      "step": 2513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147319290302925e-05,
+      "loss": 0.8209,
+      "step": 2514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467211277522982e-05,
+      "loss": 0.8496,
+      "step": 2515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1461229115453019e-05,
+      "loss": 0.7945,
+      "step": 2516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455246419007056e-05,
+      "loss": 0.8241,
+      "step": 2517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449263190372977e-05,
+      "loss": 0.7971,
+      "step": 2518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1443279431738879e-05,
+      "loss": 0.8308,
+      "step": 2519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.143729514529303e-05,
+      "loss": 0.8344,
+      "step": 2520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1431310333223911e-05,
+      "loss": 0.7574,
+      "step": 2521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142532499772018e-05,
+      "loss": 0.839,
+      "step": 2522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1419339140970693e-05,
+      "loss": 0.7793,
+      "step": 2523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1413352765164503e-05,
+      "loss": 0.8202,
+      "step": 2524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140736587249084e-05,
+      "loss": 0.8061,
+      "step": 2525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140137846513913e-05,
+      "loss": 0.8333,
+      "step": 2526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1395390545298988e-05,
+      "loss": 0.8233,
+      "step": 2527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1389402115160211e-05,
+      "loss": 0.762,
+      "step": 2528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138341317691279e-05,
+      "loss": 0.8628,
+      "step": 2529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1377423732746897e-05,
+      "loss": 0.8437,
+      "step": 2530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137143378485289e-05,
+      "loss": 0.8386,
+      "step": 2531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136544333542131e-05,
+      "loss": 0.7953,
+      "step": 2532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1359452386642892e-05,
+      "loss": 0.8606,
+      "step": 2533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1353460940708535e-05,
+      "loss": 0.8056,
+      "step": 2534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1347468999809327e-05,
+      "loss": 0.8489,
+      "step": 2535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1341476566136546e-05,
+      "loss": 0.7593,
+      "step": 2536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1335483641881642e-05,
+      "loss": 0.8161,
+      "step": 2537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1329490229236247e-05,
+      "loss": 0.8318,
+      "step": 2538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1323496330392167e-05,
+      "loss": 0.7946,
+      "step": 2539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131750194754139e-05,
+      "loss": 0.2901,
+      "step": 2540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1311507082876081e-05,
+      "loss": 0.7932,
+      "step": 2541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1305511738588583e-05,
+      "loss": 0.8317,
+      "step": 2542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129951591687141e-05,
+      "loss": 0.7913,
+      "step": 2543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1293519619917252e-05,
+      "loss": 0.794,
+      "step": 2544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287522849918973e-05,
+      "loss": 0.8626,
+      "step": 2545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281525609069612e-05,
+      "loss": 0.7323,
+      "step": 2546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1275527899562374e-05,
+      "loss": 0.8244,
+      "step": 2547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269529723590649e-05,
+      "loss": 0.8051,
+      "step": 2548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263531083347979e-05,
+      "loss": 0.8684,
+      "step": 2549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125753198102809e-05,
+      "loss": 0.8211,
+      "step": 2550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1251532418824871e-05,
+      "loss": 0.8276,
+      "step": 2551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1245532398932384e-05,
+      "loss": 0.781,
+      "step": 2552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1239531923544848e-05,
+      "loss": 0.8391,
+      "step": 2553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1233530994856658e-05,
+      "loss": 0.8413,
+      "step": 2554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1227529615062378e-05,
+      "loss": 0.7525,
+      "step": 2555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221527786356723e-05,
+      "loss": 0.8479,
+      "step": 2556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215525510934583e-05,
+      "loss": 0.7922,
+      "step": 2557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1209522790991011e-05,
+      "loss": 0.8281,
+      "step": 2558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120351962872122e-05,
+      "loss": 0.7832,
+      "step": 2559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1197516026320581e-05,
+      "loss": 0.8465,
+      "step": 2560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1191511985984633e-05,
+      "loss": 0.7962,
+      "step": 2561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1185507509909071e-05,
+      "loss": 0.804,
+      "step": 2562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1179502600289752e-05,
+      "loss": 0.8645,
+      "step": 2563
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173497259322685e-05,
+      "loss": 0.7925,
+      "step": 2564
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1167491489204051e-05,
+      "loss": 0.2804,
+      "step": 2565
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116148529213017e-05,
+      "loss": 0.8083,
+      "step": 2566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1155478670297532e-05,
+      "loss": 0.77,
+      "step": 2567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1149471625902775e-05,
+      "loss": 0.7805,
+      "step": 2568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143464161142696e-05,
+      "loss": 0.8274,
+      "step": 2569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113745627821424e-05,
+      "loss": 0.8049,
+      "step": 2570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1131447979314506e-05,
+      "loss": 0.7947,
+      "step": 2571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1125439266640756e-05,
+      "loss": 0.7938,
+      "step": 2572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119430142390388e-05,
+      "loss": 0.8192,
+      "step": 2573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113420608760959e-05,
+      "loss": 0.8599,
+      "step": 2574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1107410667950176e-05,
+      "loss": 0.744,
+      "step": 2575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1101400322155892e-05,
+      "loss": 0.7975,
+      "step": 2576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1095389573576108e-05,
+      "loss": 0.8452,
+      "step": 2577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108937842440897e-05,
+      "loss": 0.7793,
+      "step": 2578
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083366876852781e-05,
+      "loss": 0.8084,
+      "step": 2579
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1077354933105976e-05,
+      "loss": 0.7999,
+      "step": 2580
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071342595367148e-05,
+      "loss": 0.7846,
+      "step": 2581
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065329865835022e-05,
+      "loss": 0.8029,
+      "step": 2582
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105931674670847e-05,
+      "loss": 0.7983,
+      "step": 2583
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1053303240186517e-05,
+      "loss": 0.8138,
+      "step": 2584
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1047289348468316e-05,
+      "loss": 0.2576,
+      "step": 2585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1041275073753163e-05,
+      "loss": 0.8436,
+      "step": 2586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1035260418240502e-05,
+      "loss": 0.8017,
+      "step": 2587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1029245384129903e-05,
+      "loss": 0.8208,
+      "step": 2588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023229973621092e-05,
+      "loss": 0.7685,
+      "step": 2589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017214188913921e-05,
+      "loss": 0.8042,
+      "step": 2590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1011198032208374e-05,
+      "loss": 0.8309,
+      "step": 2591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1005181505704588e-05,
+      "loss": 0.7697,
+      "step": 2592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999164611602817e-05,
+      "loss": 0.791,
+      "step": 2593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993147352103466e-05,
+      "loss": 0.7743,
+      "step": 2594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0987129729407055e-05,
+      "loss": 0.8354,
+      "step": 2595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0981111745714257e-05,
+      "loss": 0.8147,
+      "step": 2596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975093403225862e-05,
+      "loss": 0.8188,
+      "step": 2597
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969074704142799e-05,
+      "loss": 0.8284,
+      "step": 2598
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0963055650666121e-05,
+      "loss": 0.8266,
+      "step": 2599
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.095703624499702e-05,
+      "loss": 0.8199,
+      "step": 2600
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0951016489336805e-05,
+      "loss": 0.7806,
+      "step": 2601
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0944996385886925e-05,
+      "loss": 0.8236,
+      "step": 2602
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938975936848948e-05,
+      "loss": 0.8556,
+      "step": 2603
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0932955144424571e-05,
+      "loss": 0.7759,
+      "step": 2604
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0926934010815613e-05,
+      "loss": 0.8241,
+      "step": 2605
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0920912538224027e-05,
+      "loss": 0.2781,
+      "step": 2606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0914890728851879e-05,
+      "loss": 0.8061,
+      "step": 2607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908868584901367e-05,
+      "loss": 0.7438,
+      "step": 2608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902846108574807e-05,
+      "loss": 0.8316,
+      "step": 2609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0896823302074633e-05,
+      "loss": 0.84,
+      "step": 2610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089080016760341e-05,
+      "loss": 0.2869,
+      "step": 2611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0884776707363815e-05,
+      "loss": 0.7687,
+      "step": 2612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0878752923558646e-05,
+      "loss": 0.8317,
+      "step": 2613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0872728818390818e-05,
+      "loss": 0.2989,
+      "step": 2614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0866704394063366e-05,
+      "loss": 0.8177,
+      "step": 2615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860679652779438e-05,
+      "loss": 0.8281,
+      "step": 2616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854654596742303e-05,
+      "loss": 0.8068,
+      "step": 2617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0848629228155344e-05,
+      "loss": 0.8315,
+      "step": 2618
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0842603549222053e-05,
+      "loss": 0.8694,
+      "step": 2619
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0836577562146042e-05,
+      "loss": 0.7736,
+      "step": 2620
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0830551269131036e-05,
+      "loss": 0.8303,
+      "step": 2621
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082452467238086e-05,
+      "loss": 0.757,
+      "step": 2622
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818497774099468e-05,
+      "loss": 0.8078,
+      "step": 2623
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812470576490912e-05,
+      "loss": 0.7683,
+      "step": 2624
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0806443081759355e-05,
+      "loss": 0.7838,
+      "step": 2625
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0800415292109072e-05,
+      "loss": 0.8284,
+      "step": 2626
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0794387209744444e-05,
+      "loss": 0.8131,
+      "step": 2627
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0788358836869963e-05,
+      "loss": 0.8211,
+      "step": 2628
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.078233017569022e-05,
+      "loss": 0.8372,
+      "step": 2629
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776301228409913e-05,
+      "loss": 0.7818,
+      "step": 2630
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077027199723385e-05,
+      "loss": 0.7922,
+      "step": 2631
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0764242484366939e-05,
+      "loss": 0.2647,
+      "step": 2632
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0758212692014189e-05,
+      "loss": 0.7793,
+      "step": 2633
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0752182622380715e-05,
+      "loss": 0.8408,
+      "step": 2634
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746152277671735e-05,
+      "loss": 0.826,
+      "step": 2635
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0740121660092564e-05,
+      "loss": 0.7927,
+      "step": 2636
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734090771848616e-05,
+      "loss": 0.7762,
+      "step": 2637
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728059615145404e-05,
+      "loss": 0.8025,
+      "step": 2638
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722028192188538e-05,
+      "loss": 0.7748,
+      "step": 2639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0715996505183737e-05,
+      "loss": 0.8337,
+      "step": 2640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0709964556336799e-05,
+      "loss": 0.7954,
+      "step": 2641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0703932347853631e-05,
+      "loss": 0.2853,
+      "step": 2642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0697899881940232e-05,
+      "loss": 0.816,
+      "step": 2643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691867160802688e-05,
+      "loss": 0.8389,
+      "step": 2644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685834186647186e-05,
+      "loss": 0.7917,
+      "step": 2645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0679800961680004e-05,
+      "loss": 0.2647,
+      "step": 2646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0673767488107514e-05,
+      "loss": 0.7674,
+      "step": 2647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066773376813617e-05,
+      "loss": 0.8311,
+      "step": 2648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0661699803972524e-05,
+      "loss": 0.791,
+      "step": 2649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0655665597823214e-05,
+      "loss": 0.8899,
+      "step": 2650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.064963115189497e-05,
+      "loss": 0.7837,
+      "step": 2651
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064359646839461e-05,
+      "loss": 0.8265,
+      "step": 2652
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0637561549529035e-05,
+      "loss": 0.8148,
+      "step": 2653
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0631526397505228e-05,
+      "loss": 0.7939,
+      "step": 2654
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0625491014530271e-05,
+      "loss": 0.8534,
+      "step": 2655
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0619455402811315e-05,
+      "loss": 0.8249,
+      "step": 2656
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613419564555605e-05,
+      "loss": 0.7963,
+      "step": 2657
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0607383501970466e-05,
+      "loss": 0.7682,
+      "step": 2658
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0601347217263305e-05,
+      "loss": 0.7766,
+      "step": 2659
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059531071264161e-05,
+      "loss": 0.7858,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589273990312944e-05,
+      "loss": 0.8141,
+      "step": 2661
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0583237052484962e-05,
+      "loss": 0.8025,
+      "step": 2662
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577199901365391e-05,
+      "loss": 0.7983,
+      "step": 2663
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571162539162032e-05,
+      "loss": 0.8686,
+      "step": 2664
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056512496808277e-05,
+      "loss": 0.8339,
+      "step": 2665
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0559087190335557e-05,
+      "loss": 0.8331,
+      "step": 2666
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0553049208128436e-05,
+      "loss": 0.8142,
+      "step": 2667
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0547011023669508e-05,
+      "loss": 0.7763,
+      "step": 2668
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540972639166962e-05,
+      "loss": 0.7913,
+      "step": 2669
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053493405682905e-05,
+      "loss": 0.788,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05288952788641e-05,
+      "loss": 0.7848,
+      "step": 2671
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522856307480514e-05,
+      "loss": 0.7707,
+      "step": 2672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0516817144886762e-05,
+      "loss": 0.8445,
+      "step": 2673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0510777793291383e-05,
+      "loss": 0.8731,
+      "step": 2674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0504738254902987e-05,
+      "loss": 0.8226,
+      "step": 2675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498698531930253e-05,
+      "loss": 0.77,
+      "step": 2676
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0492658626581927e-05,
+      "loss": 0.7539,
+      "step": 2677
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486618541066815e-05,
+      "loss": 0.7821,
+      "step": 2678
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0480578277593802e-05,
+      "loss": 0.776,
+      "step": 2679
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047453783837183e-05,
+      "loss": 0.779,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468497225609904e-05,
+      "loss": 0.8155,
+      "step": 2681
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462456441517098e-05,
+      "loss": 0.7823,
+      "step": 2682
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0456415488302535e-05,
+      "loss": 0.8008,
+      "step": 2683
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0450374368175427e-05,
+      "loss": 0.7841,
+      "step": 2684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444333083345016e-05,
+      "loss": 0.7893,
+      "step": 2685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438291636020625e-05,
+      "loss": 0.8213,
+      "step": 2686
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043225002841163e-05,
+      "loss": 0.7873,
+      "step": 2687
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0426208262727464e-05,
+      "loss": 0.7797,
+      "step": 2688
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042016634117762e-05,
+      "loss": 0.8096,
+      "step": 2689
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0414124265971649e-05,
+      "loss": 0.8479,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0408082039319154e-05,
+      "loss": 0.7678,
+      "step": 2691
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0402039663429796e-05,
+      "loss": 0.788,
+      "step": 2692
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039599714051329e-05,
+      "loss": 0.777,
+      "step": 2693
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0389954472779408e-05,
+      "loss": 0.8064,
+      "step": 2694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0383911662437969e-05,
+      "loss": 0.8274,
+      "step": 2695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0377868711698853e-05,
+      "loss": 0.7845,
+      "step": 2696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371825622771979e-05,
+      "loss": 0.7876,
+      "step": 2697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036578239786733e-05,
+      "loss": 0.7889,
+      "step": 2698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359739039194925e-05,
+      "loss": 0.8055,
+      "step": 2699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0353695548964842e-05,
+      "loss": 0.7545,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0347651929387206e-05,
+      "loss": 0.7941,
+      "step": 2701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341608182672182e-05,
+      "loss": 0.7503,
+      "step": 2702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335564311029995e-05,
+      "loss": 0.8057,
+      "step": 2703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.03295203166709e-05,
+      "loss": 0.7918,
+      "step": 2704
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.032347620180521e-05,
+      "loss": 0.7819,
+      "step": 2705
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0317431968643272e-05,
+      "loss": 0.7422,
+      "step": 2706
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311387619395486e-05,
+      "loss": 0.7998,
+      "step": 2707
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0305343156272283e-05,
+      "loss": 0.8082,
+      "step": 2708
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0299298581484143e-05,
+      "loss": 0.8077,
+      "step": 2709
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029325389724159e-05,
+      "loss": 0.8021,
+      "step": 2710
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028720910575518e-05,
+      "loss": 0.8143,
+      "step": 2711
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028116420923551e-05,
+      "loss": 0.8456,
+      "step": 2712
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275119209893226e-05,
+      "loss": 0.778,
+      "step": 2713
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0269074109938992e-05,
+      "loss": 0.7973,
+      "step": 2714
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0263028911583529e-05,
+      "loss": 0.7797,
+      "step": 2715
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0256983617037582e-05,
+      "loss": 0.7889,
+      "step": 2716
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0250938228511926e-05,
+      "loss": 0.8377,
+      "step": 2717
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024489274821739e-05,
+      "loss": 0.8163,
+      "step": 2718
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0238847178364814e-05,
+      "loss": 0.8042,
+      "step": 2719
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232801521165092e-05,
+      "loss": 0.7769,
+      "step": 2720
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0226755778829131e-05,
+      "loss": 0.8228,
+      "step": 2721
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0220709953567886e-05,
+      "loss": 0.7926,
+      "step": 2722
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0214664047592329e-05,
+      "loss": 0.7705,
+      "step": 2723
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0208618063113465e-05,
+      "loss": 0.7374,
+      "step": 2724
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.020257200234233e-05,
+      "loss": 0.8324,
+      "step": 2725
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0196525867489989e-05,
+      "loss": 0.7962,
+      "step": 2726
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0190479660767532e-05,
+      "loss": 0.835,
+      "step": 2727
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184433384386074e-05,
+      "loss": 0.7939,
+      "step": 2728
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0178387040556756e-05,
+      "loss": 0.8515,
+      "step": 2729
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0172340631490747e-05,
+      "loss": 0.8203,
+      "step": 2730
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166294159399237e-05,
+      "loss": 0.7611,
+      "step": 2731
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016024762649344e-05,
+      "loss": 0.819,
+      "step": 2732
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154201034984591e-05,
+      "loss": 0.8173,
+      "step": 2733
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0148154387083948e-05,
+      "loss": 0.7948,
+      "step": 2734
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0142107685002788e-05,
+      "loss": 0.8332,
+      "step": 2735
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0136060930952408e-05,
+      "loss": 0.8106,
+      "step": 2736
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130014127144126e-05,
+      "loss": 0.8139,
+      "step": 2737
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0123967275789272e-05,
+      "loss": 0.7768,
+      "step": 2738
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0117920379099206e-05,
+      "loss": 0.749,
+      "step": 2739
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0111873439285293e-05,
+      "loss": 0.8259,
+      "step": 2740
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105826458558918e-05,
+      "loss": 0.8349,
+      "step": 2741
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009977943913148e-05,
+      "loss": 0.8253,
+      "step": 2742
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0093732383214392e-05,
+      "loss": 0.7752,
+      "step": 2743
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087685293019084e-05,
+      "loss": 0.7973,
+      "step": 2744
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0081638170756991e-05,
+      "loss": 0.734,
+      "step": 2745
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0075591018639567e-05,
+      "loss": 0.7763,
+      "step": 2746
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0069543838878275e-05,
+      "loss": 0.8315,
+      "step": 2747
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063496633684586e-05,
+      "loss": 0.7771,
+      "step": 2748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0057449405269983e-05,
+      "loss": 0.8175,
+      "step": 2749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0051402155845955e-05,
+      "loss": 0.7922,
+      "step": 2750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0045354887623999e-05,
+      "loss": 0.7882,
+      "step": 2751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.003930760281562e-05,
+      "loss": 0.8036,
+      "step": 2752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033260303632328e-05,
+      "loss": 0.8408,
+      "step": 2753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0027212992285644e-05,
+      "loss": 0.8495,
+      "step": 2754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.002116567098708e-05,
+      "loss": 0.7323,
+      "step": 2755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0015118341948169e-05,
+      "loss": 0.7507,
+      "step": 2756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0009071007380433e-05,
+      "loss": 0.7819,
+      "step": 2757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0003023669495399e-05,
+      "loss": 0.8377,
+      "step": 2758
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.996976330504604e-06,
+      "loss": 0.8181,
+      "step": 2759
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990928992619572e-06,
+      "loss": 0.7717,
+      "step": 2760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984881658051831e-06,
+      "loss": 0.8062,
+      "step": 2761
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97883432901292e-06,
+      "loss": 0.8254,
+      "step": 2762
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97278700771436e-06,
+      "loss": 0.8198,
+      "step": 2763
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966739696367674e-06,
+      "loss": 0.7786,
+      "step": 2764
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960692397184383e-06,
+      "loss": 0.8661,
+      "step": 2765
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.954645112376003e-06,
+      "loss": 0.8144,
+      "step": 2766
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948597844154048e-06,
+      "loss": 0.7909,
+      "step": 2767
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94255059473002e-06,
+      "loss": 0.7882,
+      "step": 2768
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936503366315417e-06,
+      "loss": 0.8175,
+      "step": 2769
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930456161121729e-06,
+      "loss": 0.8312,
+      "step": 2770
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.924408981360433e-06,
+      "loss": 0.284,
+      "step": 2771
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91836182924301e-06,
+      "loss": 0.7968,
+      "step": 2772
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912314706980918e-06,
+      "loss": 0.7667,
+      "step": 2773
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.906267616785611e-06,
+      "loss": 0.8089,
+      "step": 2774
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.900220560868524e-06,
+      "loss": 0.8032,
+      "step": 2775
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894173541441084e-06,
+      "loss": 0.8085,
+      "step": 2776
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888126560714709e-06,
+      "loss": 0.8233,
+      "step": 2777
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.882079620900796e-06,
+      "loss": 0.7718,
+      "step": 2778
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.876032724210731e-06,
+      "loss": 0.288,
+      "step": 2779
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869985872855881e-06,
+      "loss": 0.7916,
+      "step": 2780
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.863939069047594e-06,
+      "loss": 0.7797,
+      "step": 2781
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.857892314997215e-06,
+      "loss": 0.8081,
+      "step": 2782
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851845612916055e-06,
+      "loss": 0.7982,
+      "step": 2783
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.84579896501541e-06,
+      "loss": 0.7458,
+      "step": 2784
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839752373506563e-06,
+      "loss": 0.8378,
+      "step": 2785
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833705840600763e-06,
+      "loss": 0.7552,
+      "step": 2786
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.827659368509255e-06,
+      "loss": 0.8445,
+      "step": 2787
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.821612959443247e-06,
+      "loss": 0.8342,
+      "step": 2788
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815566615613932e-06,
+      "loss": 0.8026,
+      "step": 2789
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80952033923247e-06,
+      "loss": 0.7718,
+      "step": 2790
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803474132510013e-06,
+      "loss": 0.7757,
+      "step": 2791
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797427997657673e-06,
+      "loss": 0.799,
+      "step": 2792
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791381936886538e-06,
+      "loss": 0.7559,
+      "step": 2793
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785335952407675e-06,
+      "loss": 0.8071,
+      "step": 2794
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.779290046432116e-06,
+      "loss": 0.824,
+      "step": 2795
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.773244221170867e-06,
+      "loss": 0.8052,
+      "step": 2796
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767198478834911e-06,
+      "loss": 0.8143,
+      "step": 2797
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761152821635189e-06,
+      "loss": 0.774,
+      "step": 2798
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.755107251782615e-06,
+      "loss": 0.7976,
+      "step": 2799
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.749061771488074e-06,
+      "loss": 0.8192,
+      "step": 2800
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743016382962422e-06,
+      "loss": 0.826,
+      "step": 2801
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.736971088416476e-06,
+      "loss": 0.7332,
+      "step": 2802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.730925890061012e-06,
+      "loss": 0.8067,
+      "step": 2803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72488079010678e-06,
+      "loss": 0.808,
+      "step": 2804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71883579076449e-06,
+      "loss": 0.8245,
+      "step": 2805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.712790894244823e-06,
+      "loss": 0.266,
+      "step": 2806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706746102758414e-06,
+      "loss": 0.8446,
+      "step": 2807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700701418515862e-06,
+      "loss": 0.7989,
+      "step": 2808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.694656843727723e-06,
+      "loss": 0.81,
+      "step": 2809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68861238060452e-06,
+      "loss": 0.8443,
+      "step": 2810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682568031356731e-06,
+      "loss": 0.8144,
+      "step": 2811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.676523798194794e-06,
+      "loss": 0.7669,
+      "step": 2812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670479683329103e-06,
+      "loss": 0.7911,
+      "step": 2813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66443568897001e-06,
+      "loss": 0.7871,
+      "step": 2814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658391817327816e-06,
+      "loss": 0.8484,
+      "step": 2815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.652348070612797e-06,
+      "loss": 0.2391,
+      "step": 2816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646304451035162e-06,
+      "loss": 0.822,
+      "step": 2817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640260960805078e-06,
+      "loss": 0.8467,
+      "step": 2818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634217602132676e-06,
+      "loss": 0.7895,
+      "step": 2819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628174377228023e-06,
+      "loss": 0.8114,
+      "step": 2820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62213128830115e-06,
+      "loss": 0.7929,
+      "step": 2821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.616088337562033e-06,
+      "loss": 0.744,
+      "step": 2822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610045527220595e-06,
+      "loss": 0.7854,
+      "step": 2823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.60400285948671e-06,
+      "loss": 0.7813,
+      "step": 2824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.597960336570206e-06,
+      "loss": 0.8102,
+      "step": 2825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59191796068085e-06,
+      "loss": 0.8546,
+      "step": 2826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.585875734028355e-06,
+      "loss": 0.8073,
+      "step": 2827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579833658822384e-06,
+      "loss": 0.8075,
+      "step": 2828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.573791737272538e-06,
+      "loss": 0.7468,
+      "step": 2829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567749971588372e-06,
+      "loss": 0.7862,
+      "step": 2830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561708363979377e-06,
+      "loss": 0.853,
+      "step": 2831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555666916654987e-06,
+      "loss": 0.7847,
+      "step": 2832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54962563182458e-06,
+      "loss": 0.8278,
+      "step": 2833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.543584511697463e-06,
+      "loss": 0.7759,
+      "step": 2834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537543558482907e-06,
+      "loss": 0.8055,
+      "step": 2835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5315027743901e-06,
+      "loss": 0.8348,
+      "step": 2836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525462161628172e-06,
+      "loss": 0.2768,
+      "step": 2837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.5194217224062e-06,
+      "loss": 0.2744,
+      "step": 2838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513381458933187e-06,
+      "loss": 0.8384,
+      "step": 2839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507341373418078e-06,
+      "loss": 0.7868,
+      "step": 2840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.50130146806975e-06,
+      "loss": 0.79,
+      "step": 2841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.495261745097017e-06,
+      "loss": 0.7923,
+      "step": 2842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489222206708622e-06,
+      "loss": 0.8281,
+      "step": 2843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.483182855113242e-06,
+      "loss": 0.7981,
+      "step": 2844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47714369251949e-06,
+      "loss": 0.8134,
+      "step": 2845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.471104721135903e-06,
+      "loss": 0.8035,
+      "step": 2846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465065943170956e-06,
+      "loss": 0.7489,
+      "step": 2847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459027360833045e-06,
+      "loss": 0.8507,
+      "step": 2848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452988976330493e-06,
+      "loss": 0.8046,
+      "step": 2849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.446950791871568e-06,
+      "loss": 0.8104,
+      "step": 2850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.440912809664445e-06,
+      "loss": 0.7965,
+      "step": 2851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.434875031917235e-06,
+      "loss": 0.7861,
+      "step": 2852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42883746083797e-06,
+      "loss": 0.7861,
+      "step": 2853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.42280009863461e-06,
+      "loss": 0.8344,
+      "step": 2854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41676294751504e-06,
+      "loss": 0.8194,
+      "step": 2855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.410726009687059e-06,
+      "loss": 0.7307,
+      "step": 2856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404689287358397e-06,
+      "loss": 0.7399,
+      "step": 2857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.398652782736696e-06,
+      "loss": 0.8621,
+      "step": 2858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.392616498029536e-06,
+      "loss": 0.7403,
+      "step": 2859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386580435444398e-06,
+      "loss": 0.7629,
+      "step": 2860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.380544597188689e-06,
+      "loss": 0.7635,
+      "step": 2861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.374508985469735e-06,
+      "loss": 0.7279,
+      "step": 2862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.368473602494774e-06,
+      "loss": 0.8254,
+      "step": 2863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.362438450470969e-06,
+      "loss": 0.8066,
+      "step": 2864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356403531605393e-06,
+      "loss": 0.799,
+      "step": 2865
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.350368848105032e-06,
+      "loss": 0.2541,
+      "step": 2866
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34433440217679e-06,
+      "loss": 0.8277,
+      "step": 2867
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.338300196027478e-06,
+      "loss": 0.8446,
+      "step": 2868
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332266231863833e-06,
+      "loss": 0.817,
+      "step": 2869
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32623251189249e-06,
+      "loss": 0.8033,
+      "step": 2870
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.320199038319997e-06,
+      "loss": 0.7729,
+      "step": 2871
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314165813352817e-06,
+      "loss": 0.823,
+      "step": 2872
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308132839197314e-06,
+      "loss": 0.2774,
+      "step": 2873
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.302100118059771e-06,
+      "loss": 0.7665,
+      "step": 2874
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29606765214637e-06,
+      "loss": 0.8684,
+      "step": 2875
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.290035443663205e-06,
+      "loss": 0.267,
+      "step": 2876
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284003494816268e-06,
+      "loss": 0.8049,
+      "step": 2877
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277971807811462e-06,
+      "loss": 0.7728,
+      "step": 2878
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271940384854601e-06,
+      "loss": 0.7762,
+      "step": 2879
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.265909228151389e-06,
+      "loss": 0.8651,
+      "step": 2880
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.259878339907441e-06,
+      "loss": 0.8027,
+      "step": 2881
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.253847722328269e-06,
+      "loss": 0.8018,
+      "step": 2882
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.247817377619285e-06,
+      "loss": 0.7959,
+      "step": 2883
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.241787307985814e-06,
+      "loss": 0.7953,
+      "step": 2884
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.235757515633066e-06,
+      "loss": 0.8292,
+      "step": 2885
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229728002766154e-06,
+      "loss": 0.8185,
+      "step": 2886
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22369877159009e-06,
+      "loss": 0.8378,
+      "step": 2887
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217669824309786e-06,
+      "loss": 0.8658,
+      "step": 2888
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21164116313004e-06,
+      "loss": 0.7841,
+      "step": 2889
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.205612790255558e-06,
+      "loss": 0.8066,
+      "step": 2890
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.199584707890933e-06,
+      "loss": 0.7482,
+      "step": 2891
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193556918240646e-06,
+      "loss": 0.8523,
+      "step": 2892
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187529423509092e-06,
+      "loss": 0.7586,
+      "step": 2893
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181502225900535e-06,
+      "loss": 0.7501,
+      "step": 2894
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.175475327619143e-06,
+      "loss": 0.7792,
+      "step": 2895
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16944873086897e-06,
+      "loss": 0.7578,
+      "step": 2896
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16342243785396e-06,
+      "loss": 0.7542,
+      "step": 2897
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15739645077795e-06,
+      "loss": 0.8197,
+      "step": 2898
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.151370771844661e-06,
+      "loss": 0.8157,
+      "step": 2899
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1453454032577e-06,
+      "loss": 0.8142,
+      "step": 2900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139320347220567e-06,
+      "loss": 0.734,
+      "step": 2901
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.133295605936636e-06,
+      "loss": 0.8248,
+      "step": 2902
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.127271181609184e-06,
+      "loss": 0.8423,
+      "step": 2903
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121247076441356e-06,
+      "loss": 0.7792,
+      "step": 2904
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115223292636187e-06,
+      "loss": 0.8162,
+      "step": 2905
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.109199832396592e-06,
+      "loss": 0.7967,
+      "step": 2906
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.103176697925367e-06,
+      "loss": 0.8526,
+      "step": 2907
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.097153891425197e-06,
+      "loss": 0.8075,
+      "step": 2908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091131415098636e-06,
+      "loss": 0.7919,
+      "step": 2909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085109271148125e-06,
+      "loss": 0.76,
+      "step": 2910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07908746177598e-06,
+      "loss": 0.8589,
+      "step": 2911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.073065989184389e-06,
+      "loss": 0.8303,
+      "step": 2912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.067044855575434e-06,
+      "loss": 0.8356,
+      "step": 2913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061024063151055e-06,
+      "loss": 0.7973,
+      "step": 2914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.055003614113078e-06,
+      "loss": 0.8394,
+      "step": 2915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048983510663195e-06,
+      "loss": 0.8359,
+      "step": 2916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.042963755002982e-06,
+      "loss": 0.7717,
+      "step": 2917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03694434933388e-06,
+      "loss": 0.8377,
+      "step": 2918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.030925295857204e-06,
+      "loss": 0.8325,
+      "step": 2919
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024906596774141e-06,
+      "loss": 0.7602,
+      "step": 2920
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.018888254285743e-06,
+      "loss": 0.8055,
+      "step": 2921
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012870270592945e-06,
+      "loss": 0.7919,
+      "step": 2922
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006852647896539e-06,
+      "loss": 0.7933,
+      "step": 2923
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000835388397186e-06,
+      "loss": 0.7975,
+      "step": 2924
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.994818494295419e-06,
+      "loss": 0.7652,
+      "step": 2925
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.988801967791626e-06,
+      "loss": 0.8519,
+      "step": 2926
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982785811086084e-06,
+      "loss": 0.7046,
+      "step": 2927
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.97677002637891e-06,
+      "loss": 0.7742,
+      "step": 2928
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970754615870098e-06,
+      "loss": 0.7675,
+      "step": 2929
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.964739581759503e-06,
+      "loss": 0.7977,
+      "step": 2930
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95872492624684e-06,
+      "loss": 0.2674,
+      "step": 2931
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.952710651531687e-06,
+      "loss": 0.8173,
+      "step": 2932
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.946696759813486e-06,
+      "loss": 0.7476,
+      "step": 2933
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940683253291533e-06,
+      "loss": 0.7947,
+      "step": 2934
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934670134164985e-06,
+      "loss": 0.8057,
+      "step": 2935
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928657404632854e-06,
+      "loss": 0.8594,
+      "step": 2936
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.922645066894025e-06,
+      "loss": 0.7756,
+      "step": 2937
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916633123147222e-06,
+      "loss": 0.7568,
+      "step": 2938
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.910621575591032e-06,
+      "loss": 0.8061,
+      "step": 2939
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904610426423897e-06,
+      "loss": 0.7959,
+      "step": 2940
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89859967784411e-06,
+      "loss": 0.8435,
+      "step": 2941
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.892589332049826e-06,
+      "loss": 0.8069,
+      "step": 2942
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886579391239043e-06,
+      "loss": 0.7897,
+      "step": 2943
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880569857609615e-06,
+      "loss": 0.8235,
+      "step": 2944
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.874560733359249e-06,
+      "loss": 0.7685,
+      "step": 2945
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.868552020685494e-06,
+      "loss": 0.8008,
+      "step": 2946
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.862543721785763e-06,
+      "loss": 0.8273,
+      "step": 2947
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856535838857307e-06,
+      "loss": 0.7569,
+      "step": 2948
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.850528374097227e-06,
+      "loss": 0.8378,
+      "step": 2949
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84452132970247e-06,
+      "loss": 0.7804,
+      "step": 2950
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83851470786983e-06,
+      "loss": 0.825,
+      "step": 2951
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832508510795952e-06,
+      "loss": 0.8672,
+      "step": 2952
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.826502740677317e-06,
+      "loss": 0.8625,
+      "step": 2953
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.820497399710253e-06,
+      "loss": 0.8004,
+      "step": 2954
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81449249009093e-06,
+      "loss": 0.8349,
+      "step": 2955
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.808488014015369e-06,
+      "loss": 0.7691,
+      "step": 2956
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.80248397367942e-06,
+      "loss": 0.7814,
+      "step": 2957
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796480371278784e-06,
+      "loss": 0.8388,
+      "step": 2958
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.790477209008992e-06,
+      "loss": 0.7898,
+      "step": 2959
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784474489065418e-06,
+      "loss": 0.793,
+      "step": 2960
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778472213643279e-06,
+      "loss": 0.8154,
+      "step": 2961
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.772470384937625e-06,
+      "loss": 0.8153,
+      "step": 2962
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.766469005143343e-06,
+      "loss": 0.7648,
+      "step": 2963
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760468076455157e-06,
+      "loss": 0.7702,
+      "step": 2964
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.754467601067617e-06,
+      "loss": 0.8237,
+      "step": 2965
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74846758117513e-06,
+      "loss": 0.7724,
+      "step": 2966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.742468018971914e-06,
+      "loss": 0.8335,
+      "step": 2967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.736468916652026e-06,
+      "loss": 0.7743,
+      "step": 2968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730470276409358e-06,
+      "loss": 0.8202,
+      "step": 2969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.724472100437627e-06,
+      "loss": 0.8152,
+      "step": 2970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.718474390930391e-06,
+      "loss": 0.7506,
+      "step": 2971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.71247715008103e-06,
+      "loss": 0.779,
+      "step": 2972
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706480380082753e-06,
+      "loss": 0.8426,
+      "step": 2973
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.700484083128595e-06,
+      "loss": 0.7844,
+      "step": 2974
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.694488261411418e-06,
+      "loss": 0.7629,
+      "step": 2975
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68849291712392e-06,
+      "loss": 0.8166,
+      "step": 2976
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682498052458614e-06,
+      "loss": 0.7683,
+      "step": 2977
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676503669607838e-06,
+      "loss": 0.7394,
+      "step": 2978
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.670509770763755e-06,
+      "loss": 0.7479,
+      "step": 2979
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66451635811836e-06,
+      "loss": 0.8127,
+      "step": 2980
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.658523433863455e-06,
+      "loss": 0.3071,
+      "step": 2981
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652531000190674e-06,
+      "loss": 0.7953,
+      "step": 2982
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64653905929147e-06,
+      "loss": 0.7929,
+      "step": 2983
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64054761335711e-06,
+      "loss": 0.821,
+      "step": 2984
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634556664578688e-06,
+      "loss": 0.7574,
+      "step": 2985
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628566215147112e-06,
+      "loss": 0.7659,
+      "step": 2986
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.622576267253107e-06,
+      "loss": 0.7485,
+      "step": 2987
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616586823087216e-06,
+      "loss": 0.8377,
+      "step": 2988
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61059788483979e-06,
+      "loss": 0.2617,
+      "step": 2989
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.604609454701015e-06,
+      "loss": 0.7688,
+      "step": 2990
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.598621534860872e-06,
+      "loss": 0.7978,
+      "step": 2991
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592634127509163e-06,
+      "loss": 0.8093,
+      "step": 2992
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5866472348355e-06,
+      "loss": 0.8026,
+      "step": 2993
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.580660859029307e-06,
+      "loss": 0.8208,
+      "step": 2994
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574675002279822e-06,
+      "loss": 0.7779,
+      "step": 2995
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.568689666776092e-06,
+      "loss": 0.7557,
+      "step": 2996
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.562704854706974e-06,
+      "loss": 0.8052,
+      "step": 2997
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556720568261128e-06,
+      "loss": 0.7554,
+      "step": 2998
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550736809627023e-06,
+      "loss": 0.7408,
+      "step": 2999
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.544753580992948e-06,
+      "loss": 0.7974,
+      "step": 3000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.538770884546984e-06,
+      "loss": 0.8182,
+      "step": 3001
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53278872247702e-06,
+      "loss": 0.7422,
+      "step": 3002
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526807096970753e-06,
+      "loss": 0.8008,
+      "step": 3003
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.520826010215677e-06,
+      "loss": 0.735,
+      "step": 3004
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.514845464399102e-06,
+      "loss": 0.8155,
+      "step": 3005
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.508865461708125e-06,
+      "loss": 0.8495,
+      "step": 3006
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502886004329658e-06,
+      "loss": 0.8408,
+      "step": 3007
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4969070944504e-06,
+      "loss": 0.7723,
+      "step": 3008
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49092873425686e-06,
+      "loss": 0.8374,
+      "step": 3009
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.484950925935344e-06,
+      "loss": 0.7899,
+      "step": 3010
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478973671671956e-06,
+      "loss": 0.7706,
+      "step": 3011
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472996973652591e-06,
+      "loss": 0.8361,
+      "step": 3012
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467020834062947e-06,
+      "loss": 0.8001,
+      "step": 3013
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461045255088523e-06,
+      "loss": 0.8395,
+      "step": 3014
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.455070238914602e-06,
+      "loss": 0.7931,
+      "step": 3015
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449095787726267e-06,
+      "loss": 0.7775,
+      "step": 3016
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.443121903708392e-06,
+      "loss": 0.7628,
+      "step": 3017
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437148589045643e-06,
+      "loss": 0.8023,
+      "step": 3018
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.431175845922485e-06,
+      "loss": 0.7621,
+      "step": 3019
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.425203676523167e-06,
+      "loss": 0.7959,
+      "step": 3020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.419232083031728e-06,
+      "loss": 0.7903,
+      "step": 3021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413261067631997e-06,
+      "loss": 0.8148,
+      "step": 3022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.407290632507592e-06,
+      "loss": 0.8007,
+      "step": 3023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401320779841927e-06,
+      "loss": 0.8229,
+      "step": 3024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.395351511818188e-06,
+      "loss": 0.2508,
+      "step": 3025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.389382830619358e-06,
+      "loss": 0.2818,
+      "step": 3026
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383414738428204e-06,
+      "loss": 0.8481,
+      "step": 3027
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37744723742727e-06,
+      "loss": 0.7603,
+      "step": 3028
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.371480329798893e-06,
+      "loss": 0.8379,
+      "step": 3029
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.36551401772519e-06,
+      "loss": 0.7659,
+      "step": 3030
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359548303388062e-06,
+      "loss": 0.8084,
+      "step": 3031
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353583188969184e-06,
+      "loss": 0.8434,
+      "step": 3032
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.347618676650016e-06,
+      "loss": 0.8017,
+      "step": 3033
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.341654768611804e-06,
+      "loss": 0.8252,
+      "step": 3034
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.335691467035567e-06,
+      "loss": 0.7573,
+      "step": 3035
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3297287741021e-06,
+      "loss": 0.8518,
+      "step": 3036
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32376669199198e-06,
+      "loss": 0.808,
+      "step": 3037
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.317805222885557e-06,
+      "loss": 0.777,
+      "step": 3038
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31184436896296e-06,
+      "loss": 0.8141,
+      "step": 3039
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305884132404091e-06,
+      "loss": 0.8517,
+      "step": 3040
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29992451538863e-06,
+      "loss": 0.7446,
+      "step": 3041
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.293965520096019e-06,
+      "loss": 0.819,
+      "step": 3042
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.28800714870549e-06,
+      "loss": 0.7998,
+      "step": 3043
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.282049403396035e-06,
+      "loss": 0.7844,
+      "step": 3044
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.276092286346422e-06,
+      "loss": 0.2701,
+      "step": 3045
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270135799735185e-06,
+      "loss": 0.8365,
+      "step": 3046
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264179945740621e-06,
+      "loss": 0.7855,
+      "step": 3047
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258224726540817e-06,
+      "loss": 0.7606,
+      "step": 3048
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.252270144313611e-06,
+      "loss": 0.7868,
+      "step": 3049
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246316201236612e-06,
+      "loss": 0.7944,
+      "step": 3050
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240362899487195e-06,
+      "loss": 0.2696,
+      "step": 3051
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234410241242496e-06,
+      "loss": 0.7847,
+      "step": 3052
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22845822867943e-06,
+      "loss": 0.7971,
+      "step": 3053
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.222506863974661e-06,
+      "loss": 0.8059,
+      "step": 3054
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21655614930462e-06,
+      "loss": 0.8441,
+      "step": 3055
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210606086845506e-06,
+      "loss": 0.7837,
+      "step": 3056
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204656678773268e-06,
+      "loss": 0.7909,
+      "step": 3057
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.198707927263633e-06,
+      "loss": 0.8348,
+      "step": 3058
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19275983449207e-06,
+      "loss": 0.8475,
+      "step": 3059
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18681240263382e-06,
+      "loss": 0.7239,
+      "step": 3060
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.180865633863875e-06,
+      "loss": 0.8303,
+      "step": 3061
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174919530356983e-06,
+      "loss": 0.8138,
+      "step": 3062
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168974094287659e-06,
+      "loss": 0.7706,
+      "step": 3063
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163029327830162e-06,
+      "loss": 0.8504,
+      "step": 3064
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.157085233158517e-06,
+      "loss": 0.2751,
+      "step": 3065
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15114181244649e-06,
+      "loss": 0.8159,
+      "step": 3066
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.145199067867612e-06,
+      "loss": 0.842,
+      "step": 3067
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139257001595167e-06,
+      "loss": 0.8359,
+      "step": 3068
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.133315615802184e-06,
+      "loss": 0.7741,
+      "step": 3069
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.127374912661446e-06,
+      "loss": 0.8397,
+      "step": 3070
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121434894345487e-06,
+      "loss": 0.7766,
+      "step": 3071
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115495563026586e-06,
+      "loss": 0.8424,
+      "step": 3072
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10955692087678e-06,
+      "loss": 0.8286,
+      "step": 3073
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.10361897006785e-06,
+      "loss": 0.8129,
+      "step": 3074
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.097681712771317e-06,
+      "loss": 0.8052,
+      "step": 3075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.091745151158454e-06,
+      "loss": 0.8487,
+      "step": 3076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085809287400288e-06,
+      "loss": 0.8228,
+      "step": 3077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.079874123667577e-06,
+      "loss": 0.7631,
+      "step": 3078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.073939662130831e-06,
+      "loss": 0.8374,
+      "step": 3079
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068005904960299e-06,
+      "loss": 0.8371,
+      "step": 3080
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062072854325972e-06,
+      "loss": 0.8472,
+      "step": 3081
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05614051239759e-06,
+      "loss": 0.8385,
+      "step": 3082
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.050208881344627e-06,
+      "loss": 0.7601,
+      "step": 3083
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.0442779633363e-06,
+      "loss": 0.7871,
+      "step": 3084
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.03834776054156e-06,
+      "loss": 0.8218,
+      "step": 3085
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.032418275129099e-06,
+      "loss": 0.8246,
+      "step": 3086
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026489509267358e-06,
+      "loss": 0.2635,
+      "step": 3087
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.020561465124498e-06,
+      "loss": 0.8284,
+      "step": 3088
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014634144868427e-06,
+      "loss": 0.8042,
+      "step": 3089
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008707550666777e-06,
+      "loss": 0.8204,
+      "step": 3090
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002781684686925e-06,
+      "loss": 0.7915,
+      "step": 3091
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.996856549095982e-06,
+      "loss": 0.8086,
+      "step": 3092
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990932146060785e-06,
+      "loss": 0.7708,
+      "step": 3093
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98500847774791e-06,
+      "loss": 0.8049,
+      "step": 3094
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.979085546323657e-06,
+      "loss": 0.7658,
+      "step": 3095
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973163353954061e-06,
+      "loss": 0.7925,
+      "step": 3096
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967241902804887e-06,
+      "loss": 0.8449,
+      "step": 3097
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.961321195041629e-06,
+      "loss": 0.2757,
+      "step": 3098
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.955401232829506e-06,
+      "loss": 0.8235,
+      "step": 3099
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.949482018333467e-06,
+      "loss": 0.7888,
+      "step": 3100
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.943563553718186e-06,
+      "loss": 0.7736,
+      "step": 3101
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937645841148065e-06,
+      "loss": 0.8466,
+      "step": 3102
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.931728882787233e-06,
+      "loss": 0.787,
+      "step": 3103
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.925812680799535e-06,
+      "loss": 0.8436,
+      "step": 3104
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.919897237348546e-06,
+      "loss": 0.8384,
+      "step": 3105
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913982554597563e-06,
+      "loss": 0.7765,
+      "step": 3106
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908068634709603e-06,
+      "loss": 0.7219,
+      "step": 3107
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.902155479847404e-06,
+      "loss": 0.703,
+      "step": 3108
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.896243092173427e-06,
+      "loss": 0.793,
+      "step": 3109
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890331473849844e-06,
+      "loss": 0.7697,
+      "step": 3110
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884420627038562e-06,
+      "loss": 0.7697,
+      "step": 3111
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87851055390119e-06,
+      "loss": 0.7645,
+      "step": 3112
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.872601256599062e-06,
+      "loss": 0.8008,
+      "step": 3113
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.866692737293224e-06,
+      "loss": 0.8022,
+      "step": 3114
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86078499814444e-06,
+      "loss": 0.8446,
+      "step": 3115
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.854878041313192e-06,
+      "loss": 0.7905,
+      "step": 3116
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84897186895967e-06,
+      "loss": 0.832,
+      "step": 3117
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.843066483243779e-06,
+      "loss": 0.8104,
+      "step": 3118
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.837161886325137e-06,
+      "loss": 0.7772,
+      "step": 3119
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.831258080363068e-06,
+      "loss": 0.7669,
+      "step": 3120
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.825355067516625e-06,
+      "loss": 0.8407,
+      "step": 3121
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.819452849944549e-06,
+      "loss": 0.8311,
+      "step": 3122
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.8135514298053e-06,
+      "loss": 0.801,
+      "step": 3123
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.807650809257048e-06,
+      "loss": 0.8032,
+      "step": 3124
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.801750990457662e-06,
+      "loss": 0.7749,
+      "step": 3125
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795851975564734e-06,
+      "loss": 0.7866,
+      "step": 3126
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789953766735543e-06,
+      "loss": 0.7817,
+      "step": 3127
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.784056366127088e-06,
+      "loss": 0.2625,
+      "step": 3128
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.778159775896063e-06,
+      "loss": 0.7926,
+      "step": 3129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772263998198867e-06,
+      "loss": 0.8324,
+      "step": 3130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766369035191614e-06,
+      "loss": 0.8127,
+      "step": 3131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760474889030103e-06,
+      "loss": 0.8229,
+      "step": 3132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75458156186984e-06,
+      "loss": 0.7667,
+      "step": 3133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748689055866037e-06,
+      "loss": 0.761,
+      "step": 3134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742797373173594e-06,
+      "loss": 0.8127,
+      "step": 3135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73690651594713e-06,
+      "loss": 0.8096,
+      "step": 3136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.731016486340941e-06,
+      "loss": 0.7996,
+      "step": 3137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72512728650903e-06,
+      "loss": 0.7735,
+      "step": 3138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719238918605098e-06,
+      "loss": 0.8105,
+      "step": 3139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713351384782539e-06,
+      "loss": 0.8556,
+      "step": 3140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70746468719444e-06,
+      "loss": 0.8649,
+      "step": 3141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.701578827993587e-06,
+      "loss": 0.8045,
+      "step": 3142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695693809332455e-06,
+      "loss": 0.7788,
+      "step": 3143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68980963336321e-06,
+      "loss": 0.8177,
+      "step": 3144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.683926302237722e-06,
+      "loss": 0.8434,
+      "step": 3145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67804381810754e-06,
+      "loss": 0.7812,
+      "step": 3146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672162183123905e-06,
+      "loss": 0.8094,
+      "step": 3147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.666281399437749e-06,
+      "loss": 0.8203,
+      "step": 3148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660401469199693e-06,
+      "loss": 0.783,
+      "step": 3149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.654522394560045e-06,
+      "loss": 0.8488,
+      "step": 3150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648644177668803e-06,
+      "loss": 0.7946,
+      "step": 3151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64276682067565e-06,
+      "loss": 0.8554,
+      "step": 3152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63689032572995e-06,
+      "loss": 0.837,
+      "step": 3153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.631014694980752e-06,
+      "loss": 0.8064,
+      "step": 3154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625139930576801e-06,
+      "loss": 0.7563,
+      "step": 3155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6192660346665136e-06,
+      "loss": 0.8095,
+      "step": 3156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6133930093979905e-06,
+      "loss": 0.7869,
+      "step": 3157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607520856919015e-06,
+      "loss": 0.823,
+      "step": 3158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.60164957937705e-06,
+      "loss": 0.7962,
+      "step": 3159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.595779178919244e-06,
+      "loss": 0.7548,
+      "step": 3160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.589909657692417e-06,
+      "loss": 0.7524,
+      "step": 3161
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.584041017843074e-06,
+      "loss": 0.8458,
+      "step": 3162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.578173261517392e-06,
+      "loss": 0.7684,
+      "step": 3163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572306390861224e-06,
+      "loss": 0.256,
+      "step": 3164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.566440408020111e-06,
+      "loss": 0.2551,
+      "step": 3165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.560575315139258e-06,
+      "loss": 0.8202,
+      "step": 3166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5547111143635475e-06,
+      "loss": 0.7568,
+      "step": 3167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5488478078375326e-06,
+      "loss": 0.7944,
+      "step": 3168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.542985397705448e-06,
+      "loss": 0.7407,
+      "step": 3169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5371238861111915e-06,
+      "loss": 0.8004,
+      "step": 3170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.531263275198337e-06,
+      "loss": 0.7993,
+      "step": 3171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525403567110128e-06,
+      "loss": 0.7865,
+      "step": 3172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.519544763989474e-06,
+      "loss": 0.2477,
+      "step": 3173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513686867978967e-06,
+      "loss": 0.8386,
+      "step": 3174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507829881220851e-06,
+      "loss": 0.8489,
+      "step": 3175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.501973805857043e-06,
+      "loss": 0.7569,
+      "step": 3176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4961186440291316e-06,
+      "loss": 0.8173,
+      "step": 3177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.490264397878359e-06,
+      "loss": 0.2683,
+      "step": 3178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484411069545655e-06,
+      "loss": 0.7432,
+      "step": 3179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.478558661171591e-06,
+      "loss": 0.8075,
+      "step": 3180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4727071748964146e-06,
+      "loss": 0.8004,
+      "step": 3181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466856612860028e-06,
+      "loss": 0.775,
+      "step": 3182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.461006977202001e-06,
+      "loss": 0.7899,
+      "step": 3183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455158270061567e-06,
+      "loss": 0.8521,
+      "step": 3184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449310493577615e-06,
+      "loss": 0.3049,
+      "step": 3185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4434636498886914e-06,
+      "loss": 0.2598,
+      "step": 3186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43761774113301e-06,
+      "loss": 0.7828,
+      "step": 3187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.431772769448429e-06,
+      "loss": 0.8032,
+      "step": 3188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.425928736972483e-06,
+      "loss": 0.7886,
+      "step": 3189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.420085645842348e-06,
+      "loss": 0.7553,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.414243498194862e-06,
+      "loss": 0.78,
+      "step": 3191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408402296166515e-06,
+      "loss": 0.2739,
+      "step": 3192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402562041893451e-06,
+      "loss": 0.2649,
+      "step": 3193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.396722737511473e-06,
+      "loss": 0.7724,
+      "step": 3194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.39088438515603e-06,
+      "loss": 0.8866,
+      "step": 3195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3850469869622265e-06,
+      "loss": 0.8013,
+      "step": 3196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379210545064817e-06,
+      "loss": 0.7979,
+      "step": 3197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.373375061598201e-06,
+      "loss": 0.812,
+      "step": 3198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.367540538696443e-06,
+      "loss": 0.82,
+      "step": 3199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.361706978493241e-06,
+      "loss": 0.7598,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.355874383121945e-06,
+      "loss": 0.8389,
+      "step": 3201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350042754715552e-06,
+      "loss": 0.846,
+      "step": 3202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344212095406711e-06,
+      "loss": 0.8298,
+      "step": 3203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.338382407327709e-06,
+      "loss": 0.8187,
+      "step": 3204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332553692610482e-06,
+      "loss": 0.7428,
+      "step": 3205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326725953386609e-06,
+      "loss": 0.7203,
+      "step": 3206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320899191787308e-06,
+      "loss": 0.7279,
+      "step": 3207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.31507340994345e-06,
+      "loss": 0.7628,
+      "step": 3208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.30924860998554e-06,
+      "loss": 0.7957,
+      "step": 3209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.303424794043723e-06,
+      "loss": 0.7653,
+      "step": 3210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297601964247791e-06,
+      "loss": 0.2401,
+      "step": 3211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.291780122727162e-06,
+      "loss": 0.8133,
+      "step": 3212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.285959271610909e-06,
+      "loss": 0.808,
+      "step": 3213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280139413027735e-06,
+      "loss": 0.7744,
+      "step": 3214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274320549105977e-06,
+      "loss": 0.7478,
+      "step": 3215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.268502681973615e-06,
+      "loss": 0.8405,
+      "step": 3216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2626858137582555e-06,
+      "loss": 0.7635,
+      "step": 3217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2568699465871505e-06,
+      "loss": 0.7939,
+      "step": 3218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2510550825871796e-06,
+      "loss": 0.736,
+      "step": 3219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.245241223884855e-06,
+      "loss": 0.7651,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2394283726063245e-06,
+      "loss": 0.8338,
+      "step": 3221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2336165308773585e-06,
+      "loss": 0.8147,
+      "step": 3222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227805700823378e-06,
+      "loss": 0.7837,
+      "step": 3223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221995884569415e-06,
+      "loss": 0.7256,
+      "step": 3224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.216187084240136e-06,
+      "loss": 0.7884,
+      "step": 3225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.21037930195984e-06,
+      "loss": 0.8229,
+      "step": 3226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2045725398524455e-06,
+      "loss": 0.8008,
+      "step": 3227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1987668000415125e-06,
+      "loss": 0.7784,
+      "step": 3228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.19296208465021e-06,
+      "loss": 0.8256,
+      "step": 3229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.187158395801344e-06,
+      "loss": 0.7454,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181355735617338e-06,
+      "loss": 0.7774,
+      "step": 3231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.175554106220249e-06,
+      "loss": 0.7838,
+      "step": 3232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.169753509731746e-06,
+      "loss": 0.7774,
+      "step": 3233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.163953948273127e-06,
+      "loss": 0.8204,
+      "step": 3234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158155423965309e-06,
+      "loss": 0.7568,
+      "step": 3235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1523579389288275e-06,
+      "loss": 0.7782,
+      "step": 3236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.146561495283846e-06,
+      "loss": 0.7604,
+      "step": 3237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.140766095150138e-06,
+      "loss": 0.2737,
+      "step": 3238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.134971740647101e-06,
+      "loss": 0.7864,
+      "step": 3239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.129178433893749e-06,
+      "loss": 0.7904,
+      "step": 3240
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1233861770087045e-06,
+      "loss": 0.7477,
+      "step": 3241
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117594972110225e-06,
+      "loss": 0.8177,
+      "step": 3242
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111804821316166e-06,
+      "loss": 0.7711,
+      "step": 3243
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1060157267440065e-06,
+      "loss": 0.2509,
+      "step": 3244
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.100227690510833e-06,
+      "loss": 0.812,
+      "step": 3245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.094440714733348e-06,
+      "loss": 0.7672,
+      "step": 3246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088654801527871e-06,
+      "loss": 0.2858,
+      "step": 3247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0828699530103255e-06,
+      "loss": 0.7936,
+      "step": 3248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.077086171296246e-06,
+      "loss": 0.7174,
+      "step": 3249
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.071303458500784e-06,
+      "loss": 0.8274,
+      "step": 3250
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065521816738687e-06,
+      "loss": 0.772,
+      "step": 3251
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.05974124812433e-06,
+      "loss": 0.75,
+      "step": 3252
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053961754771681e-06,
+      "loss": 0.7697,
+      "step": 3253
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.048183338794319e-06,
+      "loss": 0.7939,
+      "step": 3254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042406002305427e-06,
+      "loss": 0.7726,
+      "step": 3255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.036629747417788e-06,
+      "loss": 0.8278,
+      "step": 3256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0308545762438086e-06,
+      "loss": 0.7742,
+      "step": 3257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.025080490895482e-06,
+      "loss": 0.8222,
+      "step": 3258
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019307493484409e-06,
+      "loss": 0.7728,
+      "step": 3259
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013535586121791e-06,
+      "loss": 0.8606,
+      "step": 3260
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00776477091843e-06,
+      "loss": 0.8479,
+      "step": 3261
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00199504998474e-06,
+      "loss": 0.829,
+      "step": 3262
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996226425430717e-06,
+      "loss": 0.8308,
+      "step": 3263
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.990458899365969e-06,
+      "loss": 0.8054,
+      "step": 3264
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.984692473899693e-06,
+      "loss": 0.7886,
+      "step": 3265
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978927151140697e-06,
+      "loss": 0.7874,
+      "step": 3266
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.973162933197374e-06,
+      "loss": 0.8253,
+      "step": 3267
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967399822177714e-06,
+      "loss": 0.8066,
+      "step": 3268
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.961637820189307e-06,
+      "loss": 0.7647,
+      "step": 3269
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.955876929339329e-06,
+      "loss": 0.7909,
+      "step": 3270
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950117151734565e-06,
+      "loss": 0.7775,
+      "step": 3271
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.944358489481378e-06,
+      "loss": 0.7824,
+      "step": 3272
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.938600944685729e-06,
+      "loss": 0.8127,
+      "step": 3273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93284451945317e-06,
+      "loss": 0.2678,
+      "step": 3274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9270892158888405e-06,
+      "loss": 0.7686,
+      "step": 3275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92133503609748e-06,
+      "loss": 0.8061,
+      "step": 3276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915581982183406e-06,
+      "loss": 0.8291,
+      "step": 3277
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 0.8305,
+      "step": 3278
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.904079260402341e-06,
+      "loss": 0.7633,
+      "step": 3279
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8983295967419304e-06,
+      "loss": 0.8433,
+      "step": 3280
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8925810673719685e-06,
+      "loss": 0.8368,
+      "step": 3281
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886833674394709e-06,
+      "loss": 0.7214,
+      "step": 3282
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88108741991199e-06,
+      "loss": 0.7661,
+      "step": 3283
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.875342306025233e-06,
+      "loss": 0.7886,
+      "step": 3284
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869598334835443e-06,
+      "loss": 0.8129,
+      "step": 3285
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8638555084432155e-06,
+      "loss": 0.7644,
+      "step": 3286
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858113828948714e-06,
+      "loss": 0.8097,
+      "step": 3287
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8523732984516915e-06,
+      "loss": 0.7753,
+      "step": 3288
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.846633919051474e-06,
+      "loss": 0.7895,
+      "step": 3289
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84089569284697e-06,
+      "loss": 0.8103,
+      "step": 3290
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.835158621936672e-06,
+      "loss": 0.2554,
+      "step": 3291
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82942270841864e-06,
+      "loss": 0.7939,
+      "step": 3292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.823687954390517e-06,
+      "loss": 0.8236,
+      "step": 3293
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.817954361949516e-06,
+      "loss": 0.847,
+      "step": 3294
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812221933192437e-06,
+      "loss": 0.2462,
+      "step": 3295
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.806490670215644e-06,
+      "loss": 0.7558,
+      "step": 3296
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.800760575115076e-06,
+      "loss": 0.8029,
+      "step": 3297
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.795031649986248e-06,
+      "loss": 0.7742,
+      "step": 3298
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789303896924238e-06,
+      "loss": 0.8248,
+      "step": 3299
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.783577318023717e-06,
+      "loss": 0.7508,
+      "step": 3300
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.777851915378903e-06,
+      "loss": 0.8218,
+      "step": 3301
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772127691083597e-06,
+      "loss": 0.8546,
+      "step": 3302
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766404647231164e-06,
+      "loss": 0.2778,
+      "step": 3303
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7606827859145364e-06,
+      "loss": 0.7377,
+      "step": 3304
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7549621092262215e-06,
+      "loss": 0.8158,
+      "step": 3305
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.749242619258287e-06,
+      "loss": 0.7456,
+      "step": 3306
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743524318102366e-06,
+      "loss": 0.7683,
+      "step": 3307
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73780720784966e-06,
+      "loss": 0.7633,
+      "step": 3308
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7320912905909315e-06,
+      "loss": 0.8414,
+      "step": 3309
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.726376568416515e-06,
+      "loss": 0.8038,
+      "step": 3310
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720663043416299e-06,
+      "loss": 0.7769,
+      "step": 3311
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714950717679735e-06,
+      "loss": 0.7997,
+      "step": 3312
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.709239593295842e-06,
+      "loss": 0.814,
+      "step": 3313
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.703529672353193e-06,
+      "loss": 0.8415,
+      "step": 3314
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697820956939924e-06,
+      "loss": 0.8417,
+      "step": 3315
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.692113449143731e-06,
+      "loss": 0.7674,
+      "step": 3316
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686407151051868e-06,
+      "loss": 0.8015,
+      "step": 3317
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.680702064751143e-06,
+      "loss": 0.7626,
+      "step": 3318
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.674998192327923e-06,
+      "loss": 0.7197,
+      "step": 3319
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669295535868137e-06,
+      "loss": 0.8027,
+      "step": 3320
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.663594097457261e-06,
+      "loss": 0.8021,
+      "step": 3321
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.657893879180328e-06,
+      "loss": 0.8259,
+      "step": 3322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652194883121928e-06,
+      "loss": 0.8386,
+      "step": 3323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.646497111366196e-06,
+      "loss": 0.8449,
+      "step": 3324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.640800565996831e-06,
+      "loss": 0.8124,
+      "step": 3325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6351052490970735e-06,
+      "loss": 0.8391,
+      "step": 3326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629411162749722e-06,
+      "loss": 0.7362,
+      "step": 3327
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623718309037113e-06,
+      "loss": 0.7291,
+      "step": 3328
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618026690041151e-06,
+      "loss": 0.7846,
+      "step": 3329
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6123363078432745e-06,
+      "loss": 0.7858,
+      "step": 3330
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6066471645244735e-06,
+      "loss": 0.791,
+      "step": 3331
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.600959262165285e-06,
+      "loss": 0.8003,
+      "step": 3332
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.595272602845793e-06,
+      "loss": 0.8294,
+      "step": 3333
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5895871886456255e-06,
+      "loss": 0.8086,
+      "step": 3334
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.583903021643958e-06,
+      "loss": 0.777,
+      "step": 3335
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.578220103919508e-06,
+      "loss": 0.8343,
+      "step": 3336
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.572538437550533e-06,
+      "loss": 0.7422,
+      "step": 3337
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.566858024614835e-06,
+      "loss": 0.8249,
+      "step": 3338
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561178867189767e-06,
+      "loss": 0.8281,
+      "step": 3339
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5555009673522064e-06,
+      "loss": 0.2608,
+      "step": 3340
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549824327178582e-06,
+      "loss": 0.7961,
+      "step": 3341
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.544148948744856e-06,
+      "loss": 0.7782,
+      "step": 3342
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.538474834126532e-06,
+      "loss": 0.7203,
+      "step": 3343
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532801985398654e-06,
+      "loss": 0.7715,
+      "step": 3344
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.527130404635799e-06,
+      "loss": 0.7834,
+      "step": 3345
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521460093912084e-06,
+      "loss": 0.7958,
+      "step": 3346
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.515791055301156e-06,
+      "loss": 0.7456,
+      "step": 3347
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.510123290876199e-06,
+      "loss": 0.7699,
+      "step": 3348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5044568027099355e-06,
+      "loss": 0.7397,
+      "step": 3349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498791592874616e-06,
+      "loss": 0.7557,
+      "step": 3350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.493127663442025e-06,
+      "loss": 0.7831,
+      "step": 3351
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487465016483478e-06,
+      "loss": 0.8119,
+      "step": 3352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.48180365406982e-06,
+      "loss": 0.7475,
+      "step": 3353
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.476143578271438e-06,
+      "loss": 0.7656,
+      "step": 3354
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470484791158231e-06,
+      "loss": 0.7811,
+      "step": 3355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46482729479964e-06,
+      "loss": 0.7518,
+      "step": 3356
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4591710912646235e-06,
+      "loss": 0.8034,
+      "step": 3357
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453516182621677e-06,
+      "loss": 0.8119,
+      "step": 3358
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447862570938817e-06,
+      "loss": 0.2656,
+      "step": 3359
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4422102582835854e-06,
+      "loss": 0.269,
+      "step": 3360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.436559246723053e-06,
+      "loss": 0.8018,
+      "step": 3361
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.430909538323805e-06,
+      "loss": 0.2695,
+      "step": 3362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.425261135151968e-06,
+      "loss": 0.787,
+      "step": 3363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.419614039273173e-06,
+      "loss": 0.7962,
+      "step": 3364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.413968252752585e-06,
+      "loss": 0.7902,
+      "step": 3365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.408323777654881e-06,
+      "loss": 0.7548,
+      "step": 3366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.402680616044265e-06,
+      "loss": 0.7889,
+      "step": 3367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397038769984458e-06,
+      "loss": 0.8339,
+      "step": 3368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.391398241538705e-06,
+      "loss": 0.809,
+      "step": 3369
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3857590327697585e-06,
+      "loss": 0.2828,
+      "step": 3370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.380121145739897e-06,
+      "loss": 0.7733,
+      "step": 3371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.374484582510909e-06,
+      "loss": 0.7933,
+      "step": 3372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368849345144112e-06,
+      "loss": 0.763,
+      "step": 3373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3632154357003265e-06,
+      "loss": 0.8317,
+      "step": 3374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357582856239888e-06,
+      "loss": 0.7974,
+      "step": 3375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35195160882265e-06,
+      "loss": 0.7998,
+      "step": 3376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.346321695507971e-06,
+      "loss": 0.7785,
+      "step": 3377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340693118354737e-06,
+      "loss": 0.7653,
+      "step": 3378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.335065879421334e-06,
+      "loss": 0.819,
+      "step": 3379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329439980765657e-06,
+      "loss": 0.8062,
+      "step": 3380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323815424445117e-06,
+      "loss": 0.8286,
+      "step": 3381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3181922125166275e-06,
+      "loss": 0.7969,
+      "step": 3382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.31257034703662e-06,
+      "loss": 0.8436,
+      "step": 3383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3069498300610264e-06,
+      "loss": 0.7637,
+      "step": 3384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3013306636452876e-06,
+      "loss": 0.7661,
+      "step": 3385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295712849844349e-06,
+      "loss": 0.7476,
+      "step": 3386
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290096390712658e-06,
+      "loss": 0.8105,
+      "step": 3387
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.284481288304179e-06,
+      "loss": 0.7872,
+      "step": 3388
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.27886754467237e-06,
+      "loss": 0.2604,
+      "step": 3389
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273255161870193e-06,
+      "loss": 0.2591,
+      "step": 3390
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267644141950113e-06,
+      "loss": 0.7302,
+      "step": 3391
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.262034486964099e-06,
+      "loss": 0.77,
+      "step": 3392
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25642619896362e-06,
+      "loss": 0.7717,
+      "step": 3393
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250819279999642e-06,
+      "loss": 0.7905,
+      "step": 3394
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.245213732122633e-06,
+      "loss": 0.8063,
+      "step": 3395
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.239609557382556e-06,
+      "loss": 0.2857,
+      "step": 3396
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.234006757828883e-06,
+      "loss": 0.7936,
+      "step": 3397
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22840533551057e-06,
+      "loss": 0.8038,
+      "step": 3398
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.222805292476075e-06,
+      "loss": 0.7945,
+      "step": 3399
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217206630773351e-06,
+      "loss": 0.7803,
+      "step": 3400
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.211609352449846e-06,
+      "loss": 0.8174,
+      "step": 3401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206013459552501e-06,
+      "loss": 0.7934,
+      "step": 3402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200418954127754e-06,
+      "loss": 0.8125,
+      "step": 3403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1948258382215325e-06,
+      "loss": 0.7921,
+      "step": 3404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.189234113879252e-06,
+      "loss": 0.7726,
+      "step": 3405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183643783145824e-06,
+      "loss": 0.785,
+      "step": 3406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.178054848065655e-06,
+      "loss": 0.7366,
+      "step": 3407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.172467310682635e-06,
+      "loss": 0.8189,
+      "step": 3408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1668811730401405e-06,
+      "loss": 0.7563,
+      "step": 3409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.16129643718104e-06,
+      "loss": 0.7777,
+      "step": 3410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155713105147687e-06,
+      "loss": 0.7927,
+      "step": 3411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.150131178981928e-06,
+      "loss": 0.7912,
+      "step": 3412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.144550660725087e-06,
+      "loss": 0.8288,
+      "step": 3413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138971552417978e-06,
+      "loss": 0.7946,
+      "step": 3414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133393856100898e-06,
+      "loss": 0.7522,
+      "step": 3415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1278175738136236e-06,
+      "loss": 0.7874,
+      "step": 3416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.122242707595427e-06,
+      "loss": 0.7436,
+      "step": 3417
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.116669259485051e-06,
+      "loss": 0.7502,
+      "step": 3418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111097231520718e-06,
+      "loss": 0.8555,
+      "step": 3419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105526625740137e-06,
+      "loss": 0.7945,
+      "step": 3420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099957444180501e-06,
+      "loss": 0.765,
+      "step": 3421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0943896888784734e-06,
+      "loss": 0.77,
+      "step": 3422
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0888233618702e-06,
+      "loss": 0.7794,
+      "step": 3423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.083258465191304e-06,
+      "loss": 0.7272,
+      "step": 3424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077695000876882e-06,
+      "loss": 0.7754,
+      "step": 3425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.072132970961516e-06,
+      "loss": 0.7802,
+      "step": 3426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066572377479254e-06,
+      "loss": 0.8079,
+      "step": 3427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0610132224636235e-06,
+      "loss": 0.7455,
+      "step": 3428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.055455507947623e-06,
+      "loss": 0.2417,
+      "step": 3429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049899235963723e-06,
+      "loss": 0.802,
+      "step": 3430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0443444085438765e-06,
+      "loss": 0.8258,
+      "step": 3431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.038791027719498e-06,
+      "loss": 0.7468,
+      "step": 3432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.033239095521477e-06,
+      "loss": 0.7954,
+      "step": 3433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02768861398017e-06,
+      "loss": 0.8005,
+      "step": 3434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022139585125403e-06,
+      "loss": 0.7772,
+      "step": 3435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016592010986478e-06,
+      "loss": 0.7661,
+      "step": 3436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.01104589359216e-06,
+      "loss": 0.8472,
+      "step": 3437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.005501234970679e-06,
+      "loss": 0.8521,
+      "step": 3438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.999958037149732e-06,
+      "loss": 0.7767,
+      "step": 3439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994416302156484e-06,
+      "loss": 0.8325,
+      "step": 3440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.98887603201757e-06,
+      "loss": 0.79,
+      "step": 3441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9833372287590795e-06,
+      "loss": 0.7404,
+      "step": 3442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.97779989440657e-06,
+      "loss": 0.8078,
+      "step": 3443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.972264030985064e-06,
+      "loss": 0.7386,
+      "step": 3444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96672964051904e-06,
+      "loss": 0.787,
+      "step": 3445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.961196725032445e-06,
+      "loss": 0.8414,
+      "step": 3446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.955665286548684e-06,
+      "loss": 0.774,
+      "step": 3447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.950135327090619e-06,
+      "loss": 0.7462,
+      "step": 3448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.944606848680573e-06,
+      "loss": 0.7645,
+      "step": 3449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.939079853340325e-06,
+      "loss": 0.7341,
+      "step": 3450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.933554343091122e-06,
+      "loss": 0.7685,
+      "step": 3451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9280303199536534e-06,
+      "loss": 0.7918,
+      "step": 3452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.922507785948077e-06,
+      "loss": 0.772,
+      "step": 3453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916986743093993e-06,
+      "loss": 0.8269,
+      "step": 3454
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9114671934104704e-06,
+      "loss": 0.8098,
+      "step": 3455
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.905949138916025e-06,
+      "loss": 0.813,
+      "step": 3456
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900432581628624e-06,
+      "loss": 0.8438,
+      "step": 3457
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894917523565689e-06,
+      "loss": 0.842,
+      "step": 3458
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8894039667440914e-06,
+      "loss": 0.8077,
+      "step": 3459
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883891913180165e-06,
+      "loss": 0.8381,
+      "step": 3460
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87838136488968e-06,
+      "loss": 0.7929,
+      "step": 3461
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.87287232388786e-06,
+      "loss": 0.8284,
+      "step": 3462
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.867364792189376e-06,
+      "loss": 0.8219,
+      "step": 3463
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.861858771808351e-06,
+      "loss": 0.8117,
+      "step": 3464
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856354264758356e-06,
+      "loss": 0.8252,
+      "step": 3465
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.850851273052406e-06,
+      "loss": 0.797,
+      "step": 3466
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8453497987029605e-06,
+      "loss": 0.7705,
+      "step": 3467
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.839849843721926e-06,
+      "loss": 0.7537,
+      "step": 3468
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8343514101206515e-06,
+      "loss": 0.7757,
+      "step": 3469
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828854499909933e-06,
+      "loss": 0.8113,
+      "step": 3470
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823359115100011e-06,
+      "loss": 0.7945,
+      "step": 3471
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.817865257700557e-06,
+      "loss": 0.7388,
+      "step": 3472
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8123729297207e-06,
+      "loss": 0.3035,
+      "step": 3473
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806882133168992e-06,
+      "loss": 0.729,
+      "step": 3474
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801392870053439e-06,
+      "loss": 0.7758,
+      "step": 3475
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795905142381486e-06,
+      "loss": 0.263,
+      "step": 3476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790418952160004e-06,
+      "loss": 0.7671,
+      "step": 3477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784934301395317e-06,
+      "loss": 0.8519,
+      "step": 3478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.779451192093167e-06,
+      "loss": 0.7942,
+      "step": 3479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.773969626258761e-06,
+      "loss": 0.7683,
+      "step": 3480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.76848960589671e-06,
+      "loss": 0.7987,
+      "step": 3481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.763011133011085e-06,
+      "loss": 0.7141,
+      "step": 3482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757534209605371e-06,
+      "loss": 0.7919,
+      "step": 3483
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.752058837682502e-06,
+      "loss": 0.7725,
+      "step": 3484
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.746585019244838e-06,
+      "loss": 0.7728,
+      "step": 3485
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741112756294167e-06,
+      "loss": 0.7263,
+      "step": 3486
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73564205083172e-06,
+      "loss": 0.7887,
+      "step": 3487
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730172904858138e-06,
+      "loss": 0.7509,
+      "step": 3488
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724705320373521e-06,
+      "loss": 0.7436,
+      "step": 3489
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719239299377368e-06,
+      "loss": 0.7845,
+      "step": 3490
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713774843868631e-06,
+      "loss": 0.7749,
+      "step": 3491
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.708311955845667e-06,
+      "loss": 0.8074,
+      "step": 3492
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.702850637306277e-06,
+      "loss": 0.8608,
+      "step": 3493
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697390890247687e-06,
+      "loss": 0.8115,
+      "step": 3494
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.691932716666533e-06,
+      "loss": 0.7815,
+      "step": 3495
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.686476118558893e-06,
+      "loss": 0.7691,
+      "step": 3496
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.681021097920257e-06,
+      "loss": 0.8304,
+      "step": 3497
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6755676567455464e-06,
+      "loss": 0.7469,
+      "step": 3498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.670115797029099e-06,
+      "loss": 0.796,
+      "step": 3499
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664665520764684e-06,
+      "loss": 0.8287,
+      "step": 3500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6592168299454776e-06,
+      "loss": 0.7469,
+      "step": 3501
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.65376972656408e-06,
+      "loss": 0.7318,
+      "step": 3502
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.648324212612517e-06,
+      "loss": 0.7886,
+      "step": 3503
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.642880290082231e-06,
+      "loss": 0.8499,
+      "step": 3504
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.637437960964085e-06,
+      "loss": 0.7785,
+      "step": 3505
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631997227248347e-06,
+      "loss": 0.8079,
+      "step": 3506
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6265580909247184e-06,
+      "loss": 0.8115,
+      "step": 3507
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6211205539823e-06,
+      "loss": 0.7821,
+      "step": 3508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61568461840962e-06,
+      "loss": 0.78,
+      "step": 3509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610250286194622e-06,
+      "loss": 0.773,
+      "step": 3510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604817559324648e-06,
+      "loss": 0.8333,
+      "step": 3511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5993864397864715e-06,
+      "loss": 0.8124,
+      "step": 3512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5939569295662585e-06,
+      "loss": 0.8103,
+      "step": 3513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5885290306496146e-06,
+      "loss": 0.7533,
+      "step": 3514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.583102745021524e-06,
+      "loss": 0.7502,
+      "step": 3515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577678074666406e-06,
+      "loss": 0.7918,
+      "step": 3516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.572255021568073e-06,
+      "loss": 0.8162,
+      "step": 3517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566833587709752e-06,
+      "loss": 0.7484,
+      "step": 3518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.561413775074087e-06,
+      "loss": 0.8143,
+      "step": 3519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55599558564311e-06,
+      "loss": 0.7416,
+      "step": 3520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5505790213982765e-06,
+      "loss": 0.803,
+      "step": 3521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.545164084320429e-06,
+      "loss": 0.7717,
+      "step": 3522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539750776389845e-06,
+      "loss": 0.8022,
+      "step": 3523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.534339099586175e-06,
+      "loss": 0.8203,
+      "step": 3524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5289290558884916e-06,
+      "loss": 0.7874,
+      "step": 3525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5235206472752575e-06,
+      "loss": 0.7813,
+      "step": 3526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5181138757243514e-06,
+      "loss": 0.8401,
+      "step": 3527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512708743213046e-06,
+      "loss": 0.798,
+      "step": 3528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5073052517180115e-06,
+      "loss": 0.7909,
+      "step": 3529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.501903403215326e-06,
+      "loss": 0.8253,
+      "step": 3530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4965031996804555e-06,
+      "loss": 0.7735,
+      "step": 3531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.491104643088274e-06,
+      "loss": 0.8236,
+      "step": 3532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.485707735413052e-06,
+      "loss": 0.8051,
+      "step": 3533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480312478628458e-06,
+      "loss": 0.8269,
+      "step": 3534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4749188747075465e-06,
+      "loss": 0.8109,
+      "step": 3535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4695269256227836e-06,
+      "loss": 0.8133,
+      "step": 3536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.464136633346011e-06,
+      "loss": 0.823,
+      "step": 3537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.458747999848484e-06,
+      "loss": 0.8608,
+      "step": 3538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453361027100843e-06,
+      "loss": 0.8077,
+      "step": 3539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.447975717073114e-06,
+      "loss": 0.7893,
+      "step": 3540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.442592071734729e-06,
+      "loss": 0.8064,
+      "step": 3541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4372100930544925e-06,
+      "loss": 0.7544,
+      "step": 3542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.431829783000624e-06,
+      "loss": 0.7755,
+      "step": 3543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426451143540715e-06,
+      "loss": 0.7774,
+      "step": 3544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.421074176641745e-06,
+      "loss": 0.7743,
+      "step": 3545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415698884270091e-06,
+      "loss": 0.7974,
+      "step": 3546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410325268391517e-06,
+      "loss": 0.7818,
+      "step": 3547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404953330971169e-06,
+      "loss": 0.8217,
+      "step": 3548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399583073973578e-06,
+      "loss": 0.817,
+      "step": 3549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.39421449936267e-06,
+      "loss": 0.7791,
+      "step": 3550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3888476091017416e-06,
+      "loss": 0.8295,
+      "step": 3551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383482405153485e-06,
+      "loss": 0.2404,
+      "step": 3552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.378118889479975e-06,
+      "loss": 0.2498,
+      "step": 3553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3727570640426605e-06,
+      "loss": 0.8137,
+      "step": 3554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367396930802382e-06,
+      "loss": 0.7693,
+      "step": 3555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362038491719348e-06,
+      "loss": 0.7965,
+      "step": 3556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356681748753171e-06,
+      "loss": 0.7764,
+      "step": 3557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351326703862816e-06,
+      "loss": 0.7591,
+      "step": 3558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.345973359006646e-06,
+      "loss": 0.7608,
+      "step": 3559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340621716142391e-06,
+      "loss": 0.8098,
+      "step": 3560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.335271777227167e-06,
+      "loss": 0.8173,
+      "step": 3561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.329923544217463e-06,
+      "loss": 0.2739,
+      "step": 3562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.324577019069141e-06,
+      "loss": 0.7502,
+      "step": 3563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.319232203737448e-06,
+      "loss": 0.7777,
+      "step": 3564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3138891001769925e-06,
+      "loss": 0.7727,
+      "step": 3565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.308547710341766e-06,
+      "loss": 0.8073,
+      "step": 3566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303208036185132e-06,
+      "loss": 0.7912,
+      "step": 3567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.29787007965983e-06,
+      "loss": 0.7842,
+      "step": 3568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292533842717961e-06,
+      "loss": 0.2653,
+      "step": 3569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.287199327311008e-06,
+      "loss": 0.8258,
+      "step": 3570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281866535389814e-06,
+      "loss": 0.7537,
+      "step": 3571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276535468904601e-06,
+      "loss": 0.849,
+      "step": 3572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.27120612980496e-06,
+      "loss": 0.7666,
+      "step": 3573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26587852003984e-06,
+      "loss": 0.7941,
+      "step": 3574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.26055264155757e-06,
+      "loss": 0.7746,
+      "step": 3575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255228496305831e-06,
+      "loss": 0.7681,
+      "step": 3576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249906086231693e-06,
+      "loss": 0.8249,
+      "step": 3577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244585413281569e-06,
+      "loss": 0.8094,
+      "step": 3578
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239266479401248e-06,
+      "loss": 0.784,
+      "step": 3579
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.233949286535879e-06,
+      "loss": 0.7559,
+      "step": 3580
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2286338366299745e-06,
+      "loss": 0.8061,
+      "step": 3581
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.223320131627418e-06,
+      "loss": 0.8184,
+      "step": 3582
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2180081734714405e-06,
+      "loss": 0.8559,
+      "step": 3583
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2126979641046474e-06,
+      "loss": 0.793,
+      "step": 3584
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.207389505468987e-06,
+      "loss": 0.7877,
+      "step": 3585
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202082799505794e-06,
+      "loss": 0.7722,
+      "step": 3586
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.196777848155743e-06,
+      "loss": 0.7529,
+      "step": 3587
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.191474653358864e-06,
+      "loss": 0.7633,
+      "step": 3588
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.186173217054558e-06,
+      "loss": 0.7696,
+      "step": 3589
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18087354118157e-06,
+      "loss": 0.7731,
+      "step": 3590
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175575627678019e-06,
+      "loss": 0.7911,
+      "step": 3591
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.170279478481359e-06,
+      "loss": 0.7558,
+      "step": 3592
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164985095528414e-06,
+      "loss": 0.8349,
+      "step": 3593
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1596924807553515e-06,
+      "loss": 0.7926,
+      "step": 3594
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154401636097698e-06,
+      "loss": 0.8076,
+      "step": 3595
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1491125634903375e-06,
+      "loss": 0.2684,
+      "step": 3596
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1438252648674904e-06,
+      "loss": 0.7846,
+      "step": 3597
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138539742162747e-06,
+      "loss": 0.7859,
+      "step": 3598
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.133255997309032e-06,
+      "loss": 0.8213,
+      "step": 3599
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127974032238629e-06,
+      "loss": 0.7906,
+      "step": 3600
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122693848883173e-06,
+      "loss": 0.7461,
+      "step": 3601
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117415449173644e-06,
+      "loss": 0.8274,
+      "step": 3602
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112138835040364e-06,
+      "loss": 0.7326,
+      "step": 3603
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.106864008413011e-06,
+      "loss": 0.7697,
+      "step": 3604
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101590971220599e-06,
+      "loss": 0.7034,
+      "step": 3605
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0963197253915e-06,
+      "loss": 0.7717,
+      "step": 3606
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091050272853426e-06,
+      "loss": 0.7427,
+      "step": 3607
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.085782615533426e-06,
+      "loss": 0.7489,
+      "step": 3608
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0805167553579026e-06,
+      "loss": 0.7557,
+      "step": 3609
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.075252694252597e-06,
+      "loss": 0.8187,
+      "step": 3610
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069990434142595e-06,
+      "loss": 0.7508,
+      "step": 3611
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064729976952316e-06,
+      "loss": 0.7857,
+      "step": 3612
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0594713246055305e-06,
+      "loss": 0.7631,
+      "step": 3613
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054214479025338e-06,
+      "loss": 0.7796,
+      "step": 3614
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.048959442134186e-06,
+      "loss": 0.8039,
+      "step": 3615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043706215853863e-06,
+      "loss": 0.7671,
+      "step": 3616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038454802105481e-06,
+      "loss": 0.7448,
+      "step": 3617
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0332052028095055e-06,
+      "loss": 0.7729,
+      "step": 3618
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027957419885723e-06,
+      "loss": 0.7792,
+      "step": 3619
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0227114552532755e-06,
+      "loss": 0.802,
+      "step": 3620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0174673108306175e-06,
+      "loss": 0.7983,
+      "step": 3621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.012224988535558e-06,
+      "loss": 0.75,
+      "step": 3622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0069844902852264e-06,
+      "loss": 0.747,
+      "step": 3623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.00174581799608e-06,
+      "loss": 0.2475,
+      "step": 3624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996508973583934e-06,
+      "loss": 0.8164,
+      "step": 3625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991273958963907e-06,
+      "loss": 0.8073,
+      "step": 3626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.986040776050468e-06,
+      "loss": 0.8018,
+      "step": 3627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980809426757399e-06,
+      "loss": 0.7507,
+      "step": 3628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.975579912997826e-06,
+      "loss": 0.8263,
+      "step": 3629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.970352236684202e-06,
+      "loss": 0.7626,
+      "step": 3630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965126399728295e-06,
+      "loss": 0.7666,
+      "step": 3631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9599024040412196e-06,
+      "loss": 0.7526,
+      "step": 3632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954680251533398e-06,
+      "loss": 0.7857,
+      "step": 3633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.949459944114592e-06,
+      "loss": 0.7468,
+      "step": 3634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944241483693884e-06,
+      "loss": 0.7864,
+      "step": 3635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.939024872179685e-06,
+      "loss": 0.2589,
+      "step": 3636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933810111479719e-06,
+      "loss": 0.783,
+      "step": 3637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928597203501045e-06,
+      "loss": 0.8506,
+      "step": 3638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923386150150036e-06,
+      "loss": 0.6988,
+      "step": 3639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9181769533323905e-06,
+      "loss": 0.8145,
+      "step": 3640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912969614953134e-06,
+      "loss": 0.7904,
+      "step": 3641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.907764136916598e-06,
+      "loss": 0.7605,
+      "step": 3642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.902560521126445e-06,
+      "loss": 0.7545,
+      "step": 3643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.897358769485654e-06,
+      "loss": 0.7781,
+      "step": 3644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892158883896526e-06,
+      "loss": 0.7696,
+      "step": 3645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886960866260666e-06,
+      "loss": 0.7463,
+      "step": 3646
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.881764718479014e-06,
+      "loss": 0.7664,
+      "step": 3647
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876570442451809e-06,
+      "loss": 0.7974,
+      "step": 3648
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871378040078618e-06,
+      "loss": 0.7833,
+      "step": 3649
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.866187513258322e-06,
+      "loss": 0.8013,
+      "step": 3650
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.860998863889107e-06,
+      "loss": 0.7942,
+      "step": 3651
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855812093868483e-06,
+      "loss": 0.7918,
+      "step": 3652
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.850627205093257e-06,
+      "loss": 0.7259,
+      "step": 3653
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845444199459575e-06,
+      "loss": 0.8076,
+      "step": 3654
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.840263078862866e-06,
+      "loss": 0.7734,
+      "step": 3655
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.83508384519789e-06,
+      "loss": 0.2642,
+      "step": 3656
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.829906500358701e-06,
+      "loss": 0.8069,
+      "step": 3657
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8247310462386735e-06,
+      "loss": 0.7347,
+      "step": 3658
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8195574847304904e-06,
+      "loss": 0.7795,
+      "step": 3659
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814385817726134e-06,
+      "loss": 0.7874,
+      "step": 3660
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809216047116902e-06,
+      "loss": 0.8581,
+      "step": 3661
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8040481747933934e-06,
+      "loss": 0.8442,
+      "step": 3662
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.798882202645514e-06,
+      "loss": 0.738,
+      "step": 3663
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.793718132562478e-06,
+      "loss": 0.7931,
+      "step": 3664
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788555966432807e-06,
+      "loss": 0.8131,
+      "step": 3665
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783395706144313e-06,
+      "loss": 0.7777,
+      "step": 3666
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7782373535841196e-06,
+      "loss": 0.866,
+      "step": 3667
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773080910638654e-06,
+      "loss": 0.7591,
+      "step": 3668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.767926379193643e-06,
+      "loss": 0.7906,
+      "step": 3669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7627737611341204e-06,
+      "loss": 0.265,
+      "step": 3670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.757623058344406e-06,
+      "loss": 0.7756,
+      "step": 3671
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.752474272708131e-06,
+      "loss": 0.7731,
+      "step": 3672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747327406108225e-06,
+      "loss": 0.7429,
+      "step": 3673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742182460426908e-06,
+      "loss": 0.7363,
+      "step": 3674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737039437545708e-06,
+      "loss": 0.7864,
+      "step": 3675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731898339345439e-06,
+      "loss": 0.7655,
+      "step": 3676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726759167706217e-06,
+      "loss": 0.8205,
+      "step": 3677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721621924507457e-06,
+      "loss": 0.8421,
+      "step": 3678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.716486611627865e-06,
+      "loss": 0.7441,
+      "step": 3679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.711353230945435e-06,
+      "loss": 0.7637,
+      "step": 3680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.706221784337466e-06,
+      "loss": 0.7593,
+      "step": 3681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7010922736805385e-06,
+      "loss": 0.8531,
+      "step": 3682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695964700850533e-06,
+      "loss": 0.7748,
+      "step": 3683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.690839067722624e-06,
+      "loss": 0.7571,
+      "step": 3684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6857153761712605e-06,
+      "loss": 0.2695,
+      "step": 3685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680593628070203e-06,
+      "loss": 0.7726,
+      "step": 3686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.675473825292478e-06,
+      "loss": 0.7541,
+      "step": 3687
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.670355969710429e-06,
+      "loss": 0.8149,
+      "step": 3688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665240063195659e-06,
+      "loss": 0.8254,
+      "step": 3689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660126107619078e-06,
+      "loss": 0.7476,
+      "step": 3690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6550141048508715e-06,
+      "loss": 0.8086,
+      "step": 3691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649904056760515e-06,
+      "loss": 0.7961,
+      "step": 3692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644795965216773e-06,
+      "loss": 0.8073,
+      "step": 3693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6396898320876825e-06,
+      "loss": 0.8704,
+      "step": 3694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.634585659240582e-06,
+      "loss": 0.7811,
+      "step": 3695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6294834485420735e-06,
+      "loss": 0.8187,
+      "step": 3696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624383201858057e-06,
+      "loss": 0.2584,
+      "step": 3697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619284921053706e-06,
+      "loss": 0.8516,
+      "step": 3698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.614188607993483e-06,
+      "loss": 0.2962,
+      "step": 3699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.609094264541118e-06,
+      "loss": 0.7962,
+      "step": 3700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.604001892559633e-06,
+      "loss": 0.8241,
+      "step": 3701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.598911493911321e-06,
+      "loss": 0.7896,
+      "step": 3702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593823070457757e-06,
+      "loss": 0.8154,
+      "step": 3703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.588736624059798e-06,
+      "loss": 0.7894,
+      "step": 3704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.583652156577566e-06,
+      "loss": 0.802,
+      "step": 3705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.578569669870468e-06,
+      "loss": 0.7446,
+      "step": 3706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573489165797189e-06,
+      "loss": 0.7684,
+      "step": 3707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.568410646215686e-06,
+      "loss": 0.7642,
+      "step": 3708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563334112983187e-06,
+      "loss": 0.8125,
+      "step": 3709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5582595679561905e-06,
+      "loss": 0.7782,
+      "step": 3710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.553187012990478e-06,
+      "loss": 0.8095,
+      "step": 3711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.548116449941098e-06,
+      "loss": 0.855,
+      "step": 3712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543047880662376e-06,
+      "loss": 0.7977,
+      "step": 3713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537981307007894e-06,
+      "loss": 0.8519,
+      "step": 3714
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532916730830523e-06,
+      "loss": 0.7022,
+      "step": 3715
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527854153982386e-06,
+      "loss": 0.269,
+      "step": 3716
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5227935783148865e-06,
+      "loss": 0.769,
+      "step": 3717
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.517735005678696e-06,
+      "loss": 0.765,
+      "step": 3718
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512678437923744e-06,
+      "loss": 0.2479,
+      "step": 3719
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5076238768992395e-06,
+      "loss": 0.7811,
+      "step": 3720
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.502571324453641e-06,
+      "loss": 0.749,
+      "step": 3721
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4975207824346955e-06,
+      "loss": 0.8315,
+      "step": 3722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492472252689392e-06,
+      "loss": 0.7704,
+      "step": 3723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487425737063999e-06,
+      "loss": 0.7395,
+      "step": 3724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482381237404039e-06,
+      "loss": 0.8247,
+      "step": 3725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477338755554302e-06,
+      "loss": 0.7616,
+      "step": 3726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.472298293358843e-06,
+      "loss": 0.7448,
+      "step": 3727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.467259852660969e-06,
+      "loss": 0.244,
+      "step": 3728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.462223435303259e-06,
+      "loss": 0.8226,
+      "step": 3729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.457189043127539e-06,
+      "loss": 0.7948,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.452156677974906e-06,
+      "loss": 0.2504,
+      "step": 3731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.447126341685712e-06,
+      "loss": 0.2365,
+      "step": 3732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.442098036099569e-06,
+      "loss": 0.7263,
+      "step": 3733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.437071763055339e-06,
+      "loss": 0.8228,
+      "step": 3734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4320475243911455e-06,
+      "loss": 0.769,
+      "step": 3735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.427025321944374e-06,
+      "loss": 0.7475,
+      "step": 3736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.422005157551652e-06,
+      "loss": 0.7267,
+      "step": 3737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4169870330488764e-06,
+      "loss": 0.7294,
+      "step": 3738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.411970950271184e-06,
+      "loss": 0.2371,
+      "step": 3739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406956911052974e-06,
+      "loss": 0.7428,
+      "step": 3740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.401944917227897e-06,
+      "loss": 0.7633,
+      "step": 3741
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.396934970628857e-06,
+      "loss": 0.7594,
+      "step": 3742
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391927073088004e-06,
+      "loss": 0.7321,
+      "step": 3743
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386921226436742e-06,
+      "loss": 0.742,
+      "step": 3744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381917432505724e-06,
+      "loss": 0.7211,
+      "step": 3745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3769156931248514e-06,
+      "loss": 0.2609,
+      "step": 3746
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371916010123281e-06,
+      "loss": 0.8107,
+      "step": 3747
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3669183853294075e-06,
+      "loss": 0.742,
+      "step": 3748
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361922820570882e-06,
+      "loss": 0.7362,
+      "step": 3749
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356929317674589e-06,
+      "loss": 0.7865,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35193787846668e-06,
+      "loss": 0.8318,
+      "step": 3751
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346948504772534e-06,
+      "loss": 0.7453,
+      "step": 3752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341961198416778e-06,
+      "loss": 0.7889,
+      "step": 3753
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336975961223291e-06,
+      "loss": 0.842,
+      "step": 3754
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.331992795015178e-06,
+      "loss": 0.7758,
+      "step": 3755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327011701614815e-06,
+      "loss": 0.2384,
+      "step": 3756
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3220326828437906e-06,
+      "loss": 0.7691,
+      "step": 3757
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317055740522956e-06,
+      "loss": 0.7838,
+      "step": 3758
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3120808764723855e-06,
+      "loss": 0.7822,
+      "step": 3759
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307108092511407e-06,
+      "loss": 0.8456,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.302137390458587e-06,
+      "loss": 0.8012,
+      "step": 3761
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297168772131718e-06,
+      "loss": 0.7364,
+      "step": 3762
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292202239347847e-06,
+      "loss": 0.8547,
+      "step": 3763
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2872377939232435e-06,
+      "loss": 0.7728,
+      "step": 3764
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282275437673424e-06,
+      "loss": 0.8,
+      "step": 3765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277315172413135e-06,
+      "loss": 0.7468,
+      "step": 3766
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.272356999956368e-06,
+      "loss": 0.7367,
+      "step": 3767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.267400922116335e-06,
+      "loss": 0.7976,
+      "step": 3768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262446940705489e-06,
+      "loss": 0.7846,
+      "step": 3769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2574950575355235e-06,
+      "loss": 0.7978,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.252545274417348e-06,
+      "loss": 0.7983,
+      "step": 3771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247597593161122e-06,
+      "loss": 0.7948,
+      "step": 3772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242652015576219e-06,
+      "loss": 0.8343,
+      "step": 3773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237708543471256e-06,
+      "loss": 0.8404,
+      "step": 3774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.232767178654078e-06,
+      "loss": 0.7657,
+      "step": 3775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.22782792293176e-06,
+      "loss": 0.8198,
+      "step": 3776
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222890778110595e-06,
+      "loss": 0.257,
+      "step": 3777
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.217955745996118e-06,
+      "loss": 0.7425,
+      "step": 3778
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.213022828393082e-06,
+      "loss": 0.7689,
+      "step": 3779
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208092027105471e-06,
+      "loss": 0.8113,
+      "step": 3780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2031633439364995e-06,
+      "loss": 0.7612,
+      "step": 3781
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.198236780688594e-06,
+      "loss": 0.7837,
+      "step": 3782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.193312339163422e-06,
+      "loss": 0.8039,
+      "step": 3783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.188390021161853e-06,
+      "loss": 0.7509,
+      "step": 3784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1834698284840125e-06,
+      "loss": 0.7846,
+      "step": 3785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.178551762929218e-06,
+      "loss": 0.7406,
+      "step": 3786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.173635826296027e-06,
+      "loss": 0.2676,
+      "step": 3787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.168722020382211e-06,
+      "loss": 0.6872,
+      "step": 3788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163810346984755e-06,
+      "loss": 0.7969,
+      "step": 3789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1589008078998905e-06,
+      "loss": 0.7536,
+      "step": 3790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.153993404923037e-06,
+      "loss": 0.8726,
+      "step": 3791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149088139848855e-06,
+      "loss": 0.7532,
+      "step": 3792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.14418501447121e-06,
+      "loss": 0.8148,
+      "step": 3793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139284030583189e-06,
+      "loss": 0.7973,
+      "step": 3794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.134385189977106e-06,
+      "loss": 0.2535,
+      "step": 3795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.129488494444472e-06,
+      "loss": 0.8096,
+      "step": 3796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124593945776032e-06,
+      "loss": 0.7215,
+      "step": 3797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119701545761725e-06,
+      "loss": 0.7874,
+      "step": 3798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1148112961907335e-06,
+      "loss": 0.8595,
+      "step": 3799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109923198851425e-06,
+      "loss": 0.7818,
+      "step": 3800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105037255531398e-06,
+      "loss": 0.7817,
+      "step": 3801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.10015346801745e-06,
+      "loss": 0.7765,
+      "step": 3802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.095271838095604e-06,
+      "loss": 0.7813,
+      "step": 3803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.090392367551087e-06,
+      "loss": 0.7497,
+      "step": 3804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.08551505816833e-06,
+      "loss": 0.2639,
+      "step": 3805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080639911730989e-06,
+      "loss": 0.8628,
+      "step": 3806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075766930021912e-06,
+      "loss": 0.7679,
+      "step": 3807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.070896114823165e-06,
+      "loss": 0.8328,
+      "step": 3808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.066027467916024e-06,
+      "loss": 0.7326,
+      "step": 3809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061160991080968e-06,
+      "loss": 0.7694,
+      "step": 3810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.056296686097679e-06,
+      "loss": 0.8182,
+      "step": 3811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.051434554745053e-06,
+      "loss": 0.2757,
+      "step": 3812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0465745988011796e-06,
+      "loss": 0.7614,
+      "step": 3813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041716820043364e-06,
+      "loss": 0.2455,
+      "step": 3814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036861220248115e-06,
+      "loss": 0.7626,
+      "step": 3815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.032007801191132e-06,
+      "loss": 0.7888,
+      "step": 3816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.027156564647332e-06,
+      "loss": 0.8089,
+      "step": 3817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022307512390818e-06,
+      "loss": 0.8106,
+      "step": 3818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.017460646194918e-06,
+      "loss": 0.7428,
+      "step": 3819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.012615967832134e-06,
+      "loss": 0.7816,
+      "step": 3820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007773479074187e-06,
+      "loss": 0.7385,
+      "step": 3821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.002933181691984e-06,
+      "loss": 0.7831,
+      "step": 3822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998095077455639e-06,
+      "loss": 0.7918,
+      "step": 3823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.993259168134465e-06,
+      "loss": 0.7872,
+      "step": 3824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.988425455496964e-06,
+      "loss": 0.7852,
+      "step": 3825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.983593941310845e-06,
+      "loss": 0.2626,
+      "step": 3826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978764627343001e-06,
+      "loss": 0.8213,
+      "step": 3827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973937515359531e-06,
+      "loss": 0.8008,
+      "step": 3828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969112607125724e-06,
+      "loss": 0.7831,
+      "step": 3829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964289904406067e-06,
+      "loss": 0.7725,
+      "step": 3830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959469408964235e-06,
+      "loss": 0.7935,
+      "step": 3831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9546511225630905e-06,
+      "loss": 0.7498,
+      "step": 3832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949835046964709e-06,
+      "loss": 0.7966,
+      "step": 3833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945021183930335e-06,
+      "loss": 0.7794,
+      "step": 3834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9402095352204195e-06,
+      "loss": 0.802,
+      "step": 3835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93540010259459e-06,
+      "loss": 0.8329,
+      "step": 3836
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.930592887811675e-06,
+      "loss": 0.7625,
+      "step": 3837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.925787892629691e-06,
+      "loss": 0.795,
+      "step": 3838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.920985118805833e-06,
+      "loss": 0.7944,
+      "step": 3839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.916184568096496e-06,
+      "loss": 0.8036,
+      "step": 3840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911386242257251e-06,
+      "loss": 0.7724,
+      "step": 3841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.906590143042862e-06,
+      "loss": 0.7794,
+      "step": 3842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.901796272207279e-06,
+      "loss": 0.785,
+      "step": 3843
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.897004631503638e-06,
+      "loss": 0.7623,
+      "step": 3844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.89221522268425e-06,
+      "loss": 0.741,
+      "step": 3845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8874280475006245e-06,
+      "loss": 0.7592,
+      "step": 3846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.882643107703438e-06,
+      "loss": 0.768,
+      "step": 3847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877860405042561e-06,
+      "loss": 0.8464,
+      "step": 3848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873079941267047e-06,
+      "loss": 0.7775,
+      "step": 3849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.868301718125119e-06,
+      "loss": 0.7813,
+      "step": 3850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.863525737364196e-06,
+      "loss": 0.7504,
+      "step": 3851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858752000730856e-06,
+      "loss": 0.259,
+      "step": 3852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.853980509970887e-06,
+      "loss": 0.818,
+      "step": 3853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.849211266829224e-06,
+      "loss": 0.7735,
+      "step": 3854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844444273050002e-06,
+      "loss": 0.8347,
+      "step": 3855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839679530376519e-06,
+      "loss": 0.7966,
+      "step": 3856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.834917040551258e-06,
+      "loss": 0.7688,
+      "step": 3857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830156805315882e-06,
+      "loss": 0.76,
+      "step": 3858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8253988264112155e-06,
+      "loss": 0.836,
+      "step": 3859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820643105577273e-06,
+      "loss": 0.7589,
+      "step": 3860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8158896445532255e-06,
+      "loss": 0.7451,
+      "step": 3861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8111384450774424e-06,
+      "loss": 0.8224,
+      "step": 3862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8063895088874438e-06,
+      "loss": 0.2439,
+      "step": 3863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801642837719933e-06,
+      "loss": 0.8045,
+      "step": 3864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7968984333107795e-06,
+      "loss": 0.7637,
+      "step": 3865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7921562973950287e-06,
+      "loss": 0.7987,
+      "step": 3866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7874164317068964e-06,
+      "loss": 0.7664,
+      "step": 3867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782678837979761e-06,
+      "loss": 0.765,
+      "step": 3868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7779435179461808e-06,
+      "loss": 0.793,
+      "step": 3869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7732104733378716e-06,
+      "loss": 0.7858,
+      "step": 3870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.768479705885725e-06,
+      "loss": 0.8198,
+      "step": 3871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7637512173197967e-06,
+      "loss": 0.7805,
+      "step": 3872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.759025009369314e-06,
+      "loss": 0.7823,
+      "step": 3873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7543010837626604e-06,
+      "loss": 0.784,
+      "step": 3874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.749579442227389e-06,
+      "loss": 0.7475,
+      "step": 3875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7448600864902218e-06,
+      "loss": 0.7295,
+      "step": 3876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7401430182770403e-06,
+      "loss": 0.7997,
+      "step": 3877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.735428239312896e-06,
+      "loss": 0.7671,
+      "step": 3878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7307157513219905e-06,
+      "loss": 0.7352,
+      "step": 3879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.726005556027702e-06,
+      "loss": 0.8122,
+      "step": 3880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.721297655152556e-06,
+      "loss": 0.74,
+      "step": 3881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7165920504182516e-06,
+      "loss": 0.8066,
+      "step": 3882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.711888743545644e-06,
+      "loss": 0.8048,
+      "step": 3883
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7071877362547424e-06,
+      "loss": 0.7752,
+      "step": 3884
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7024890302647253e-06,
+      "loss": 0.7637,
+      "step": 3885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6977926272939145e-06,
+      "loss": 0.7723,
+      "step": 3886
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.693098529059812e-06,
+      "loss": 0.8276,
+      "step": 3887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6884067372790545e-06,
+      "loss": 0.7471,
+      "step": 3888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6837172536674516e-06,
+      "loss": 0.7228,
+      "step": 3889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6790300799399535e-06,
+      "loss": 0.7575,
+      "step": 3890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6743452178106797e-06,
+      "loss": 0.7965,
+      "step": 3891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6696626689929004e-06,
+      "loss": 0.7748,
+      "step": 3892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.664982435199033e-06,
+      "loss": 0.7932,
+      "step": 3893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6603045181406583e-06,
+      "loss": 0.7748,
+      "step": 3894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6556289195284966e-06,
+      "loss": 0.7719,
+      "step": 3895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.650955641072441e-06,
+      "loss": 0.2369,
+      "step": 3896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.646284684481516e-06,
+      "loss": 0.7652,
+      "step": 3897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.641616051463909e-06,
+      "loss": 0.7627,
+      "step": 3898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.636949743726946e-06,
+      "loss": 0.7942,
+      "step": 3899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.632285762977118e-06,
+      "loss": 0.8072,
+      "step": 3900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.627624110920056e-06,
+      "loss": 0.791,
+      "step": 3901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6229647892605367e-06,
+      "loss": 0.7806,
+      "step": 3902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618307799702493e-06,
+      "loss": 0.7581,
+      "step": 3903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.613653143948995e-06,
+      "loss": 0.7588,
+      "step": 3904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.609000823702269e-06,
+      "loss": 0.8131,
+      "step": 3905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6043508406636805e-06,
+      "loss": 0.8525,
+      "step": 3906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.599703196533747e-06,
+      "loss": 0.8211,
+      "step": 3907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5950578930121203e-06,
+      "loss": 0.7389,
+      "step": 3908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5904149317976077e-06,
+      "loss": 0.849,
+      "step": 3909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5857743145881486e-06,
+      "loss": 0.7837,
+      "step": 3910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.581136043080835e-06,
+      "loss": 0.8172,
+      "step": 3911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5765001189718993e-06,
+      "loss": 0.8063,
+      "step": 3912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5718665439567067e-06,
+      "loss": 0.746,
+      "step": 3913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.567235319729777e-06,
+      "loss": 0.8318,
+      "step": 3914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5626064479847535e-06,
+      "loss": 0.7627,
+      "step": 3915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5579799304144413e-06,
+      "loss": 0.7401,
+      "step": 3916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5533557687107655e-06,
+      "loss": 0.7896,
+      "step": 3917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5487339645647943e-06,
+      "loss": 0.8142,
+      "step": 3918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5441145196667415e-06,
+      "loss": 0.8153,
+      "step": 3919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.539497435705943e-06,
+      "loss": 0.8254,
+      "step": 3920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.534882714370892e-06,
+      "loss": 0.7529,
+      "step": 3921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5302703573491993e-06,
+      "loss": 0.773,
+      "step": 3922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.525660366327622e-06,
+      "loss": 0.7276,
+      "step": 3923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5210527429920437e-06,
+      "loss": 0.763,
+      "step": 3924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.516447489027488e-06,
+      "loss": 0.7818,
+      "step": 3925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5118446061181156e-06,
+      "loss": 0.7709,
+      "step": 3926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5072440959472065e-06,
+      "loss": 0.7728,
+      "step": 3927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5026459601971884e-06,
+      "loss": 0.8134,
+      "step": 3928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.498050200549605e-06,
+      "loss": 0.7193,
+      "step": 3929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4934568186851535e-06,
+      "loss": 0.821,
+      "step": 3930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.488865816283635e-06,
+      "loss": 0.7521,
+      "step": 3931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.484277195024004e-06,
+      "loss": 0.8079,
+      "step": 3932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479690956584323e-06,
+      "loss": 0.7909,
+      "step": 3933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4751071026417983e-06,
+      "loss": 0.7216,
+      "step": 3934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4705256348727634e-06,
+      "loss": 0.7583,
+      "step": 3935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4659465549526684e-06,
+      "loss": 0.795,
+      "step": 3936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.461369864556102e-06,
+      "loss": 0.7853,
+      "step": 3937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.45679556535677e-06,
+      "loss": 0.8022,
+      "step": 3938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4522236590275092e-06,
+      "loss": 0.8216,
+      "step": 3939
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4476541472402802e-06,
+      "loss": 0.282,
+      "step": 3940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4430870316661713e-06,
+      "loss": 0.7688,
+      "step": 3941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.438522313975383e-06,
+      "loss": 0.2462,
+      "step": 3942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433959995837256e-06,
+      "loss": 0.2556,
+      "step": 3943
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.429400078920234e-06,
+      "loss": 0.7838,
+      "step": 3944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.424842564891899e-06,
+      "loss": 0.743,
+      "step": 3945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4202874554189493e-06,
+      "loss": 0.6976,
+      "step": 3946
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4157347521671967e-06,
+      "loss": 0.746,
+      "step": 3947
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.411184456801585e-06,
+      "loss": 0.7538,
+      "step": 3948
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4066365709861628e-06,
+      "loss": 0.7644,
+      "step": 3949
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.402091096384117e-06,
+      "loss": 0.792,
+      "step": 3950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397548034657734e-06,
+      "loss": 0.8062,
+      "step": 3951
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.39300738746843e-06,
+      "loss": 0.7768,
+      "step": 3952
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3884691564767315e-06,
+      "loss": 0.85,
+      "step": 3953
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3839333433422774e-06,
+      "loss": 0.2373,
+      "step": 3954
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3793999497238404e-06,
+      "loss": 0.8252,
+      "step": 3955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3748689772792876e-06,
+      "loss": 0.7292,
+      "step": 3956
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.370340427665617e-06,
+      "loss": 0.7981,
+      "step": 3957
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3658143025389256e-06,
+      "loss": 0.7365,
+      "step": 3958
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3612906035544337e-06,
+      "loss": 0.7612,
+      "step": 3959
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.356769332366477e-06,
+      "loss": 0.7584,
+      "step": 3960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3522504906284915e-06,
+      "loss": 0.8473,
+      "step": 3961
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.347734079993037e-06,
+      "loss": 0.2741,
+      "step": 3962
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3432201021117705e-06,
+      "loss": 0.7449,
+      "step": 3963
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3387085586354795e-06,
+      "loss": 0.7626,
+      "step": 3964
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3341994512140397e-06,
+      "loss": 0.8128,
+      "step": 3965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.329692781496453e-06,
+      "loss": 0.7383,
+      "step": 3966
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.325188551130816e-06,
+      "loss": 0.758,
+      "step": 3967
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3206867617643413e-06,
+      "loss": 0.7713,
+      "step": 3968
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3161874150433515e-06,
+      "loss": 0.297,
+      "step": 3969
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311690512613265e-06,
+      "loss": 0.7634,
+      "step": 3970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30719605611862e-06,
+      "loss": 0.7741,
+      "step": 3971
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3027040472030458e-06,
+      "loss": 0.7855,
+      "step": 3972
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2982144875092868e-06,
+      "loss": 0.8283,
+      "step": 3973
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2937273786791903e-06,
+      "loss": 0.7519,
+      "step": 3974
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2892427223537082e-06,
+      "loss": 0.7576,
+      "step": 3975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2847605201728873e-06,
+      "loss": 0.7783,
+      "step": 3976
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.280280773775888e-06,
+      "loss": 0.7642,
+      "step": 3977
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.275803484800961e-06,
+      "loss": 0.8118,
+      "step": 3978
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2713286548854695e-06,
+      "loss": 0.7696,
+      "step": 3979
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.266856285665876e-06,
+      "loss": 0.8077,
+      "step": 3980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.26238637877773e-06,
+      "loss": 0.7466,
+      "step": 3981
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2579189358557006e-06,
+      "loss": 0.7811,
+      "step": 3982
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2534539585335333e-06,
+      "loss": 0.7763,
+      "step": 3983
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2489914484440986e-06,
+      "loss": 0.7923,
+      "step": 3984
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244531407219339e-06,
+      "loss": 0.7952,
+      "step": 3985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2400738364903127e-06,
+      "loss": 0.782,
+      "step": 3986
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2356187378871616e-06,
+      "loss": 0.7849,
+      "step": 3987
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2311661130391303e-06,
+      "loss": 0.2578,
+      "step": 3988
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2267159635745615e-06,
+      "loss": 0.7611,
+      "step": 3989
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2222682911208825e-06,
+      "loss": 0.7646,
+      "step": 3990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2178230973046267e-06,
+      "loss": 0.297,
+      "step": 3991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2133803837514056e-06,
+      "loss": 0.7773,
+      "step": 3992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2089401520859466e-06,
+      "loss": 0.7996,
+      "step": 3993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.204502403932047e-06,
+      "loss": 0.7609,
+      "step": 3994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2000671409126105e-06,
+      "loss": 0.8187,
+      "step": 3995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1956343646496223e-06,
+      "loss": 0.8042,
+      "step": 3996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1912040767641584e-06,
+      "loss": 0.7775,
+      "step": 3997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1867762788764003e-06,
+      "loss": 0.7744,
+      "step": 3998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.182350972605598e-06,
+      "loss": 0.8029,
+      "step": 3999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.177928159570105e-06,
+      "loss": 0.2488,
+      "step": 4000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.173507841387352e-06,
+      "loss": 0.7905,
+      "step": 4001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1690900196738674e-06,
+      "loss": 0.7845,
+      "step": 4002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1646746960452634e-06,
+      "loss": 0.7928,
+      "step": 4003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1602618721162326e-06,
+      "loss": 0.7738,
+      "step": 4004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1558515495005647e-06,
+      "loss": 0.754,
+      "step": 4005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.151443729811121e-06,
+      "loss": 0.7409,
+      "step": 4006
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1470384146598576e-06,
+      "loss": 0.75,
+      "step": 4007
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1426356056578143e-06,
+      "loss": 0.7618,
+      "step": 4008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1382353044151127e-06,
+      "loss": 0.7556,
+      "step": 4009
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1338375125409505e-06,
+      "loss": 0.8179,
+      "step": 4010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1294422316436214e-06,
+      "loss": 0.7535,
+      "step": 4011
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1250494633304853e-06,
+      "loss": 0.8074,
+      "step": 4012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.120659209207996e-06,
+      "loss": 0.812,
+      "step": 4013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1162714708816844e-06,
+      "loss": 0.8053,
+      "step": 4014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.111886249956155e-06,
+      "loss": 0.768,
+      "step": 4015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1075035480351024e-06,
+      "loss": 0.2439,
+      "step": 4016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1031233667212844e-06,
+      "loss": 0.8211,
+      "step": 4017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0987457076165608e-06,
+      "loss": 0.725,
+      "step": 4018
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.094370572321844e-06,
+      "loss": 0.7844,
+      "step": 4019
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.089997962437141e-06,
+      "loss": 0.8046,
+      "step": 4020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0856278795615237e-06,
+      "loss": 0.8069,
+      "step": 4021
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.081260325293147e-06,
+      "loss": 0.7576,
+      "step": 4022
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.076895301229242e-06,
+      "loss": 0.8076,
+      "step": 4023
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0725328089661065e-06,
+      "loss": 0.7808,
+      "step": 4024
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0681728500991216e-06,
+      "loss": 0.8075,
+      "step": 4025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0638154262227303e-06,
+      "loss": 0.7632,
+      "step": 4026
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0594605389304677e-06,
+      "loss": 0.7573,
+      "step": 4027
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.055108189814919e-06,
+      "loss": 0.7602,
+      "step": 4028
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.05075838046776e-06,
+      "loss": 0.763,
+      "step": 4029
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0464111124797224e-06,
+      "loss": 0.7869,
+      "step": 4030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0420663874406174e-06,
+      "loss": 0.7945,
+      "step": 4031
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.03772420693933e-06,
+      "loss": 0.8227,
+      "step": 4032
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0333845725638e-06,
+      "loss": 0.7305,
+      "step": 4033
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0290474859010545e-06,
+      "loss": 0.7615,
+      "step": 4034
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0247129485371696e-06,
+      "loss": 0.7576,
+      "step": 4035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0203809620573044e-06,
+      "loss": 0.2339,
+      "step": 4036
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0160515280456802e-06,
+      "loss": 0.7278,
+      "step": 4037
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0117246480855854e-06,
+      "loss": 0.2681,
+      "step": 4038
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.007400323759372e-06,
+      "loss": 0.7877,
+      "step": 4039
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0030785566484556e-06,
+      "loss": 0.7726,
+      "step": 4040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.998759348333322e-06,
+      "loss": 0.8168,
+      "step": 4041
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9944427003935196e-06,
+      "loss": 0.7395,
+      "step": 4042
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9901286144076623e-06,
+      "loss": 0.7734,
+      "step": 4043
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.985817091953419e-06,
+      "loss": 0.7606,
+      "step": 4044
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9815081346075338e-06,
+      "loss": 0.267,
+      "step": 4045
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.977201743945798e-06,
+      "loss": 0.8554,
+      "step": 4046
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9728979215430755e-06,
+      "loss": 0.7792,
+      "step": 4047
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9685966689732904e-06,
+      "loss": 0.8002,
+      "step": 4048
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9642979878094193e-06,
+      "loss": 0.2564,
+      "step": 4049
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.960001879623504e-06,
+      "loss": 0.7618,
+      "step": 4050
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9557083459866477e-06,
+      "loss": 0.7807,
+      "step": 4051
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.951417388469008e-06,
+      "loss": 0.8007,
+      "step": 4052
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9471290086397986e-06,
+      "loss": 0.7701,
+      "step": 4053
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9428432080672977e-06,
+      "loss": 0.8101,
+      "step": 4054
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9385599883188285e-06,
+      "loss": 0.7596,
+      "step": 4055
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934279350960784e-06,
+      "loss": 0.7956,
+      "step": 4056
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9300012975586066e-06,
+      "loss": 0.7656,
+      "step": 4057
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9257258296767897e-06,
+      "loss": 0.2612,
+      "step": 4058
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.92145294887889e-06,
+      "loss": 0.7426,
+      "step": 4059
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9171826567275043e-06,
+      "loss": 0.8052,
+      "step": 4060
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.912914954784303e-06,
+      "loss": 0.8057,
+      "step": 4061
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9086498446099907e-06,
+      "loss": 0.7846,
+      "step": 4062
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9043873277643354e-06,
+      "loss": 0.7535,
+      "step": 4063
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.900127405806148e-06,
+      "loss": 0.7605,
+      "step": 4064
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.895870080293297e-06,
+      "loss": 0.7354,
+      "step": 4065
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.891615352782704e-06,
+      "loss": 0.8182,
+      "step": 4066
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8873632248303287e-06,
+      "loss": 0.7716,
+      "step": 4067
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8831136979911933e-06,
+      "loss": 0.7778,
+      "step": 4068
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.878866773819359e-06,
+      "loss": 0.7781,
+      "step": 4069
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874622453867939e-06,
+      "loss": 0.7891,
+      "step": 4070
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.870380739689096e-06,
+      "loss": 0.8153,
+      "step": 4071
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.866141632834042e-06,
+      "loss": 0.7952,
+      "step": 4072
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8619051348530226e-06,
+      "loss": 0.7586,
+      "step": 4073
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.857671247295346e-06,
+      "loss": 0.7744,
+      "step": 4074
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.853439971709353e-06,
+      "loss": 0.7317,
+      "step": 4075
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8492113096424347e-06,
+      "loss": 0.7536,
+      "step": 4076
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8449852626410303e-06,
+      "loss": 0.8281,
+      "step": 4077
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8407618322506125e-06,
+      "loss": 0.7728,
+      "step": 4078
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8365410200157074e-06,
+      "loss": 0.7583,
+      "step": 4079
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8323228274798754e-06,
+      "loss": 0.7836,
+      "step": 4080
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8281072561857227e-06,
+      "loss": 0.8367,
+      "step": 4081
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8238943076749027e-06,
+      "loss": 0.7146,
+      "step": 4082
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.819683983488095e-06,
+      "loss": 0.7482,
+      "step": 4083
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.815476285165033e-06,
+      "loss": 0.776,
+      "step": 4084
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8112712142444843e-06,
+      "loss": 0.7364,
+      "step": 4085
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8070687722642586e-06,
+      "loss": 0.8267,
+      "step": 4086
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.802868960761197e-06,
+      "loss": 0.7769,
+      "step": 4087
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7986717812711895e-06,
+      "loss": 0.7536,
+      "step": 4088
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7944772353291517e-06,
+      "loss": 0.7846,
+      "step": 4089
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7902853244690444e-06,
+      "loss": 0.7826,
+      "step": 4090
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.786096050223864e-06,
+      "loss": 0.7462,
+      "step": 4091
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7819094141256388e-06,
+      "loss": 0.7458,
+      "step": 4092
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.777725417705437e-06,
+      "loss": 0.7674,
+      "step": 4093
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.773544062493352e-06,
+      "loss": 0.809,
+      "step": 4094
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7693653500185293e-06,
+      "loss": 0.7643,
+      "step": 4095
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7651892818091284e-06,
+      "loss": 0.7372,
+      "step": 4096
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.761015859392356e-06,
+      "loss": 0.7556,
+      "step": 4097
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7568450842944406e-06,
+      "loss": 0.7366,
+      "step": 4098
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.75267695804065e-06,
+      "loss": 0.7505,
+      "step": 4099
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7485114821552827e-06,
+      "loss": 0.6911,
+      "step": 4100
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7443486581616607e-06,
+      "loss": 0.7894,
+      "step": 4101
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7401884875821495e-06,
+      "loss": 0.76,
+      "step": 4102
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7360309719381275e-06,
+      "loss": 0.7416,
+      "step": 4103
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.731876112750015e-06,
+      "loss": 0.7367,
+      "step": 4104
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.727723911537258e-06,
+      "loss": 0.8692,
+      "step": 4105
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7235743698183314e-06,
+      "loss": 0.7607,
+      "step": 4106
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7194274891107286e-06,
+      "loss": 0.8123,
+      "step": 4107
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.715283270930984e-06,
+      "loss": 0.7616,
+      "step": 4108
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7111417167946443e-06,
+      "loss": 0.7832,
+      "step": 4109
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7070028282162917e-06,
+      "loss": 0.7674,
+      "step": 4110
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7028666067095343e-06,
+      "loss": 0.8293,
+      "step": 4111
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6987330537869936e-06,
+      "loss": 0.8082,
+      "step": 4112
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6946021709603275e-06,
+      "loss": 0.7362,
+      "step": 4113
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.690473959740212e-06,
+      "loss": 0.797,
+      "step": 4114
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6863484216363477e-06,
+      "loss": 0.2468,
+      "step": 4115
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.682225558157453e-06,
+      "loss": 0.8063,
+      "step": 4116
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.678105370811277e-06,
+      "loss": 0.7842,
+      "step": 4117
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.67398786110458e-06,
+      "loss": 0.7647,
+      "step": 4118
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6698730305431487e-06,
+      "loss": 0.8051,
+      "step": 4119
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6657608806317938e-06,
+      "loss": 0.7308,
+      "step": 4120
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6616514128743356e-06,
+      "loss": 0.7475,
+      "step": 4121
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.657544628773625e-06,
+      "loss": 0.7912,
+      "step": 4122
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.653440529831518e-06,
+      "loss": 0.7442,
+      "step": 4123
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.649339117548901e-06,
+      "loss": 0.6991,
+      "step": 4124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6452403934256764e-06,
+      "loss": 0.7718,
+      "step": 4125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6411443589607532e-06,
+      "loss": 0.8079,
+      "step": 4126
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.63705101565207e-06,
+      "loss": 0.8,
+      "step": 4127
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6329603649965663e-06,
+      "loss": 0.6469,
+      "step": 4128
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6288724084902184e-06,
+      "loss": 0.7308,
+      "step": 4129
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.624787147627994e-06,
+      "loss": 0.7287,
+      "step": 4130
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6207045839038925e-06,
+      "loss": 0.765,
+      "step": 4131
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.616624718810914e-06,
+      "loss": 0.7535,
+      "step": 4132
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6125475538410817e-06,
+      "loss": 0.8143,
+      "step": 4133
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6084730904854283e-06,
+      "loss": 0.7836,
+      "step": 4134
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6044013302339945e-06,
+      "loss": 0.8169,
+      "step": 4135
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6003322745758387e-06,
+      "loss": 0.7683,
+      "step": 4136
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.596265924999023e-06,
+      "loss": 0.7582,
+      "step": 4137
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5922022829906257e-06,
+      "loss": 0.2397,
+      "step": 4138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5881413500367347e-06,
+      "loss": 0.7591,
+      "step": 4139
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5840831276224475e-06,
+      "loss": 0.7207,
+      "step": 4140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.580027617231863e-06,
+      "loss": 0.764,
+      "step": 4141
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5759748203481006e-06,
+      "loss": 0.6718,
+      "step": 4142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5719247384532732e-06,
+      "loss": 0.7924,
+      "step": 4143
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5678773730285124e-06,
+      "loss": 0.7445,
+      "step": 4144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5638327255539552e-06,
+      "loss": 0.742,
+      "step": 4145
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5597907975087343e-06,
+      "loss": 0.8214,
+      "step": 4146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5557515903710005e-06,
+      "loss": 0.8414,
+      "step": 4147
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.551715105617902e-06,
+      "loss": 0.7603,
+      "step": 4148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5476813447255988e-06,
+      "loss": 0.7601,
+      "step": 4149
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.543650309169242e-06,
+      "loss": 0.7719,
+      "step": 4150
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.539622000423001e-06,
+      "loss": 0.7837,
+      "step": 4151
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5355964199600345e-06,
+      "loss": 0.7705,
+      "step": 4152
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.531573569252512e-06,
+      "loss": 0.7762,
+      "step": 4153
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5275534497716057e-06,
+      "loss": 0.8013,
+      "step": 4154
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5235360629874806e-06,
+      "loss": 0.781,
+      "step": 4155
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.519521410369312e-06,
+      "loss": 0.7723,
+      "step": 4156
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5155094933852644e-06,
+      "loss": 0.7369,
+      "step": 4157
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.511500313502516e-06,
+      "loss": 0.7354,
+      "step": 4158
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.507493872187231e-06,
+      "loss": 0.7593,
+      "step": 4159
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5034901709045824e-06,
+      "loss": 0.2663,
+      "step": 4160
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.499489211118732e-06,
+      "loss": 0.7877,
+      "step": 4161
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4954909942928373e-06,
+      "loss": 0.8268,
+      "step": 4162
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4914955218890724e-06,
+      "loss": 0.7662,
+      "step": 4163
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4875027953685826e-06,
+      "loss": 0.7298,
+      "step": 4164
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.483512816191528e-06,
+      "loss": 0.7796,
+      "step": 4165
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4795255858170498e-06,
+      "loss": 0.754,
+      "step": 4166
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.475541105703293e-06,
+      "loss": 0.7933,
+      "step": 4167
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4715593773073977e-06,
+      "loss": 0.7728,
+      "step": 4168
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.467580402085488e-06,
+      "loss": 0.8002,
+      "step": 4169
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4636041814926935e-06,
+      "loss": 0.8183,
+      "step": 4170
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4596307169831246e-06,
+      "loss": 0.7464,
+      "step": 4171
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4556600100098925e-06,
+      "loss": 0.7528,
+      "step": 4172
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451692062025097e-06,
+      "loss": 0.7989,
+      "step": 4173
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.447726874479832e-06,
+      "loss": 0.7547,
+      "step": 4174
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.443764448824173e-06,
+      "loss": 0.7945,
+      "step": 4175
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.439804786507193e-06,
+      "loss": 0.7729,
+      "step": 4176
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4358478889769565e-06,
+      "loss": 0.7516,
+      "step": 4177
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.431893757680509e-06,
+      "loss": 0.7763,
+      "step": 4178
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.427942394063891e-06,
+      "loss": 0.8215,
+      "step": 4179
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.423993799572125e-06,
+      "loss": 0.7158,
+      "step": 4180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.420047975649227e-06,
+      "loss": 0.7726,
+      "step": 4181
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.416104923738196e-06,
+      "loss": 0.6648,
+      "step": 4182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.412164645281022e-06,
+      "loss": 0.7634,
+      "step": 4183
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4082271417186708e-06,
+      "loss": 0.7861,
+      "step": 4184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4042924144911053e-06,
+      "loss": 0.773,
+      "step": 4185
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4003604650372614e-06,
+      "loss": 0.7722,
+      "step": 4186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3964312947950694e-06,
+      "loss": 0.7159,
+      "step": 4187
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3925049052014404e-06,
+      "loss": 0.7688,
+      "step": 4188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.388581297692262e-06,
+      "loss": 0.8055,
+      "step": 4189
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3846604737024147e-06,
+      "loss": 0.7077,
+      "step": 4190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3807424346657483e-06,
+      "loss": 0.7511,
+      "step": 4191
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3768271820151123e-06,
+      "loss": 0.7715,
+      "step": 4192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3729147171823197e-06,
+      "loss": 0.755,
+      "step": 4193
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3690050415981747e-06,
+      "loss": 0.8163,
+      "step": 4194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3650981566924538e-06,
+      "loss": 0.7347,
+      "step": 4195
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.361194063893918e-06,
+      "loss": 0.8227,
+      "step": 4196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.35729276463031e-06,
+      "loss": 0.7489,
+      "step": 4197
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3533942603283423e-06,
+      "loss": 0.7346,
+      "step": 4198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3494985524137137e-06,
+      "loss": 0.7179,
+      "step": 4199
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3456056423110906e-06,
+      "loss": 0.7572,
+      "step": 4200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3417155314441274e-06,
+      "loss": 0.806,
+      "step": 4201
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3378282212354486e-06,
+      "loss": 0.8004,
+      "step": 4202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.333943713106658e-06,
+      "loss": 0.7989,
+      "step": 4203
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33006200847833e-06,
+      "loss": 0.8045,
+      "step": 4204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3261831087700114e-06,
+      "loss": 0.7169,
+      "step": 4205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3223070154002336e-06,
+      "loss": 0.7717,
+      "step": 4206
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3184337297864933e-06,
+      "loss": 0.7789,
+      "step": 4207
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.314563253345268e-06,
+      "loss": 0.6991,
+      "step": 4208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3106955874919944e-06,
+      "loss": 0.7242,
+      "step": 4209
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3068307336410956e-06,
+      "loss": 0.243,
+      "step": 4210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.302968693205961e-06,
+      "loss": 0.7639,
+      "step": 4211
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.299109467598948e-06,
+      "loss": 0.7846,
+      "step": 4212
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2952530582313904e-06,
+      "loss": 0.7092,
+      "step": 4213
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.291399466513584e-06,
+      "loss": 0.2511,
+      "step": 4214
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2875486938548032e-06,
+      "loss": 0.7153,
+      "step": 4215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.283700741663285e-06,
+      "loss": 0.7313,
+      "step": 4216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.279855611346242e-06,
+      "loss": 0.7216,
+      "step": 4217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2760133043098454e-06,
+      "loss": 0.7761,
+      "step": 4218
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2721738219592426e-06,
+      "loss": 0.7155,
+      "step": 4219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2683371656985387e-06,
+      "loss": 0.7904,
+      "step": 4220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2645033369308146e-06,
+      "loss": 0.8193,
+      "step": 4221
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2606723370581163e-06,
+      "loss": 0.762,
+      "step": 4222
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2568441674814446e-06,
+      "loss": 0.8472,
+      "step": 4223
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2530188296007804e-06,
+      "loss": 0.8006,
+      "step": 4224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2491963248150526e-06,
+      "loss": 0.7261,
+      "step": 4225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2453766545221754e-06,
+      "loss": 0.7833,
+      "step": 4226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2415598201190037e-06,
+      "loss": 0.807,
+      "step": 4227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2377458230013714e-06,
+      "loss": 0.7681,
+      "step": 4228
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2339346645640645e-06,
+      "loss": 0.7484,
+      "step": 4229
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2301263462008384e-06,
+      "loss": 0.7601,
+      "step": 4230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.226320869304408e-06,
+      "loss": 0.8026,
+      "step": 4231
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2225182352664445e-06,
+      "loss": 0.7236,
+      "step": 4232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2187184454775867e-06,
+      "loss": 0.7692,
+      "step": 4233
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.214921501327425e-06,
+      "loss": 0.8018,
+      "step": 4234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2111274042045163e-06,
+      "loss": 0.7771,
+      "step": 4235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2073361554963723e-06,
+      "loss": 0.7882,
+      "step": 4236
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.203547756589469e-06,
+      "loss": 0.74,
+      "step": 4237
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1997622088692292e-06,
+      "loss": 0.7943,
+      "step": 4238
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1959795137200423e-06,
+      "loss": 0.7465,
+      "step": 4239
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1921996725252536e-06,
+      "loss": 0.8048,
+      "step": 4240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1884226866671575e-06,
+      "loss": 0.7472,
+      "step": 4241
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.184648557527017e-06,
+      "loss": 0.7406,
+      "step": 4242
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1808772864850337e-06,
+      "loss": 0.2646,
+      "step": 4243
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1771088749203785e-06,
+      "loss": 0.7546,
+      "step": 4244
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1733433242111734e-06,
+      "loss": 0.7915,
+      "step": 4245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1695806357344852e-06,
+      "loss": 0.7411,
+      "step": 4246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165820810866347e-06,
+      "loss": 0.7403,
+      "step": 4247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1620638509817336e-06,
+      "loss": 0.7713,
+      "step": 4248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1583097574545787e-06,
+      "loss": 0.7916,
+      "step": 4249
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1545585316577657e-06,
+      "loss": 0.7321,
+      "step": 4250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.150810174963133e-06,
+      "loss": 0.7439,
+      "step": 4251
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1470646887414604e-06,
+      "loss": 0.7712,
+      "step": 4252
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1433220743624883e-06,
+      "loss": 0.7789,
+      "step": 4253
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1395823331948985e-06,
+      "loss": 0.7354,
+      "step": 4254
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.135845466606329e-06,
+      "loss": 0.7526,
+      "step": 4255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1321114759633633e-06,
+      "loss": 0.7581,
+      "step": 4256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1283803626315303e-06,
+      "loss": 0.8091,
+      "step": 4257
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1246521279753142e-06,
+      "loss": 0.7533,
+      "step": 4258
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.120926773358134e-06,
+      "loss": 0.7996,
+      "step": 4259
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.117204300142375e-06,
+      "loss": 0.7449,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1134847096893473e-06,
+      "loss": 0.7714,
+      "step": 4261
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109768003359324e-06,
+      "loss": 0.2428,
+      "step": 4262
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106054182511509e-06,
+      "loss": 0.7641,
+      "step": 4263
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1023432485040616e-06,
+      "loss": 0.7812,
+      "step": 4264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0986352026940848e-06,
+      "loss": 0.8069,
+      "step": 4265
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0949300464376166e-06,
+      "loss": 0.7825,
+      "step": 4266
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0912277810896486e-06,
+      "loss": 0.7667,
+      "step": 4267
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.087528408004107e-06,
+      "loss": 0.8293,
+      "step": 4268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.083831928533865e-06,
+      "loss": 0.7393,
+      "step": 4269
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0801383440307375e-06,
+      "loss": 0.7951,
+      "step": 4270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0764476558454836e-06,
+      "loss": 0.753,
+      "step": 4271
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0727598653277915e-06,
+      "loss": 0.2538,
+      "step": 4272
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0690749738263018e-06,
+      "loss": 0.7474,
+      "step": 4273
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0653929826885932e-06,
+      "loss": 0.7685,
+      "step": 4274
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.061713893261176e-06,
+      "loss": 0.6973,
+      "step": 4275
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0580377068895085e-06,
+      "loss": 0.7395,
+      "step": 4276
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0543644249179807e-06,
+      "loss": 0.7652,
+      "step": 4277
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.050694048689923e-06,
+      "loss": 0.7668,
+      "step": 4278
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047026579547605e-06,
+      "loss": 0.7698,
+      "step": 4279
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0433620188322347e-06,
+      "loss": 0.2441,
+      "step": 4280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0397003678839465e-06,
+      "loss": 0.7833,
+      "step": 4281
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.036041628041824e-06,
+      "loss": 0.7417,
+      "step": 4282
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0323858006438733e-06,
+      "loss": 0.755,
+      "step": 4283
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0287328870270438e-06,
+      "loss": 0.7559,
+      "step": 4284
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.025082888527222e-06,
+      "loss": 0.7351,
+      "step": 4285
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0214358064792173e-06,
+      "loss": 0.7688,
+      "step": 4286
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.017791642216784e-06,
+      "loss": 0.7232,
+      "step": 4287
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0141503970725997e-06,
+      "loss": 0.824,
+      "step": 4288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.01051207237828e-06,
+      "loss": 0.7806,
+      "step": 4289
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006876669464376e-06,
+      "loss": 0.8104,
+      "step": 4290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0032441896603594e-06,
+      "loss": 0.7626,
+      "step": 4291
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9996146342946453e-06,
+      "loss": 0.7942,
+      "step": 4292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9959880046945646e-06,
+      "loss": 0.7882,
+      "step": 4293
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.992364302186397e-06,
+      "loss": 0.7541,
+      "step": 4294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988743528095336e-06,
+      "loss": 0.7535,
+      "step": 4295
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9851256837455126e-06,
+      "loss": 0.7534,
+      "step": 4296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9815107704599803e-06,
+      "loss": 0.7726,
+      "step": 4297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9778987895607247e-06,
+      "loss": 0.8026,
+      "step": 4298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9742897423686636e-06,
+      "loss": 0.8033,
+      "step": 4299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9706836302036303e-06,
+      "loss": 0.7523,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.967080454384395e-06,
+      "loss": 0.7569,
+      "step": 4301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9634802162286436e-06,
+      "loss": 0.7738,
+      "step": 4302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9598829170530055e-06,
+      "loss": 0.7726,
+      "step": 4303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956288558173015e-06,
+      "loss": 0.7821,
+      "step": 4304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.952697140903146e-06,
+      "loss": 0.7302,
+      "step": 4305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.949108666556786e-06,
+      "loss": 0.7833,
+      "step": 4306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9455231364462533e-06,
+      "loss": 0.797,
+      "step": 4307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9419405518827894e-06,
+      "loss": 0.8175,
+      "step": 4308
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9383609141765524e-06,
+      "loss": 0.2497,
+      "step": 4309
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.934784224636632e-06,
+      "loss": 0.7445,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9312104845710287e-06,
+      "loss": 0.7385,
+      "step": 4311
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.927639695286673e-06,
+      "loss": 0.824,
+      "step": 4312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9240718580894146e-06,
+      "loss": 0.7953,
+      "step": 4313
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.920506974284023e-06,
+      "loss": 0.809,
+      "step": 4314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9169450451741843e-06,
+      "loss": 0.7748,
+      "step": 4315
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.913386072062512e-06,
+      "loss": 0.8022,
+      "step": 4316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 0.7491,
+      "step": 4317
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9062769990386776e-06,
+      "loss": 0.8156,
+      "step": 4318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9027269017263327e-06,
+      "loss": 0.8403,
+      "step": 4319
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.899179765611766e-06,
+      "loss": 0.7952,
+      "step": 4320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8956355919921832e-06,
+      "loss": 0.8125,
+      "step": 4321
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8920943821636917e-06,
+      "loss": 0.8126,
+      "step": 4322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8885561374213324e-06,
+      "loss": 0.7368,
+      "step": 4323
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8850208590590447e-06,
+      "loss": 0.7613,
+      "step": 4324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8814885483696954e-06,
+      "loss": 0.7427,
+      "step": 4325
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8779592066450602e-06,
+      "loss": 0.7627,
+      "step": 4326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8744328351758246e-06,
+      "loss": 0.7292,
+      "step": 4327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8709094352516033e-06,
+      "loss": 0.7591,
+      "step": 4328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8673890081609081e-06,
+      "loss": 0.7484,
+      "step": 4329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8638715551911735e-06,
+      "loss": 0.7854,
+      "step": 4330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8603570776287395e-06,
+      "loss": 0.7847,
+      "step": 4331
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8568455767588622e-06,
+      "loss": 0.7967,
+      "step": 4332
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8533370538657115e-06,
+      "loss": 0.7963,
+      "step": 4333
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8498315102323617e-06,
+      "loss": 0.8233,
+      "step": 4334
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846328947140803e-06,
+      "loss": 0.7954,
+      "step": 4335
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8428293658719288e-06,
+      "loss": 0.7568,
+      "step": 4336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.839332767705555e-06,
+      "loss": 0.7667,
+      "step": 4337
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8358391539203912e-06,
+      "loss": 0.744,
+      "step": 4338
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8323485257940677e-06,
+      "loss": 0.7996,
+      "step": 4339
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.828860884603113e-06,
+      "loss": 0.7985,
+      "step": 4340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8253762316229718e-06,
+      "loss": 0.7763,
+      "step": 4341
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8218945681279942e-06,
+      "loss": 0.7704,
+      "step": 4342
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.81841589539143e-06,
+      "loss": 0.718,
+      "step": 4343
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8149402146854455e-06,
+      "loss": 0.6907,
+      "step": 4344
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8114675272811033e-06,
+      "loss": 0.7424,
+      "step": 4345
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8079978344483774e-06,
+      "loss": 0.7722,
+      "step": 4346
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.804531137456147e-06,
+      "loss": 0.7976,
+      "step": 4347
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8010674375721959e-06,
+      "loss": 0.7265,
+      "step": 4348
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7976067360632033e-06,
+      "loss": 0.7866,
+      "step": 4349
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7941490341947654e-06,
+      "loss": 0.7656,
+      "step": 4350
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7906943332313687e-06,
+      "loss": 0.8373,
+      "step": 4351
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787242634436409e-06,
+      "loss": 0.7978,
+      "step": 4352
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.783793939072188e-06,
+      "loss": 0.717,
+      "step": 4353
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.780348248399899e-06,
+      "loss": 0.8423,
+      "step": 4354
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.776905563679645e-06,
+      "loss": 0.7799,
+      "step": 4355
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7734658861704202e-06,
+      "loss": 0.792,
+      "step": 4356
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7700292171301358e-06,
+      "loss": 0.8502,
+      "step": 4357
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7665955578155835e-06,
+      "loss": 0.7824,
+      "step": 4358
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7631649094824687e-06,
+      "loss": 0.7917,
+      "step": 4359
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7597372733853846e-06,
+      "loss": 0.7706,
+      "step": 4360
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7563126507778329e-06,
+      "loss": 0.2918,
+      "step": 4361
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7528910429122082e-06,
+      "loss": 0.7529,
+      "step": 4362
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7494724510397998e-06,
+      "loss": 0.7932,
+      "step": 4363
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.746056876410801e-06,
+      "loss": 0.7818,
+      "step": 4364
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7426443202742927e-06,
+      "loss": 0.7458,
+      "step": 4365
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7392347838782653e-06,
+      "loss": 0.7426,
+      "step": 4366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7358282684695925e-06,
+      "loss": 0.8558,
+      "step": 4367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.732424775294046e-06,
+      "loss": 0.7322,
+      "step": 4368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7290243055962962e-06,
+      "loss": 0.8208,
+      "step": 4369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7256268606199e-06,
+      "loss": 0.7381,
+      "step": 4370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7222324416073233e-06,
+      "loss": 0.7918,
+      "step": 4371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7188410497999065e-06,
+      "loss": 0.7883,
+      "step": 4372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7154526864378995e-06,
+      "loss": 0.7995,
+      "step": 4373
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7120673527604304e-06,
+      "loss": 0.8094,
+      "step": 4374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7086850500055297e-06,
+      "loss": 0.7687,
+      "step": 4375
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705305779410118e-06,
+      "loss": 0.7717,
+      "step": 4376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019295422100002e-06,
+      "loss": 0.8447,
+      "step": 4377
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6985563396398807e-06,
+      "loss": 0.7349,
+      "step": 4378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6951861729333464e-06,
+      "loss": 0.7596,
+      "step": 4379
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6918190433228798e-06,
+      "loss": 0.7703,
+      "step": 4380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6884549520398509e-06,
+      "loss": 0.813,
+      "step": 4381
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6850939003145195e-06,
+      "loss": 0.7732,
+      "step": 4382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6817358893760295e-06,
+      "loss": 0.8085,
+      "step": 4383
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6783809204524182e-06,
+      "loss": 0.7694,
+      "step": 4384
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6750289947706066e-06,
+      "loss": 0.717,
+      "step": 4385
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6716801135564043e-06,
+      "loss": 0.755,
+      "step": 4386
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6683342780345103e-06,
+      "loss": 0.8566,
+      "step": 4387
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6649914894285024e-06,
+      "loss": 0.7457,
+      "step": 4388
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.661651748960853e-06,
+      "loss": 0.7644,
+      "step": 4389
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6583150578529094e-06,
+      "loss": 0.7295,
+      "step": 4390
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.654981417324919e-06,
+      "loss": 0.8014,
+      "step": 4391
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6516508285959975e-06,
+      "loss": 0.804,
+      "step": 4392
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6483232928841552e-06,
+      "loss": 0.7619,
+      "step": 4393
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6449988114062787e-06,
+      "loss": 0.7291,
+      "step": 4394
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.641677385378142e-06,
+      "loss": 0.7794,
+      "step": 4395
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6383590160144047e-06,
+      "loss": 0.7829,
+      "step": 4396
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6350437045285993e-06,
+      "loss": 0.7618,
+      "step": 4397
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6317314521331505e-06,
+      "loss": 0.8294,
+      "step": 4398
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6284222600393517e-06,
+      "loss": 0.2434,
+      "step": 4399
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6251161294573947e-06,
+      "loss": 0.7898,
+      "step": 4400
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6218130615963346e-06,
+      "loss": 0.7442,
+      "step": 4401
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6185130576641173e-06,
+      "loss": 0.735,
+      "step": 4402
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6152161188675608e-06,
+      "loss": 0.7799,
+      "step": 4403
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6119222464123674e-06,
+      "loss": 0.747,
+      "step": 4404
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6086314415031191e-06,
+      "loss": 0.7738,
+      "step": 4405
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6053437053432685e-06,
+      "loss": 0.7836,
+      "step": 4406
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6020590391351564e-06,
+      "loss": 0.7156,
+      "step": 4407
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5987774440799897e-06,
+      "loss": 0.8073,
+      "step": 4408
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5954989213778593e-06,
+      "loss": 0.7633,
+      "step": 4409
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5922234722277362e-06,
+      "loss": 0.7409,
+      "step": 4410
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5889510978274558e-06,
+      "loss": 0.7713,
+      "step": 4411
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5856817993737394e-06,
+      "loss": 0.8283,
+      "step": 4412
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5824155780621776e-06,
+      "loss": 0.8023,
+      "step": 4413
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5791524350872368e-06,
+      "loss": 0.7855,
+      "step": 4414
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5758923716422602e-06,
+      "loss": 0.8279,
+      "step": 4415
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5726353889194657e-06,
+      "loss": 0.8033,
+      "step": 4416
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5693814881099368e-06,
+      "loss": 0.6905,
+      "step": 4417
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5661306704036394e-06,
+      "loss": 0.7306,
+      "step": 4418
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5628829369894027e-06,
+      "loss": 0.7905,
+      "step": 4419
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5596382890549355e-06,
+      "loss": 0.7914,
+      "step": 4420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5563967277868186e-06,
+      "loss": 0.7917,
+      "step": 4421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.553158254370496e-06,
+      "loss": 0.7671,
+      "step": 4422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5499228699902912e-06,
+      "loss": 0.7945,
+      "step": 4423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5466905758293881e-06,
+      "loss": 0.7148,
+      "step": 4424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5434613730698566e-06,
+      "loss": 0.7779,
+      "step": 4425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5402352628926176e-06,
+      "loss": 0.7265,
+      "step": 4426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.537012246477474e-06,
+      "loss": 0.7772,
+      "step": 4427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5337923250030907e-06,
+      "loss": 0.7591,
+      "step": 4428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5305754996470023e-06,
+      "loss": 0.7858,
+      "step": 4429
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.527361771585617e-06,
+      "loss": 0.7627,
+      "step": 4430
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.524151141994198e-06,
+      "loss": 0.8138,
+      "step": 4431
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5209436120468879e-06,
+      "loss": 0.7749,
+      "step": 4432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5177391829166844e-06,
+      "loss": 0.7223,
+      "step": 4433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5145378557754642e-06,
+      "loss": 0.7806,
+      "step": 4434
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5113396317939578e-06,
+      "loss": 0.7552,
+      "step": 4435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5081445121417682e-06,
+      "loss": 0.7451,
+      "step": 4436
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5049524979873574e-06,
+      "loss": 0.7665,
+      "step": 4437
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5017635904980577e-06,
+      "loss": 0.7765,
+      "step": 4438
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4985777908400634e-06,
+      "loss": 0.7438,
+      "step": 4439
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4953951001784261e-06,
+      "loss": 0.7726,
+      "step": 4440
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4922155196770715e-06,
+      "loss": 0.8106,
+      "step": 4441
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4890390504987773e-06,
+      "loss": 0.8153,
+      "step": 4442
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4858656938051897e-06,
+      "loss": 0.7157,
+      "step": 4443
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.482695450756816e-06,
+      "loss": 0.7874,
+      "step": 4444
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4795283225130252e-06,
+      "loss": 0.7686,
+      "step": 4445
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4763643102320414e-06,
+      "loss": 0.2586,
+      "step": 4446
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4732034150709595e-06,
+      "loss": 0.8108,
+      "step": 4447
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.470045638185722e-06,
+      "loss": 0.7458,
+      "step": 4448
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4668909807311415e-06,
+      "loss": 0.7862,
+      "step": 4449
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4637394438608888e-06,
+      "loss": 0.8006,
+      "step": 4450
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4605910287274839e-06,
+      "loss": 0.7658,
+      "step": 4451
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.457445736482318e-06,
+      "loss": 0.7738,
+      "step": 4452
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.454303568275629e-06,
+      "loss": 0.7699,
+      "step": 4453
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4511645252565209e-06,
+      "loss": 0.7274,
+      "step": 4454
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4480286085729523e-06,
+      "loss": 0.8023,
+      "step": 4455
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4448958193717345e-06,
+      "loss": 0.7709,
+      "step": 4456
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.441766158798541e-06,
+      "loss": 0.7838,
+      "step": 4457
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.438639627997893e-06,
+      "loss": 0.7836,
+      "step": 4458
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4355162281131806e-06,
+      "loss": 0.7449,
+      "step": 4459
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4323959602866355e-06,
+      "loss": 0.7914,
+      "step": 4460
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4292788256593516e-06,
+      "loss": 0.2823,
+      "step": 4461
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4261648253712713e-06,
+      "loss": 0.8104,
+      "step": 4462
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.423053960561197e-06,
+      "loss": 0.7592,
+      "step": 4463
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4199462323667824e-06,
+      "loss": 0.7209,
+      "step": 4464
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4168416419245302e-06,
+      "loss": 0.7015,
+      "step": 4465
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4137401903698023e-06,
+      "loss": 0.7901,
+      "step": 4466
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4106418788368027e-06,
+      "loss": 0.7916,
+      "step": 4467
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4075467084586026e-06,
+      "loss": 0.8205,
+      "step": 4468
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4044546803671077e-06,
+      "loss": 0.7671,
+      "step": 4469
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4013657956930883e-06,
+      "loss": 0.7425,
+      "step": 4470
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3982800555661536e-06,
+      "loss": 0.8184,
+      "step": 4471
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.39519746111477e-06,
+      "loss": 0.7977,
+      "step": 4472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3921180134662548e-06,
+      "loss": 0.8208,
+      "step": 4473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3890417137467681e-06,
+      "loss": 0.7975,
+      "step": 4474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3859685630813246e-06,
+      "loss": 0.7605,
+      "step": 4475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.382898562593782e-06,
+      "loss": 0.7414,
+      "step": 4476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3798317134068505e-06,
+      "loss": 0.8049,
+      "step": 4477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.376768016642086e-06,
+      "loss": 0.7817,
+      "step": 4478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3737074734198952e-06,
+      "loss": 0.806,
+      "step": 4479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3706500848595227e-06,
+      "loss": 0.7655,
+      "step": 4480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.36759585207907e-06,
+      "loss": 0.7526,
+      "step": 4481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3645447761954745e-06,
+      "loss": 0.7645,
+      "step": 4482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.361496858324527e-06,
+      "loss": 0.7575,
+      "step": 4483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3584520995808626e-06,
+      "loss": 0.7427,
+      "step": 4484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.355410501077954e-06,
+      "loss": 0.8147,
+      "step": 4485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3523720639281278e-06,
+      "loss": 0.7539,
+      "step": 4486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3493367892425435e-06,
+      "loss": 0.7644,
+      "step": 4487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3463046781312195e-06,
+      "loss": 0.7553,
+      "step": 4488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.343275731703e-06,
+      "loss": 0.7854,
+      "step": 4489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3402499510655863e-06,
+      "loss": 0.7904,
+      "step": 4490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3372273373255108e-06,
+      "loss": 0.7429,
+      "step": 4491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.334207891588154e-06,
+      "loss": 0.7616,
+      "step": 4492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.331191614957741e-06,
+      "loss": 0.7818,
+      "step": 4493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3281785085373255e-06,
+      "loss": 0.7869,
+      "step": 4494
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3251685734288177e-06,
+      "loss": 0.7465,
+      "step": 4495
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3221618107329537e-06,
+      "loss": 0.8069,
+      "step": 4496
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3191582215493193e-06,
+      "loss": 0.7432,
+      "step": 4497
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.316157806976337e-06,
+      "loss": 0.8027,
+      "step": 4498
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3131605681112646e-06,
+      "loss": 0.7354,
+      "step": 4499
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3101665060502056e-06,
+      "loss": 0.8394,
+      "step": 4500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3071756218880915e-06,
+      "loss": 0.785,
+      "step": 4501
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3041879167187065e-06,
+      "loss": 0.7875,
+      "step": 4502
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3012033916346566e-06,
+      "loss": 0.8218,
+      "step": 4503
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2982220477273965e-06,
+      "loss": 0.7762,
+      "step": 4504
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2952438860872085e-06,
+      "loss": 0.7783,
+      "step": 4505
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.292268907803218e-06,
+      "loss": 0.7637,
+      "step": 4506
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2892971139633847e-06,
+      "loss": 0.7739,
+      "step": 4507
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2863285056545005e-06,
+      "loss": 0.8004,
+      "step": 4508
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.283363083962198e-06,
+      "loss": 0.7246,
+      "step": 4509
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2804008499709353e-06,
+      "loss": 0.7212,
+      "step": 4510
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2774418047640147e-06,
+      "loss": 0.7658,
+      "step": 4511
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2744859494235673e-06,
+      "loss": 0.7565,
+      "step": 4512
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2715332850305607e-06,
+      "loss": 0.8169,
+      "step": 4513
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2685838126647897e-06,
+      "loss": 0.7238,
+      "step": 4514
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2656375334048886e-06,
+      "loss": 0.2568,
+      "step": 4515
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2626944483283176e-06,
+      "loss": 0.8141,
+      "step": 4516
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2597545585113736e-06,
+      "loss": 0.7943,
+      "step": 4517
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2568178650291863e-06,
+      "loss": 0.7082,
+      "step": 4518
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.253884368955708e-06,
+      "loss": 0.767,
+      "step": 4519
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2509540713637313e-06,
+      "loss": 0.7945,
+      "step": 4520
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2480269733248707e-06,
+      "loss": 0.2525,
+      "step": 4521
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2451030759095827e-06,
+      "loss": 0.2791,
+      "step": 4522
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.242182380187138e-06,
+      "loss": 0.8031,
+      "step": 4523
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2392648872256497e-06,
+      "loss": 0.7619,
+      "step": 4524
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.23635059809205e-06,
+      "loss": 0.2537,
+      "step": 4525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.233439513852104e-06,
+      "loss": 0.754,
+      "step": 4526
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2305316355704066e-06,
+      "loss": 0.811,
+      "step": 4527
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2276269643103744e-06,
+      "loss": 0.7621,
+      "step": 4528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2247255011342596e-06,
+      "loss": 0.7808,
+      "step": 4529
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.221827247103128e-06,
+      "loss": 0.7339,
+      "step": 4530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2189322032768901e-06,
+      "loss": 0.8177,
+      "step": 4531
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2160403707142675e-06,
+      "loss": 0.8335,
+      "step": 4532
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2131517504728097e-06,
+      "loss": 0.7887,
+      "step": 4533
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.210266343608899e-06,
+      "loss": 0.8027,
+      "step": 4534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2073841511777317e-06,
+      "loss": 0.7614,
+      "step": 4535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2045051742333413e-06,
+      "loss": 0.8331,
+      "step": 4536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2016294138285733e-06,
+      "loss": 0.8278,
+      "step": 4537
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1987568710151066e-06,
+      "loss": 0.2693,
+      "step": 4538
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.195887546843435e-06,
+      "loss": 0.8076,
+      "step": 4539
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1930214423628794e-06,
+      "loss": 0.75,
+      "step": 4540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.190158558621587e-06,
+      "loss": 0.2435,
+      "step": 4541
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1872988966665189e-06,
+      "loss": 0.2518,
+      "step": 4542
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1844424575434665e-06,
+      "loss": 0.7527,
+      "step": 4543
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1815892422970333e-06,
+      "loss": 0.7526,
+      "step": 4544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1787392519706509e-06,
+      "loss": 0.8106,
+      "step": 4545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1758924876065713e-06,
+      "loss": 0.7829,
+      "step": 4546
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1730489502458652e-06,
+      "loss": 0.7313,
+      "step": 4547
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1702086409284208e-06,
+      "loss": 0.7314,
+      "step": 4548
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1673715606929502e-06,
+      "loss": 0.7217,
+      "step": 4549
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1645377105769796e-06,
+      "loss": 0.7271,
+      "step": 4550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1617070916168583e-06,
+      "loss": 0.7988,
+      "step": 4551
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.158879704847754e-06,
+      "loss": 0.77,
+      "step": 4552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1560555513036476e-06,
+      "loss": 0.7896,
+      "step": 4553
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1532346320173416e-06,
+      "loss": 0.7747,
+      "step": 4554
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1504169480204564e-06,
+      "loss": 0.7135,
+      "step": 4555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1476025003434289e-06,
+      "loss": 0.7528,
+      "step": 4556
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1447912900155066e-06,
+      "loss": 0.8183,
+      "step": 4557
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1419833180647632e-06,
+      "loss": 0.7319,
+      "step": 4558
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1391785855180781e-06,
+      "loss": 0.803,
+      "step": 4559
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1363770934011532e-06,
+      "loss": 0.8624,
+      "step": 4560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.133578842738503e-06,
+      "loss": 0.8344,
+      "step": 4561
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.130783834553455e-06,
+      "loss": 0.7697,
+      "step": 4562
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1279920698681545e-06,
+      "loss": 0.7919,
+      "step": 4563
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1252035497035529e-06,
+      "loss": 0.7396,
+      "step": 4564
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1224182750794288e-06,
+      "loss": 0.6807,
+      "step": 4565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1196362470143597e-06,
+      "loss": 0.7081,
+      "step": 4566
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1168574665257448e-06,
+      "loss": 0.786,
+      "step": 4567
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1140819346297916e-06,
+      "loss": 0.8116,
+      "step": 4568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.111309652341519e-06,
+      "loss": 0.7659,
+      "step": 4569
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.108540620674764e-06,
+      "loss": 0.746,
+      "step": 4570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1057748406421654e-06,
+      "loss": 0.7351,
+      "step": 4571
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1030123132551807e-06,
+      "loss": 0.7261,
+      "step": 4572
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1002530395240706e-06,
+      "loss": 0.7549,
+      "step": 4573
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0974970204579126e-06,
+      "loss": 0.8219,
+      "step": 4574
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0947442570645939e-06,
+      "loss": 0.7675,
+      "step": 4575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0919947503508043e-06,
+      "loss": 0.806,
+      "step": 4576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.08924850132205e-06,
+      "loss": 0.7241,
+      "step": 4577
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0865055109826395e-06,
+      "loss": 0.7597,
+      "step": 4578
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.083765780335695e-06,
+      "loss": 0.7697,
+      "step": 4579
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0810293103831438e-06,
+      "loss": 0.7334,
+      "step": 4580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0782961021257244e-06,
+      "loss": 0.7848,
+      "step": 4581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0755661565629739e-06,
+      "loss": 0.778,
+      "step": 4582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0728394746932469e-06,
+      "loss": 0.798,
+      "step": 4583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0701160575136959e-06,
+      "loss": 0.803,
+      "step": 4584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0673959060202842e-06,
+      "loss": 0.7341,
+      "step": 4585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0646790212077818e-06,
+      "loss": 0.763,
+      "step": 4586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.061965404069758e-06,
+      "loss": 0.784,
+      "step": 4587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0592550555985926e-06,
+      "loss": 0.7232,
+      "step": 4588
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0565479767854693e-06,
+      "loss": 0.7756,
+      "step": 4589
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0538441686203749e-06,
+      "loss": 0.7484,
+      "step": 4590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0511436320920986e-06,
+      "loss": 0.8264,
+      "step": 4591
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0484463681882396e-06,
+      "loss": 0.7732,
+      "step": 4592
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0457523778951895e-06,
+      "loss": 0.724,
+      "step": 4593
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0430616621981516e-06,
+      "loss": 0.7246,
+      "step": 4594
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0403742220811308e-06,
+      "loss": 0.778,
+      "step": 4595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.037690058526929e-06,
+      "loss": 0.7097,
+      "step": 4596
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0350091725171562e-06,
+      "loss": 0.773,
+      "step": 4597
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0323315650322164e-06,
+      "loss": 0.8191,
+      "step": 4598
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0296572370513247e-06,
+      "loss": 0.7866,
+      "step": 4599
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0269861895524858e-06,
+      "loss": 0.8006,
+      "step": 4600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0243184235125148e-06,
+      "loss": 0.7944,
+      "step": 4601
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0216539399070168e-06,
+      "loss": 0.8294,
+      "step": 4602
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.018992739710405e-06,
+      "loss": 0.7565,
+      "step": 4603
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0163348238958904e-06,
+      "loss": 0.814,
+      "step": 4604
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0136801934354757e-06,
+      "loss": 0.7799,
+      "step": 4605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.011028849299972e-06,
+      "loss": 0.728,
+      "step": 4606
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0083807924589806e-06,
+      "loss": 0.7822,
+      "step": 4607
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0057360238809067e-06,
+      "loss": 0.7925,
+      "step": 4608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0030945445329488e-06,
+      "loss": 0.8064,
+      "step": 4609
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0004563553811063e-06,
+      "loss": 0.8004,
+      "step": 4610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.97821457390169e-07,
+      "loss": 0.7651,
+      "step": 4611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.951898515237323e-07,
+      "loss": 0.7969,
+      "step": 4612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.925615387441789e-07,
+      "loss": 0.2777,
+      "step": 4613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.899365200126908e-07,
+      "loss": 0.7599,
+      "step": 4614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.873147962892494e-07,
+      "loss": 0.7982,
+      "step": 4615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.846963685326228e-07,
+      "loss": 0.8602,
+      "step": 4616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.820812377003786e-07,
+      "loss": 0.7574,
+      "step": 4617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.794694047488818e-07,
+      "loss": 0.7436,
+      "step": 4618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.768608706332838e-07,
+      "loss": 0.7898,
+      "step": 4619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.742556363075373e-07,
+      "loss": 0.8072,
+      "step": 4620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.7165370272438e-07,
+      "loss": 0.7396,
+      "step": 4621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.690550708353486e-07,
+      "loss": 0.7182,
+      "step": 4622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.664597415907728e-07,
+      "loss": 0.7483,
+      "step": 4623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.638677159397736e-07,
+      "loss": 0.8076,
+      "step": 4624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.61278994830257e-07,
+      "loss": 0.7736,
+      "step": 4625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.586935792089335e-07,
+      "loss": 0.802,
+      "step": 4626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.561114700212914e-07,
+      "loss": 0.7824,
+      "step": 4627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.535326682116175e-07,
+      "loss": 0.2493,
+      "step": 4628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.509571747229918e-07,
+      "loss": 0.7162,
+      "step": 4629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.483849904972753e-07,
+      "loss": 0.7114,
+      "step": 4630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.458161164751267e-07,
+      "loss": 0.279,
+      "step": 4631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.432505535959857e-07,
+      "loss": 0.7712,
+      "step": 4632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.406883027980951e-07,
+      "loss": 0.781,
+      "step": 4633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.38129365018472e-07,
+      "loss": 0.7131,
+      "step": 4634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.35573741192931e-07,
+      "loss": 0.772,
+      "step": 4635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.330214322560693e-07,
+      "loss": 0.24,
+      "step": 4636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.304724391412733e-07,
+      "loss": 0.732,
+      "step": 4637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.27926762780722e-07,
+      "loss": 0.7456,
+      "step": 4638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.253844041053727e-07,
+      "loss": 0.818,
+      "step": 4639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.228453640449774e-07,
+      "loss": 0.7789,
+      "step": 4640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.203096435280667e-07,
+      "loss": 0.7713,
+      "step": 4641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.177772434819642e-07,
+      "loss": 0.7223,
+      "step": 4642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.15248164832776e-07,
+      "loss": 0.7669,
+      "step": 4643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.127224085053954e-07,
+      "loss": 0.7211,
+      "step": 4644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.101999754234969e-07,
+      "loss": 0.7463,
+      "step": 4645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.076808665095437e-07,
+      "loss": 0.7944,
+      "step": 4646
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.051650826847802e-07,
+      "loss": 0.8005,
+      "step": 4647
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.026526248692369e-07,
+      "loss": 0.8044,
+      "step": 4648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.001434939817288e-07,
+      "loss": 0.7738,
+      "step": 4649
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.976376909398499e-07,
+      "loss": 0.7278,
+      "step": 4650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.951352166599825e-07,
+      "loss": 0.7543,
+      "step": 4651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.926360720572879e-07,
+      "loss": 0.7356,
+      "step": 4652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.901402580457141e-07,
+      "loss": 0.816,
+      "step": 4653
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.876477755379831e-07,
+      "loss": 0.7667,
+      "step": 4654
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.851586254456079e-07,
+      "loss": 0.7737,
+      "step": 4655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.82672808678875e-07,
+      "loss": 0.7453,
+      "step": 4656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.801903261468558e-07,
+      "loss": 0.787,
+      "step": 4657
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.777111787574055e-07,
+      "loss": 0.7812,
+      "step": 4658
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.752353674171498e-07,
+      "loss": 0.7713,
+      "step": 4659
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.727628930315069e-07,
+      "loss": 0.7603,
+      "step": 4660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.702937565046631e-07,
+      "loss": 0.8062,
+      "step": 4661
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.678279587395922e-07,
+      "loss": 0.8118,
+      "step": 4662
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.653655006380435e-07,
+      "loss": 0.7524,
+      "step": 4663
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.629063831005446e-07,
+      "loss": 0.239,
+      "step": 4664
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.604506070264051e-07,
+      "loss": 0.8129,
+      "step": 4665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.579981733137032e-07,
+      "loss": 0.7682,
+      "step": 4666
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.555490828593104e-07,
+      "loss": 0.7768,
+      "step": 4667
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.531033365588615e-07,
+      "loss": 0.7274,
+      "step": 4668
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.506609353067751e-07,
+      "loss": 0.7765,
+      "step": 4669
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.482218799962427e-07,
+      "loss": 0.8129,
+      "step": 4670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.457861715192361e-07,
+      "loss": 0.7467,
+      "step": 4671
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.433538107665029e-07,
+      "loss": 0.753,
+      "step": 4672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.409247986275614e-07,
+      "loss": 0.7931,
+      "step": 4673
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.384991359907113e-07,
+      "loss": 0.7197,
+      "step": 4674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.360768237430239e-07,
+      "loss": 0.7879,
+      "step": 4675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.336578627703462e-07,
+      "loss": 0.7481,
+      "step": 4676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.312422539573006e-07,
+      "loss": 0.7338,
+      "step": 4677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.288299981872827e-07,
+      "loss": 0.7901,
+      "step": 4678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.264210963424613e-07,
+      "loss": 0.7464,
+      "step": 4679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.240155493037782e-07,
+      "loss": 0.7329,
+      "step": 4680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.216133579509522e-07,
+      "loss": 0.7491,
+      "step": 4681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.192145231624693e-07,
+      "loss": 0.7682,
+      "step": 4682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.168190458155933e-07,
+      "loss": 0.7601,
+      "step": 4683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.144269267863547e-07,
+      "loss": 0.7347,
+      "step": 4684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.120381669495603e-07,
+      "loss": 0.7237,
+      "step": 4685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.096527671787879e-07,
+      "loss": 0.7499,
+      "step": 4686
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.072707283463854e-07,
+      "loss": 0.7874,
+      "step": 4687
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.048920513234703e-07,
+      "loss": 0.7218,
+      "step": 4688
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.025167369799336e-07,
+      "loss": 0.776,
+      "step": 4689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.001447861844336e-07,
+      "loss": 0.7601,
+      "step": 4690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.977761998043998e-07,
+      "loss": 0.7625,
+      "step": 4691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.95410978706036e-07,
+      "loss": 0.2385,
+      "step": 4692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.930491237543047e-07,
+      "loss": 0.8034,
+      "step": 4693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.906906358129474e-07,
+      "loss": 0.2456,
+      "step": 4694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.883355157444672e-07,
+      "loss": 0.7915,
+      "step": 4695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.859837644101431e-07,
+      "loss": 0.7701,
+      "step": 4696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.83635382670016e-07,
+      "loss": 0.7876,
+      "step": 4697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.812903713828957e-07,
+      "loss": 0.7744,
+      "step": 4698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.789487314063604e-07,
+      "loss": 0.2425,
+      "step": 4699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.766104635967519e-07,
+      "loss": 0.7202,
+      "step": 4700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.742755688091897e-07,
+      "loss": 0.792,
+      "step": 4701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.719440478975449e-07,
+      "loss": 0.7132,
+      "step": 4702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.696159017144655e-07,
+      "loss": 0.7912,
+      "step": 4703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.672911311113596e-07,
+      "loss": 0.7414,
+      "step": 4704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.649697369384046e-07,
+      "loss": 0.7564,
+      "step": 4705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.626517200445416e-07,
+      "loss": 0.7726,
+      "step": 4706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.603370812774746e-07,
+      "loss": 0.7783,
+      "step": 4707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.580258214836778e-07,
+      "loss": 0.7874,
+      "step": 4708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.557179415083815e-07,
+      "loss": 0.7775,
+      "step": 4709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.534134421955874e-07,
+      "loss": 0.7577,
+      "step": 4710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.511123243880569e-07,
+      "loss": 0.7031,
+      "step": 4711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.488145889273179e-07,
+      "loss": 0.8258,
+      "step": 4712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.465202366536562e-07,
+      "loss": 0.8105,
+      "step": 4713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.442292684061258e-07,
+      "loss": 0.7741,
+      "step": 4714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.419416850225414e-07,
+      "loss": 0.8383,
+      "step": 4715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.396574873394757e-07,
+      "loss": 0.7464,
+      "step": 4716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.373766761922718e-07,
+      "loss": 0.7335,
+      "step": 4717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.350992524150258e-07,
+      "loss": 0.7827,
+      "step": 4718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.32825216840598e-07,
+      "loss": 0.8226,
+      "step": 4719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.305545703006134e-07,
+      "loss": 0.7704,
+      "step": 4720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.282873136254542e-07,
+      "loss": 0.761,
+      "step": 4721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.260234476442596e-07,
+      "loss": 0.8105,
+      "step": 4722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.237629731849372e-07,
+      "loss": 0.7642,
+      "step": 4723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.215058910741446e-07,
+      "loss": 0.7841,
+      "step": 4724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.19252202137306e-07,
+      "loss": 0.747,
+      "step": 4725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.170019071986046e-07,
+      "loss": 0.7822,
+      "step": 4726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.147550070809773e-07,
+      "loss": 0.7183,
+      "step": 4727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.125115026061236e-07,
+      "loss": 0.7886,
+      "step": 4728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.102713945944961e-07,
+      "loss": 0.7719,
+      "step": 4729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.080346838653174e-07,
+      "loss": 0.7043,
+      "step": 4730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.058013712365519e-07,
+      "loss": 0.2431,
+      "step": 4731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.035714575249342e-07,
+      "loss": 0.7394,
+      "step": 4732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.013449435459463e-07,
+      "loss": 0.7644,
+      "step": 4733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.991218301138336e-07,
+      "loss": 0.747,
+      "step": 4734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.96902118041597e-07,
+      "loss": 0.2229,
+      "step": 4735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.94685808140988e-07,
+      "loss": 0.7386,
+      "step": 4736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.924729012225229e-07,
+      "loss": 0.8567,
+      "step": 4737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.902633980954631e-07,
+      "loss": 0.7447,
+      "step": 4738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.880572995678347e-07,
+      "loss": 0.7708,
+      "step": 4739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.858546064464145e-07,
+      "loss": 0.7999,
+      "step": 4740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.836553195367324e-07,
+      "loss": 0.7671,
+      "step": 4741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.814594396430763e-07,
+      "loss": 0.2777,
+      "step": 4742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.792669675684815e-07,
+      "loss": 0.7815,
+      "step": 4743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.770779041147501e-07,
+      "loss": 0.7697,
+      "step": 4744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.748922500824229e-07,
+      "loss": 0.2518,
+      "step": 4745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.727100062708025e-07,
+      "loss": 0.7759,
+      "step": 4746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.705311734779418e-07,
+      "loss": 0.7554,
+      "step": 4747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.683557525006468e-07,
+      "loss": 0.7849,
+      "step": 4748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.661837441344765e-07,
+      "loss": 0.7373,
+      "step": 4749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.64015149173739e-07,
+      "loss": 0.7968,
+      "step": 4750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.618499684114999e-07,
+      "loss": 0.7732,
+      "step": 4751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.596882026395679e-07,
+      "loss": 0.7754,
+      "step": 4752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.575298526485096e-07,
+      "loss": 0.729,
+      "step": 4753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.553749192276393e-07,
+      "loss": 0.7829,
+      "step": 4754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.532234031650264e-07,
+      "loss": 0.7434,
+      "step": 4755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.510753052474828e-07,
+      "loss": 0.7845,
+      "step": 4756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.489306262605777e-07,
+      "loss": 0.7911,
+      "step": 4757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.467893669886227e-07,
+      "loss": 0.7838,
+      "step": 4758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.446515282146859e-07,
+      "loss": 0.7781,
+      "step": 4759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.42517110720583e-07,
+      "loss": 0.791,
+      "step": 4760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.403861152868729e-07,
+      "loss": 0.732,
+      "step": 4761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.382585426928723e-07,
+      "loss": 0.2494,
+      "step": 4762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.361343937166331e-07,
+      "loss": 0.8084,
+      "step": 4763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.34013669134973e-07,
+      "loss": 0.7291,
+      "step": 4764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.318963697234404e-07,
+      "loss": 0.7593,
+      "step": 4765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.297824962563426e-07,
+      "loss": 0.7566,
+      "step": 4766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.276720495067257e-07,
+      "loss": 0.7924,
+      "step": 4767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.25565030246389e-07,
+      "loss": 0.7791,
+      "step": 4768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.234614392458771e-07,
+      "loss": 0.8021,
+      "step": 4769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.21361277274476e-07,
+      "loss": 0.8239,
+      "step": 4770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.192645451002255e-07,
+      "loss": 0.6962,
+      "step": 4771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.171712434899035e-07,
+      "loss": 0.7568,
+      "step": 4772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.150813732090389e-07,
+      "loss": 0.7458,
+      "step": 4773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.129949350219034e-07,
+      "loss": 0.2281,
+      "step": 4774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.109119296915144e-07,
+      "loss": 0.7817,
+      "step": 4775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.088323579796318e-07,
+      "loss": 0.7615,
+      "step": 4776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.067562206467626e-07,
+      "loss": 0.8459,
+      "step": 4777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.046835184521571e-07,
+      "loss": 0.8322,
+      "step": 4778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.02614252153807e-07,
+      "loss": 0.7636,
+      "step": 4779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.005484225084524e-07,
+      "loss": 0.7557,
+      "step": 4780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.9848603027157e-07,
+      "loss": 0.6786,
+      "step": 4781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.964270761973845e-07,
+      "loss": 0.7368,
+      "step": 4782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.943715610388634e-07,
+      "loss": 0.2648,
+      "step": 4783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.923194855477121e-07,
+      "loss": 0.8039,
+      "step": 4784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.902708504743848e-07,
+      "loss": 0.7294,
+      "step": 4785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.882256565680688e-07,
+      "loss": 0.7916,
+      "step": 4786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.861839045767004e-07,
+      "loss": 0.8042,
+      "step": 4787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.841455952469555e-07,
+      "loss": 0.7616,
+      "step": 4788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.821107293242501e-07,
+      "loss": 0.8036,
+      "step": 4789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.800793075527378e-07,
+      "loss": 0.7683,
+      "step": 4790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7805133067532e-07,
+      "loss": 0.82,
+      "step": 4791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.760267994336311e-07,
+      "loss": 0.7873,
+      "step": 4792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.740057145680489e-07,
+      "loss": 0.7479,
+      "step": 4793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.719880768176922e-07,
+      "loss": 0.7706,
+      "step": 4794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.699738869204152e-07,
+      "loss": 0.7712,
+      "step": 4795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.679631456128154e-07,
+      "loss": 0.8103,
+      "step": 4796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.659558536302223e-07,
+      "loss": 0.7219,
+      "step": 4797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.639520117067165e-07,
+      "loss": 0.7611,
+      "step": 4798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.61951620575103e-07,
+      "loss": 0.7326,
+      "step": 4799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.599546809669342e-07,
+      "loss": 0.7863,
+      "step": 4800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.579611936124951e-07,
+      "loss": 0.7763,
+      "step": 4801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.559711592408112e-07,
+      "loss": 0.2486,
+      "step": 4802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.53984578579646e-07,
+      "loss": 0.7662,
+      "step": 4803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.52001452355494e-07,
+      "loss": 0.7924,
+      "step": 4804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.500217812935948e-07,
+      "loss": 0.7126,
+      "step": 4805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.480455661179151e-07,
+      "loss": 0.771,
+      "step": 4806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.46072807551169e-07,
+      "loss": 0.7588,
+      "step": 4807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.441035063147959e-07,
+      "loss": 0.6922,
+      "step": 4808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.421376631289788e-07,
+      "loss": 0.7801,
+      "step": 4809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.401752787126291e-07,
+      "loss": 0.7251,
+      "step": 4810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.382163537833984e-07,
+      "loss": 0.7938,
+      "step": 4811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.362608890576737e-07,
+      "loss": 0.8405,
+      "step": 4812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.343088852505696e-07,
+      "loss": 0.7476,
+      "step": 4813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.32360343075946e-07,
+      "loss": 0.8007,
+      "step": 4814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.304152632463855e-07,
+      "loss": 0.2528,
+      "step": 4815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.28473646473211e-07,
+      "loss": 0.7697,
+      "step": 4816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.265354934664802e-07,
+      "loss": 0.7238,
+      "step": 4817
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.246008049349805e-07,
+      "loss": 0.7249,
+      "step": 4818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.226695815862337e-07,
+      "loss": 0.8076,
+      "step": 4819
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.207418241264917e-07,
+      "loss": 0.7611,
+      "step": 4820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.188175332607426e-07,
+      "loss": 0.7158,
+      "step": 4821
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.168967096927069e-07,
+      "loss": 0.777,
+      "step": 4822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.14979354124836e-07,
+      "loss": 0.7932,
+      "step": 4823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.130654672583102e-07,
+      "loss": 0.7593,
+      "step": 4824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.111550497930473e-07,
+      "loss": 0.7715,
+      "step": 4825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.092481024276886e-07,
+      "loss": 0.2397,
+      "step": 4826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.073446258596126e-07,
+      "loss": 0.7863,
+      "step": 4827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.054446207849295e-07,
+      "loss": 0.8198,
+      "step": 4828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.035480878984722e-07,
+      "loss": 0.7871,
+      "step": 4829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.016550278938126e-07,
+      "loss": 0.7112,
+      "step": 4830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.997654414632436e-07,
+      "loss": 0.7843,
+      "step": 4831
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.97879329297799e-07,
+      "loss": 0.2134,
+      "step": 4832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.959966920872316e-07,
+      "loss": 0.7898,
+      "step": 4833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.941175305200297e-07,
+      "loss": 0.7419,
+      "step": 4834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.922418452834066e-07,
+      "loss": 0.7421,
+      "step": 4835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.903696370633071e-07,
+      "loss": 0.807,
+      "step": 4836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.885009065444057e-07,
+      "loss": 0.7225,
+      "step": 4837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.866356544100992e-07,
+      "loss": 0.7997,
+      "step": 4838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.847738813425184e-07,
+      "loss": 0.7252,
+      "step": 4839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.829155880225168e-07,
+      "loss": 0.8298,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.810607751296837e-07,
+      "loss": 0.7812,
+      "step": 4841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.792094433423233e-07,
+      "loss": 0.7704,
+      "step": 4842
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.773615933374787e-07,
+      "loss": 0.7752,
+      "step": 4843
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.755172257909102e-07,
+      "loss": 0.7249,
+      "step": 4844
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.736763413771106e-07,
+      "loss": 0.7413,
+      "step": 4845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.71838940769298e-07,
+      "loss": 0.7781,
+      "step": 4846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.700050246394128e-07,
+      "loss": 0.7392,
+      "step": 4847
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.681745936581272e-07,
+      "loss": 0.258,
+      "step": 4848
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.663476484948315e-07,
+      "loss": 0.7771,
+      "step": 4849
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.645241898176478e-07,
+      "loss": 0.7241,
+      "step": 4850
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.627042182934205e-07,
+      "loss": 0.7395,
+      "step": 4851
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.608877345877194e-07,
+      "loss": 0.2642,
+      "step": 4852
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.590747393648354e-07,
+      "loss": 0.7455,
+      "step": 4853
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.572652332877903e-07,
+      "loss": 0.7816,
+      "step": 4854
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5545921701832254e-07,
+      "loss": 0.7971,
+      "step": 4855
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5365669121689826e-07,
+      "loss": 0.7759,
+      "step": 4856
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.518576565427102e-07,
+      "loss": 0.7515,
+      "step": 4857
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.500621136536665e-07,
+      "loss": 0.8517,
+      "step": 4858
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.482700632064063e-07,
+      "loss": 0.7538,
+      "step": 4859
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4648150585628437e-07,
+      "loss": 0.7911,
+      "step": 4860
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4469644225738516e-07,
+      "loss": 0.7759,
+      "step": 4861
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.42914873062511e-07,
+      "loss": 0.7951,
+      "step": 4862
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4113679892318515e-07,
+      "loss": 0.7746,
+      "step": 4863
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.393622204896586e-07,
+      "loss": 0.7293,
+      "step": 4864
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.375911384108944e-07,
+      "loss": 0.7748,
+      "step": 4865
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3582355333458894e-07,
+      "loss": 0.7408,
+      "step": 4866
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3405946590714955e-07,
+      "loss": 0.8459,
+      "step": 4867
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.322988767737113e-07,
+      "loss": 0.8001,
+      "step": 4868
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.305417865781247e-07,
+      "loss": 0.7799,
+      "step": 4869
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.287881959629625e-07,
+      "loss": 0.7567,
+      "step": 4870
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.270381055695216e-07,
+      "loss": 0.7445,
+      "step": 4871
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2529151603781236e-07,
+      "loss": 0.7778,
+      "step": 4872
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2354842800657045e-07,
+      "loss": 0.7539,
+      "step": 4873
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.218088421132449e-07,
+      "loss": 0.7347,
+      "step": 4874
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2007275899401236e-07,
+      "loss": 0.676,
+      "step": 4875
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1834017928375936e-07,
+      "loss": 0.7766,
+      "step": 4876
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1661110361609916e-07,
+      "loss": 0.738,
+      "step": 4877
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.14885532623357e-07,
+      "loss": 0.7995,
+      "step": 4878
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.131634669365814e-07,
+      "loss": 0.7326,
+      "step": 4879
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1144490718553865e-07,
+      "loss": 0.8058,
+      "step": 4880
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.097298539987071e-07,
+      "loss": 0.7916,
+      "step": 4881
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.080183080032918e-07,
+      "loss": 0.7512,
+      "step": 4882
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.063102698252053e-07,
+      "loss": 0.8143,
+      "step": 4883
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.046057400890868e-07,
+      "loss": 0.7722,
+      "step": 4884
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.029047194182856e-07,
+      "loss": 0.6997,
+      "step": 4885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.012072084348717e-07,
+      "loss": 0.7619,
+      "step": 4886
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9951320775962867e-07,
+      "loss": 0.2468,
+      "step": 4887
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9782271801205975e-07,
+      "loss": 0.7344,
+      "step": 4888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9613573981038046e-07,
+      "loss": 0.7638,
+      "step": 4889
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9445227377152396e-07,
+      "loss": 0.7772,
+      "step": 4890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9277232051114e-07,
+      "loss": 0.7957,
+      "step": 4891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.910958806435916e-07,
+      "loss": 0.7327,
+      "step": 4892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8942295478196057e-07,
+      "loss": 0.7561,
+      "step": 4893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.877535435380364e-07,
+      "loss": 0.7149,
+      "step": 4894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8608764752233295e-07,
+      "loss": 0.7477,
+      "step": 4895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.844252673440696e-07,
+      "loss": 0.7824,
+      "step": 4896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.827664036111878e-07,
+      "loss": 0.7897,
+      "step": 4897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.811110569303345e-07,
+      "loss": 0.7315,
+      "step": 4898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.794592279068776e-07,
+      "loss": 0.7535,
+      "step": 4899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.778109171448985e-07,
+      "loss": 0.7331,
+      "step": 4900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.761661252471849e-07,
+      "loss": 0.7673,
+      "step": 4901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.745248528152456e-07,
+      "loss": 0.756,
+      "step": 4902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.728871004492951e-07,
+      "loss": 0.7383,
+      "step": 4903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.712528687482697e-07,
+      "loss": 0.751,
+      "step": 4904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.696221583098114e-07,
+      "loss": 0.782,
+      "step": 4905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6799496973027314e-07,
+      "loss": 0.7556,
+      "step": 4906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.663713036047256e-07,
+      "loss": 0.8024,
+      "step": 4907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6475116052694603e-07,
+      "loss": 0.739,
+      "step": 4908
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6313454108942937e-07,
+      "loss": 0.7743,
+      "step": 4909
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6152144588337603e-07,
+      "loss": 0.7913,
+      "step": 4910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.599118754987008e-07,
+      "loss": 0.7847,
+      "step": 4911
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5830583052402614e-07,
+      "loss": 0.7075,
+      "step": 4912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5670331154669e-07,
+      "loss": 0.7017,
+      "step": 4913
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5510431915273925e-07,
+      "loss": 0.741,
+      "step": 4914
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5350885392692823e-07,
+      "loss": 0.7763,
+      "step": 4915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5191691645272585e-07,
+      "loss": 0.7349,
+      "step": 4916
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5032850731230525e-07,
+      "loss": 0.7452,
+      "step": 4917
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.487436270865552e-07,
+      "loss": 0.7539,
+      "step": 4918
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471622763550697e-07,
+      "loss": 0.7729,
+      "step": 4919
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.455844556961563e-07,
+      "loss": 0.7717,
+      "step": 4920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4401016568682554e-07,
+      "loss": 0.7579,
+      "step": 4921
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4243940690280255e-07,
+      "loss": 0.7495,
+      "step": 4922
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.408721799185166e-07,
+      "loss": 0.7687,
+      "step": 4923
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3930848530710937e-07,
+      "loss": 0.8014,
+      "step": 4924
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3774832364042776e-07,
+      "loss": 0.7224,
+      "step": 4925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3619169548902765e-07,
+      "loss": 0.7387,
+      "step": 4926
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3463860142217477e-07,
+      "loss": 0.779,
+      "step": 4927
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.330890420078359e-07,
+      "loss": 0.769,
+      "step": 4928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.315430178126944e-07,
+      "loss": 0.7539,
+      "step": 4929
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.300005294021336e-07,
+      "loss": 0.7744,
+      "step": 4930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2846157734024776e-07,
+      "loss": 0.7331,
+      "step": 4931
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.269261621898345e-07,
+      "loss": 0.7682,
+      "step": 4932
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.253942845124003e-07,
+      "loss": 0.7929,
+      "step": 4933
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2386594486816023e-07,
+      "loss": 0.7826,
+      "step": 4934
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2234114381602845e-07,
+      "loss": 0.2518,
+      "step": 4935
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2081988191363433e-07,
+      "loss": 0.7581,
+      "step": 4936
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.193021597173018e-07,
+      "loss": 0.7647,
+      "step": 4937
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.177879777820736e-07,
+      "loss": 0.7798,
+      "step": 4938
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1627733666168556e-07,
+      "loss": 0.2552,
+      "step": 4939
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.14770236908587e-07,
+      "loss": 0.8207,
+      "step": 4940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1326667907392583e-07,
+      "loss": 0.7288,
+      "step": 4941
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1176666370756004e-07,
+      "loss": 0.8225,
+      "step": 4942
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.102701913580508e-07,
+      "loss": 0.7866,
+      "step": 4943
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.087772625726593e-07,
+      "loss": 0.7498,
+      "step": 4944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.072878778973576e-07,
+      "loss": 0.7419,
+      "step": 4945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.058020378768156e-07,
+      "loss": 0.7207,
+      "step": 4946
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0431974305441094e-07,
+      "loss": 0.7541,
+      "step": 4947
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0284099397222434e-07,
+      "loss": 0.7851,
+      "step": 4948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.013657911710377e-07,
+      "loss": 0.7779,
+      "step": 4949
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.998941351903373e-07,
+      "loss": 0.7712,
+      "step": 4950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9842602656831145e-07,
+      "loss": 0.8217,
+      "step": 4951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.969614658418529e-07,
+      "loss": 0.7704,
+      "step": 4952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9550045354655555e-07,
+      "loss": 0.7672,
+      "step": 4953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9404299021671857e-07,
+      "loss": 0.2405,
+      "step": 4954
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.925890763853367e-07,
+      "loss": 0.7592,
+      "step": 4955
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9113871258411475e-07,
+      "loss": 0.7425,
+      "step": 4956
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8969189934345164e-07,
+      "loss": 0.7222,
+      "step": 4957
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8824863719245423e-07,
+      "loss": 0.7542,
+      "step": 4958
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.868089266589269e-07,
+      "loss": 0.7292,
+      "step": 4959
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.853727682693752e-07,
+      "loss": 0.7839,
+      "step": 4960
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.83940162549009e-07,
+      "loss": 0.7072,
+      "step": 4961
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.825111100217326e-07,
+      "loss": 0.7598,
+      "step": 4962
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.810856112101612e-07,
+      "loss": 0.719,
+      "step": 4963
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7966366663559787e-07,
+      "loss": 0.7394,
+      "step": 4964
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.782452768180566e-07,
+      "loss": 0.7549,
+      "step": 4965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.768304422762447e-07,
+      "loss": 0.727,
+      "step": 4966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.754191635275705e-07,
+      "loss": 0.7479,
+      "step": 4967
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.740114410881467e-07,
+      "loss": 0.7538,
+      "step": 4968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7260727547277823e-07,
+      "loss": 0.751,
+      "step": 4969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.712066671949753e-07,
+      "loss": 0.8282,
+      "step": 4970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.698096167669406e-07,
+      "loss": 0.7651,
+      "step": 4971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6841612469958536e-07,
+      "loss": 0.7434,
+      "step": 4972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6702619150250963e-07,
+      "loss": 0.7914,
+      "step": 4973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6563981768402025e-07,
+      "loss": 0.8073,
+      "step": 4974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6425700375111387e-07,
+      "loss": 0.7788,
+      "step": 4975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6287775020949256e-07,
+      "loss": 0.7308,
+      "step": 4976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6150205756355294e-07,
+      "loss": 0.7829,
+      "step": 4977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6012992631639035e-07,
+      "loss": 0.6892,
+      "step": 4978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5876135696979774e-07,
+      "loss": 0.7999,
+      "step": 4979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.573963500242638e-07,
+      "loss": 0.7535,
+      "step": 4980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5603490597897575e-07,
+      "loss": 0.2399,
+      "step": 4981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.546770253318198e-07,
+      "loss": 0.2623,
+      "step": 4982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.533227085793766e-07,
+      "loss": 0.7643,
+      "step": 4983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.519719562169232e-07,
+      "loss": 0.7283,
+      "step": 4984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5062476873843336e-07,
+      "loss": 0.7543,
+      "step": 4985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4928114663657855e-07,
+      "loss": 0.7319,
+      "step": 4986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4794109040272464e-07,
+      "loss": 0.8646,
+      "step": 4987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4660460052693735e-07,
+      "loss": 0.8041,
+      "step": 4988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.452716774979713e-07,
+      "loss": 0.269,
+      "step": 4989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4394232180328216e-07,
+      "loss": 0.7563,
+      "step": 4990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4261653392901987e-07,
+      "loss": 0.795,
+      "step": 4991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.412943143600288e-07,
+      "loss": 0.7671,
+      "step": 4992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.399756635798489e-07,
+      "loss": 0.8073,
+      "step": 4993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.386605820707133e-07,
+      "loss": 0.8492,
+      "step": 4994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3734907031355392e-07,
+      "loss": 0.7665,
+      "step": 4995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.360411287879927e-07,
+      "loss": 0.7704,
+      "step": 4996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3473675797234918e-07,
+      "loss": 0.2538,
+      "step": 4997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3343595834363407e-07,
+      "loss": 0.7642,
+      "step": 4998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3213873037755685e-07,
+      "loss": 0.7419,
+      "step": 4999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.308450745485147e-07,
+      "loss": 0.7517,
+      "step": 5000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2955499132960245e-07,
+      "loss": 0.7855,
+      "step": 5001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2826848119260837e-07,
+      "loss": 0.7157,
+      "step": 5002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.269855446080127e-07,
+      "loss": 0.7972,
+      "step": 5003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2570618204498906e-07,
+      "loss": 0.8244,
+      "step": 5004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.244303939714032e-07,
+      "loss": 0.778,
+      "step": 5005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2315818085381856e-07,
+      "loss": 0.7351,
+      "step": 5006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.21889543157483e-07,
+      "loss": 0.7468,
+      "step": 5007
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2062448134634428e-07,
+      "loss": 0.7693,
+      "step": 5008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1936299588303678e-07,
+      "loss": 0.7515,
+      "step": 5009
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1810508722889144e-07,
+      "loss": 0.7965,
+      "step": 5010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1685075584393034e-07,
+      "loss": 0.7784,
+      "step": 5011
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.156000021868643e-07,
+      "loss": 0.7183,
+      "step": 5012
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.143528267150985e-07,
+      "loss": 0.794,
+      "step": 5013
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1310922988472926e-07,
+      "loss": 0.7442,
+      "step": 5014
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.118692121505417e-07,
+      "loss": 0.7724,
+      "step": 5015
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1063277396601744e-07,
+      "loss": 0.7804,
+      "step": 5016
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0939991578332375e-07,
+      "loss": 0.7159,
+      "step": 5017
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0817063805332104e-07,
+      "loss": 0.7506,
+      "step": 5018
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0694494122556085e-07,
+      "loss": 0.8552,
+      "step": 5019
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.057228257482813e-07,
+      "loss": 0.7491,
+      "step": 5020
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0450429206841704e-07,
+      "loss": 0.7377,
+      "step": 5021
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0328934063158946e-07,
+      "loss": 0.787,
+      "step": 5022
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0207797188210754e-07,
+      "loss": 0.7984,
+      "step": 5023
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0087018626297584e-07,
+      "loss": 0.8218,
+      "step": 5024
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9966598421588102e-07,
+      "loss": 0.8209,
+      "step": 5025
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9846536618120748e-07,
+      "loss": 0.2471,
+      "step": 5026
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9726833259802514e-07,
+      "loss": 0.7435,
+      "step": 5027
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9607488390408825e-07,
+      "loss": 0.793,
+      "step": 5028
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.948850205358488e-07,
+      "loss": 0.7776,
+      "step": 5029
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9369874292844203e-07,
+      "loss": 0.7792,
+      "step": 5030
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9251605151569542e-07,
+      "loss": 0.7217,
+      "step": 5031
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.913369467301196e-07,
+      "loss": 0.7343,
+      "step": 5032
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9016142900291856e-07,
+      "loss": 0.7642,
+      "step": 5033
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8898949876398288e-07,
+      "loss": 0.7065,
+      "step": 5034
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8782115644188969e-07,
+      "loss": 0.7962,
+      "step": 5035
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8665640246390838e-07,
+      "loss": 0.7985,
+      "step": 5036
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8549523725598929e-07,
+      "loss": 0.7396,
+      "step": 5037
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.843376612427772e-07,
+      "loss": 0.7352,
+      "step": 5038
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.831836748475979e-07,
+      "loss": 0.7945,
+      "step": 5039
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8203327849247165e-07,
+      "loss": 0.2332,
+      "step": 5040
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8088647259809967e-07,
+      "loss": 0.7589,
+      "step": 5041
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7974325758387424e-07,
+      "loss": 0.7757,
+      "step": 5042
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7860363386786982e-07,
+      "loss": 0.7607,
+      "step": 5043
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7746760186685198e-07,
+      "loss": 0.7733,
+      "step": 5044
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7633516199627165e-07,
+      "loss": 0.7845,
+      "step": 5045
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7520631467026428e-07,
+      "loss": 0.2388,
+      "step": 5046
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7408106030165407e-07,
+      "loss": 0.2848,
+      "step": 5047
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.729593993019485e-07,
+      "loss": 0.7449,
+      "step": 5048
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.718413320813428e-07,
+      "loss": 0.7655,
+      "step": 5049
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7072685904871878e-07,
+      "loss": 0.8363,
+      "step": 5050
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6961598061164374e-07,
+      "loss": 0.7764,
+      "step": 5051
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6850869717636608e-07,
+      "loss": 0.7483,
+      "step": 5052
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.674050091478241e-07,
+      "loss": 0.8454,
+      "step": 5053
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6630491692964046e-07,
+      "loss": 0.727,
+      "step": 5054
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6520842092412226e-07,
+      "loss": 0.7789,
+      "step": 5055
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6411552153226097e-07,
+      "loss": 0.7914,
+      "step": 5056
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.630262191537335e-07,
+      "loss": 0.8063,
+      "step": 5057
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6194051418690125e-07,
+      "loss": 0.2631,
+      "step": 5058
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6085840702880994e-07,
+      "loss": 0.7889,
+      "step": 5059
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5977989807518968e-07,
+      "loss": 0.7556,
+      "step": 5060
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5870498772045274e-07,
+      "loss": 0.7617,
+      "step": 5061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5763367635770022e-07,
+      "loss": 0.802,
+      "step": 5062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5656596437871097e-07,
+      "loss": 0.7705,
+      "step": 5063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5550185217395265e-07,
+      "loss": 0.7701,
+      "step": 5064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5444134013257396e-07,
+      "loss": 0.7496,
+      "step": 5065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5338442864240576e-07,
+      "loss": 0.7453,
+      "step": 5066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.523311180899667e-07,
+      "loss": 0.7702,
+      "step": 5067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5128140886045306e-07,
+      "loss": 0.8597,
+      "step": 5068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5023530133774888e-07,
+      "loss": 0.7398,
+      "step": 5069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4919279590441705e-07,
+      "loss": 0.7902,
+      "step": 5070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.48153892941707e-07,
+      "loss": 0.7698,
+      "step": 5071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.471185928295471e-07,
+      "loss": 0.7754,
+      "step": 5072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.460868959465489e-07,
+      "loss": 0.781,
+      "step": 5073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.450588026700106e-07,
+      "loss": 0.7649,
+      "step": 5074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.440343133759059e-07,
+      "loss": 0.8262,
+      "step": 5075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.430134284388962e-07,
+      "loss": 0.7684,
+      "step": 5076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4199614823231954e-07,
+      "loss": 0.737,
+      "step": 5077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4098247312820056e-07,
+      "loss": 0.7584,
+      "step": 5078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3997240349724273e-07,
+      "loss": 0.7184,
+      "step": 5079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3896593970883165e-07,
+      "loss": 0.7966,
+      "step": 5080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3796308213103405e-07,
+      "loss": 0.7795,
+      "step": 5081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3696383113059763e-07,
+      "loss": 0.7502,
+      "step": 5082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3596818707295233e-07,
+      "loss": 0.7252,
+      "step": 5083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3497615032220802e-07,
+      "loss": 0.7367,
+      "step": 5084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3398772124115667e-07,
+      "loss": 0.775,
+      "step": 5085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3300290019126695e-07,
+      "loss": 0.784,
+      "step": 5086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3202168753269406e-07,
+      "loss": 0.775,
+      "step": 5087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.310440836242688e-07,
+      "loss": 0.774,
+      "step": 5088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3007008882350401e-07,
+      "loss": 0.7265,
+      "step": 5089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2909970348659485e-07,
+      "loss": 0.7682,
+      "step": 5090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.281329279684096e-07,
+      "loss": 0.7688,
+      "step": 5091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2716976262250545e-07,
+      "loss": 0.776,
+      "step": 5092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2621020780111294e-07,
+      "loss": 0.7533,
+      "step": 5093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.252542638551446e-07,
+      "loss": 0.7599,
+      "step": 5094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2430193113419197e-07,
+      "loss": 0.7641,
+      "step": 5095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2335320998652643e-07,
+      "loss": 0.2559,
+      "step": 5096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2240810075909715e-07,
+      "loss": 0.7788,
+      "step": 5097
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2146660379753427e-07,
+      "loss": 0.7503,
+      "step": 5098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2052871944614797e-07,
+      "loss": 0.8238,
+      "step": 5099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1959444804792165e-07,
+      "loss": 0.7712,
+      "step": 5100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1866378994452421e-07,
+      "loss": 0.7457,
+      "step": 5101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1773674547629898e-07,
+      "loss": 0.838,
+      "step": 5102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1681331498226922e-07,
+      "loss": 0.7507,
+      "step": 5103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.158934988001359e-07,
+      "loss": 0.79,
+      "step": 5104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1497729726628104e-07,
+      "loss": 0.8173,
+      "step": 5105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1406471071575776e-07,
+      "loss": 0.8077,
+      "step": 5106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1315573948230574e-07,
+      "loss": 0.724,
+      "step": 5107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1225038389833797e-07,
+      "loss": 0.7468,
+      "step": 5108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1134864429494408e-07,
+      "loss": 0.73,
+      "step": 5109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1045052100189468e-07,
+      "loss": 0.7341,
+      "step": 5110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0955601434763375e-07,
+      "loss": 0.8075,
+      "step": 5111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0866512465928847e-07,
+      "loss": 0.7428,
+      "step": 5112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0777785226265824e-07,
+      "loss": 0.7968,
+      "step": 5113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.068941974822213e-07,
+      "loss": 0.788,
+      "step": 5114
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0601416064113357e-07,
+      "loss": 0.7417,
+      "step": 5115
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0513774206122651e-07,
+      "loss": 0.7671,
+      "step": 5116
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.042649420630093e-07,
+      "loss": 0.8644,
+      "step": 5117
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0339576096566772e-07,
+      "loss": 0.7595,
+      "step": 5118
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0253019908706419e-07,
+      "loss": 0.7151,
+      "step": 5119
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.016682567437377e-07,
+      "loss": 0.7635,
+      "step": 5120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0080993425090168e-07,
+      "loss": 0.796,
+      "step": 5121
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.995523192244838e-08,
+      "loss": 0.7924,
+      "step": 5122
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.910415007094553e-08,
+      "loss": 0.8019,
+      "step": 5123
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.82566890076353e-08,
+      "loss": 0.7833,
+      "step": 5124
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.741284904243752e-08,
+      "loss": 0.7795,
+      "step": 5125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.65726304839465e-08,
+      "loss": 0.7367,
+      "step": 5126
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.573603363943418e-08,
+      "loss": 0.8015,
+      "step": 5127
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.490305881484585e-08,
+      "loss": 0.7521,
+      "step": 5128
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.407370631480118e-08,
+      "loss": 0.6835,
+      "step": 5129
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.324797644259975e-08,
+      "loss": 0.8043,
+      "step": 5130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.242586950021226e-08,
+      "loss": 0.7919,
+      "step": 5131
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.160738578828487e-08,
+      "loss": 0.7867,
+      "step": 5132
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.07925256061415e-08,
+      "loss": 0.8432,
+      "step": 5133
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.998128925177707e-08,
+      "loss": 0.7524,
+      "step": 5134
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.917367702186541e-08,
+      "loss": 0.7827,
+      "step": 5135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.836968921175027e-08,
+      "loss": 0.7123,
+      "step": 5136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.756932611545422e-08,
+      "loss": 0.2749,
+      "step": 5137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.677258802567201e-08,
+      "loss": 0.801,
+      "step": 5138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.597947523377504e-08,
+      "loss": 0.7709,
+      "step": 5139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.51899880298035e-08,
+      "loss": 0.7347,
+      "step": 5140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.440412670247755e-08,
+      "loss": 0.7813,
+      "step": 5141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.362189153918843e-08,
+      "loss": 0.789,
+      "step": 5142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.284328282600285e-08,
+      "loss": 0.7456,
+      "step": 5143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.20683008476597e-08,
+      "loss": 0.7853,
+      "step": 5144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.12969458875723e-08,
+      "loss": 0.7683,
+      "step": 5145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.052921822782722e-08,
+      "loss": 0.8925,
+      "step": 5146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.976511814918542e-08,
+      "loss": 0.8144,
+      "step": 5147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.900464593108115e-08,
+      "loss": 0.7867,
+      "step": 5148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.824780185162084e-08,
+      "loss": 0.7427,
+      "step": 5149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.749458618758421e-08,
+      "loss": 0.7151,
+      "step": 5150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.674499921442536e-08,
+      "loss": 0.7361,
+      "step": 5151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.599904120626944e-08,
+      "loss": 0.7703,
+      "step": 5152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.525671243591714e-08,
+      "loss": 0.7724,
+      "step": 5153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.451801317483909e-08,
+      "loss": 0.7801,
+      "step": 5154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.37829436931814e-08,
+      "loss": 0.6727,
+      "step": 5155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.305150425975904e-08,
+      "loss": 0.7289,
+      "step": 5156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.232369514206361e-08,
+      "loss": 0.778,
+      "step": 5157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.159951660625664e-08,
+      "loss": 0.7523,
+      "step": 5158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.087896891717183e-08,
+      "loss": 0.7385,
+      "step": 5159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.016205233831619e-08,
+      "loss": 0.7291,
+      "step": 5160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.944876713186776e-08,
+      "loss": 0.7953,
+      "step": 5161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.873911355867791e-08,
+      "loss": 0.7819,
+      "step": 5162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.8033091878269e-08,
+      "loss": 0.7125,
+      "step": 5163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.733070234883566e-08,
+      "loss": 0.7438,
+      "step": 5164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.663194522724348e-08,
+      "loss": 0.2559,
+      "step": 5165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.593682076902919e-08,
+      "loss": 0.7622,
+      "step": 5166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.524532922840388e-08,
+      "loss": 0.7681,
+      "step": 5167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.455747085824638e-08,
+      "loss": 0.7628,
+      "step": 5168
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.387324591010991e-08,
+      "loss": 0.794,
+      "step": 5169
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.319265463421653e-08,
+      "loss": 0.7793,
+      "step": 5170
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.251569727946161e-08,
+      "loss": 0.832,
+      "step": 5171
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.184237409340932e-08,
+      "loss": 0.7896,
+      "step": 5172
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.117268532229826e-08,
+      "loss": 0.735,
+      "step": 5173
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.050663121103251e-08,
+      "loss": 0.7786,
+      "step": 5174
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.984421200319279e-08,
+      "loss": 0.7688,
+      "step": 5175
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9185427941026444e-08,
+      "loss": 0.7895,
+      "step": 5176
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.853027926545296e-08,
+      "loss": 0.7351,
+      "step": 5177
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.787876621606403e-08,
+      "loss": 0.7969,
+      "step": 5178
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7230889031116846e-08,
+      "loss": 0.771,
+      "step": 5179
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.658664794754409e-08,
+      "loss": 0.8214,
+      "step": 5180
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.594604320094621e-08,
+      "loss": 0.777,
+      "step": 5181
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.530907502559579e-08,
+      "loss": 0.7496,
+      "step": 5182
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4675743654430956e-08,
+      "loss": 0.7471,
+      "step": 5183
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.404604931906532e-08,
+      "loss": 0.791,
+      "step": 5184
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.341999224978023e-08,
+      "loss": 0.8087,
+      "step": 5185
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.279757267552588e-08,
+      "loss": 0.7487,
+      "step": 5186
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.217879082392241e-08,
+      "loss": 0.7836,
+      "step": 5187
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1563646921261036e-08,
+      "loss": 0.7831,
+      "step": 5188
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0952141192501804e-08,
+      "loss": 0.7202,
+      "step": 5189
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0344273861274716e-08,
+      "loss": 0.7947,
+      "step": 5190
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9740045149877515e-08,
+      "loss": 0.8449,
+      "step": 5191
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.91394552792801e-08,
+      "loss": 0.7785,
+      "step": 5192
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.85425044691179e-08,
+      "loss": 0.761,
+      "step": 5193
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.794919293769851e-08,
+      "loss": 0.7301,
+      "step": 5194
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.735952090199836e-08,
+      "loss": 0.7904,
+      "step": 5195
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.677348857766384e-08,
+      "loss": 0.8041,
+      "step": 5196
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6191096179004635e-08,
+      "loss": 0.2651,
+      "step": 5197
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.561234391900815e-08,
+      "loss": 0.8112,
+      "step": 5198
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.503723200932175e-08,
+      "loss": 0.6967,
+      "step": 5199
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.446576066026831e-08,
+      "loss": 0.7633,
+      "step": 5200
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.389793008083509e-08,
+      "loss": 0.7893,
+      "step": 5201
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.333374047868044e-08,
+      "loss": 0.7957,
+      "step": 5202
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.277319206013042e-08,
+      "loss": 0.7414,
+      "step": 5203
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2216285030178825e-08,
+      "loss": 0.7462,
+      "step": 5204
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.166301959248831e-08,
+      "loss": 0.7323,
+      "step": 5205
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1113395949390343e-08,
+      "loss": 0.7616,
+      "step": 5206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.056741430188305e-08,
+      "loss": 0.8455,
+      "step": 5207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.002507484963336e-08,
+      "loss": 0.7255,
+      "step": 5208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.948637779097708e-08,
+      "loss": 0.754,
+      "step": 5209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.895132332291885e-08,
+      "loss": 0.7942,
+      "step": 5210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8419911641126574e-08,
+      "loss": 0.7571,
+      "step": 5211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.789214293994259e-08,
+      "loss": 0.8217,
+      "step": 5212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7368017412371396e-08,
+      "loss": 0.7448,
+      "step": 5213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.684753525008744e-08,
+      "loss": 0.764,
+      "step": 5214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6330696643432913e-08,
+      "loss": 0.8164,
+      "step": 5215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.581750178141774e-08,
+      "loss": 0.7564,
+      "step": 5216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.530795085171734e-08,
+      "loss": 0.7521,
+      "step": 5217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.480204404067822e-08,
+      "loss": 0.7249,
+      "step": 5218
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.429978153331126e-08,
+      "loss": 0.8241,
+      "step": 5219
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38011635132951e-08,
+      "loss": 0.7226,
+      "step": 5220
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3306190162976096e-08,
+      "loss": 0.8152,
+      "step": 5221
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.281486166336611e-08,
+      "loss": 0.7985,
+      "step": 5222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2327178194148104e-08,
+      "loss": 0.778,
+      "step": 5223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1843139933668275e-08,
+      "loss": 0.7804,
+      "step": 5224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.13627470589406e-08,
+      "loss": 0.7401,
+      "step": 5225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.088599974564566e-08,
+      "loss": 0.7497,
+      "step": 5226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0412898168132867e-08,
+      "loss": 0.7443,
+      "step": 5227
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.994344249941605e-08,
+      "loss": 0.7606,
+      "step": 5228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.947763291117678e-08,
+      "loss": 0.251,
+      "step": 5229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9015469573763224e-08,
+      "loss": 0.7661,
+      "step": 5230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8556952656190184e-08,
+      "loss": 0.7761,
+      "step": 5231
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.810208232613798e-08,
+      "loss": 0.7587,
+      "step": 5232
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7650858749953547e-08,
+      "loss": 0.767,
+      "step": 5233
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7203282092652662e-08,
+      "loss": 0.7562,
+      "step": 5234
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6759352517914396e-08,
+      "loss": 0.7808,
+      "step": 5235
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.631907018808555e-08,
+      "loss": 0.8047,
+      "step": 5236
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.588243526417733e-08,
+      "loss": 0.727,
+      "step": 5237
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5449447905869783e-08,
+      "loss": 0.8273,
+      "step": 5238
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.502010827150847e-08,
+      "loss": 0.8002,
+      "step": 5239
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4594416518102236e-08,
+      "loss": 0.7648,
+      "step": 5240
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.417237280132878e-08,
+      "loss": 0.7268,
+      "step": 5241
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3753977275531303e-08,
+      "loss": 0.7824,
+      "step": 5242
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3339230093718523e-08,
+      "loss": 0.7948,
+      "step": 5243
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2928131407563558e-08,
+      "loss": 0.7941,
+      "step": 5244
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.252068136740726e-08,
+      "loss": 0.7284,
+      "step": 5245
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.211688012225488e-08,
+      "loss": 0.7287,
+      "step": 5246
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1716727819778293e-08,
+      "loss": 0.8418,
+      "step": 5247
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1320224606313777e-08,
+      "loss": 0.7995,
+      "step": 5248
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.092737062686534e-08,
+      "loss": 0.7308,
+      "step": 5249
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.053816602509917e-08,
+      "loss": 0.7664,
+      "step": 5250
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0152610943348084e-08,
+      "loss": 0.7893,
+      "step": 5251
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.977070552261373e-08,
+      "loss": 0.7974,
+      "step": 5252
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.939244990255773e-08,
+      "loss": 0.8125,
+      "step": 5253
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.901784422150943e-08,
+      "loss": 0.7386,
+      "step": 5254
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.864688861646369e-08,
+      "loss": 0.7463,
+      "step": 5255
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8279583223080875e-08,
+      "loss": 0.7898,
+      "step": 5256
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.791592817568355e-08,
+      "loss": 0.7688,
+      "step": 5257
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7555923607264215e-08,
+      "loss": 0.7699,
+      "step": 5258
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7199569649476445e-08,
+      "loss": 0.7273,
+      "step": 5259
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6846866432639332e-08,
+      "loss": 0.7909,
+      "step": 5260
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6497814085737472e-08,
+      "loss": 0.7794,
+      "step": 5261
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6152412736422095e-08,
+      "loss": 0.7625,
+      "step": 5262
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.581066251100549e-08,
+      "loss": 0.7557,
+      "step": 5263
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.54725635344688e-08,
+      "loss": 0.7168,
+      "step": 5264
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.513811593045422e-08,
+      "loss": 0.7495,
+      "step": 5265
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4807319821271705e-08,
+      "loss": 0.8026,
+      "step": 5266
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4480175327892255e-08,
+      "loss": 0.7911,
+      "step": 5267
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4156682569955727e-08,
+      "loss": 0.772,
+      "step": 5268
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3836841665763045e-08,
+      "loss": 0.7399,
+      "step": 5269
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3520652732281758e-08,
+      "loss": 0.8095,
+      "step": 5270
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3208115885143813e-08,
+      "loss": 0.7848,
+      "step": 5271
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.289923123864223e-08,
+      "loss": 0.8596,
+      "step": 5272
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.259399890574109e-08,
+      "loss": 0.8284,
+      "step": 5273
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2292418998061107e-08,
+      "loss": 0.765,
+      "step": 5274
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1994491625892946e-08,
+      "loss": 0.7797,
+      "step": 5275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1700216898189454e-08,
+      "loss": 0.7991,
+      "step": 5276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1409594922566768e-08,
+      "loss": 0.2636,
+      "step": 5277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1122625805307652e-08,
+      "loss": 0.7112,
+      "step": 5278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0839309651355933e-08,
+      "loss": 0.2566,
+      "step": 5279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0559646564323179e-08,
+      "loss": 0.7493,
+      "step": 5280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.028363664648091e-08,
+      "loss": 0.7542,
+      "step": 5281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0011279998769497e-08,
+      "loss": 0.7413,
+      "step": 5282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.742576720788155e-09,
+      "loss": 0.7959,
+      "step": 5283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.477526910802726e-09,
+      "loss": 0.8072,
+      "step": 5284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.21613066574456e-09,
+      "loss": 0.7965,
+      "step": 5285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.958388081204972e-09,
+      "loss": 0.74,
+      "step": 5286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.704299251443005e-09,
+      "loss": 0.7014,
+      "step": 5287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.453864269377665e-09,
+      "loss": 0.7797,
+      "step": 5288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.207083226595692e-09,
+      "loss": 0.7593,
+      "step": 5289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.963956213346002e-09,
+      "loss": 0.7724,
+      "step": 5290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.724483318538589e-09,
+      "loss": 0.7212,
+      "step": 5291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.488664629751174e-09,
+      "loss": 0.7795,
+      "step": 5292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.256500233222552e-09,
+      "loss": 0.7575,
+      "step": 5293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.0279902138570275e-09,
+      "loss": 0.8471,
+      "step": 5294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.803134655219978e-09,
+      "loss": 0.8386,
+      "step": 5295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.581933639543403e-09,
+      "loss": 0.7774,
+      "step": 5296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.364387247719261e-09,
+      "loss": 0.702,
+      "step": 5297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.150495559307246e-09,
+      "loss": 0.7666,
+      "step": 5298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.940258652525899e-09,
+      "loss": 0.7775,
+      "step": 5299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.733676604261495e-09,
+      "loss": 0.7671,
+      "step": 5300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.530749490059162e-09,
+      "loss": 0.8286,
+      "step": 5301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.331477384132866e-09,
+      "loss": 0.815,
+      "step": 5302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.135860359356537e-09,
+      "loss": 0.8747,
+      "step": 5303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.943898487266285e-09,
+      "loss": 0.7547,
+      "step": 5304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.755591838063734e-09,
+      "loss": 0.818,
+      "step": 5305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.570940480613795e-09,
+      "loss": 0.7677,
+      "step": 5306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.389944482443564e-09,
+      "loss": 0.7844,
+      "step": 5307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.212603909744539e-09,
+      "loss": 0.8183,
+      "step": 5308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.038918827370397e-09,
+      "loss": 0.7054,
+      "step": 5309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.868889298836997e-09,
+      "loss": 0.7878,
+      "step": 5310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7025153863268215e-09,
+      "loss": 0.74,
+      "step": 5311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5397971506823115e-09,
+      "loss": 0.733,
+      "step": 5312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.380734651409201e-09,
+      "loss": 0.8041,
+      "step": 5313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2253279466787357e-09,
+      "loss": 0.7699,
+      "step": 5314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.073577093322122e-09,
+      "loss": 0.8171,
+      "step": 5315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9254821468371885e-09,
+      "loss": 0.7701,
+      "step": 5316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7810431613806142e-09,
+      "loss": 0.7661,
+      "step": 5317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6402601897745903e-09,
+      "loss": 0.7447,
+      "step": 5318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5031332835034895e-09,
+      "loss": 0.7773,
+      "step": 5319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3696624927171952e-09,
+      "loss": 0.7097,
+      "step": 5320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2398478662244426e-09,
+      "loss": 0.2443,
+      "step": 5321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.113689451499479e-09,
+      "loss": 0.7269,
+      "step": 5322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9911872946787314e-09,
+      "loss": 0.7694,
+      "step": 5323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.87234144056081e-09,
+      "loss": 0.8382,
+      "step": 5324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7571519326098353e-09,
+      "loss": 0.7513,
+      "step": 5325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6456188129498897e-09,
+      "loss": 0.7853,
+      "step": 5326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5377421223683463e-09,
+      "loss": 0.7695,
+      "step": 5327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.433521900316981e-09,
+      "loss": 0.8235,
+      "step": 5328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3329581849097494e-09,
+      "loss": 0.2455,
+      "step": 5329
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2360510129216797e-09,
+      "loss": 0.7777,
+      "step": 5330
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1428004197933108e-09,
+      "loss": 0.7834,
+      "step": 5331
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0532064396262531e-09,
+      "loss": 0.763,
+      "step": 5332
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.672691051854088e-10,
+      "loss": 0.8242,
+      "step": 5333
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.84988447897861e-10,
+      "loss": 0.8207,
+      "step": 5334
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.063644978550944e-10,
+      "loss": 0.7548,
+      "step": 5335
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.313972838085548e-10,
+      "loss": 0.7573,
+      "step": 5336
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.600868331751998e-10,
+      "loss": 0.2594,
+      "step": 5337
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.924331720319476e-10,
+      "loss": 0.7427,
+      "step": 5338
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.28436325122339e-10,
+      "loss": 0.7863,
+      "step": 5339
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.680963158487651e-10,
+      "loss": 0.793,
+      "step": 5340
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.114131662769083e-10,
+      "loss": 0.2501,
+      "step": 5341
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.583868971379634e-10,
+      "loss": 0.7625,
+      "step": 5342
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.090175278219754e-10,
+      "loss": 0.7266,
+      "step": 5343
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6330507638450133e-10,
+      "loss": 0.7714,
+      "step": 5344
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2124955954327953e-10,
+      "loss": 0.7081,
+      "step": 5345
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8285099267600914e-10,
+      "loss": 0.7639,
+      "step": 5346
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4810938982812163e-10,
+      "loss": 0.7901,
+      "step": 5347
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.170247637016786e-10,
+      "loss": 0.7989,
+      "step": 5348
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.959712566647404e-11,
+      "loss": 0.259,
+      "step": 5349
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.58264857522628e-11,
+      "loss": 0.7102,
+      "step": 5350
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.571285265209113e-11,
+      "loss": 0.7701,
+      "step": 5351
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.92562337222968e-11,
+      "loss": 0.7522,
+      "step": 5352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6456634979178378e-11,
+      "loss": 0.7545,
+      "step": 5353
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.314061105656578e-12,
+      "loss": 0.8023,
+      "step": 5354
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8285154435027097e-12,
+      "loss": 0.3655,
+      "step": 5355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.6735,
+      "step": 5356
+    },
+    {
+      "epoch": 1.0,
+      "step": 5356,
+      "total_flos": 2200339809828864.0,
+      "train_loss": 0.7849369076059058,
+      "train_runtime": 66915.2023,
+      "train_samples_per_second": 10.244,
+      "train_steps_per_second": 0.08
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 5356,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 2200339809828864.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}