diff --git a/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/0_trainer_state.json b/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/0_trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..a3fbd3f74d662639e3ef28fd318ed170e420dd68
--- /dev/null
+++ b/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/0_trainer_state.json
@@ -0,0 +1,30031 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0002,
+      "learning_rate": 1.8044772347286555e-06,
+      "loss": 0.5368,
+      "step": 2
+    },
+    {
+      "epoch": 0.0004,
+      "learning_rate": 1.8084795571100865e-06,
+      "loss": 0.056,
+      "step": 4
+    },
+    {
+      "epoch": 0.0006,
+      "learning_rate": 1.8124858719390482e-06,
+      "loss": 0.7622,
+      "step": 6
+    },
+    {
+      "epoch": 0.0008,
+      "learning_rate": 1.8164961772629441e-06,
+      "loss": 1.0897,
+      "step": 8
+    },
+    {
+      "epoch": 0.001,
+      "learning_rate": 1.820510471127198e-06,
+      "loss": 1.0255,
+      "step": 10
+    },
+    {
+      "epoch": 0.0012,
+      "learning_rate": 1.824528751575263e-06,
+      "loss": 0.5419,
+      "step": 12
+    },
+    {
+      "epoch": 0.0014,
+      "learning_rate": 1.8285510166487075e-06,
+      "loss": 0.9364,
+      "step": 14
+    },
+    {
+      "epoch": 0.0016,
+      "learning_rate": 1.8325772643871287e-06,
+      "loss": 1.2027,
+      "step": 16
+    },
+    {
+      "epoch": 0.0018,
+      "learning_rate": 1.836607492828163e-06,
+      "loss": 0.5672,
+      "step": 18
+    },
+    {
+      "epoch": 0.002,
+      "learning_rate": 1.8406417000075238e-06,
+      "loss": 0.5454,
+      "step": 20
+    },
+    {
+      "epoch": 0.0022,
+      "learning_rate": 1.8446798839590207e-06,
+      "loss": 1.0434,
+      "step": 22
+    },
+    {
+      "epoch": 0.0024,
+      "learning_rate": 1.8487220427144593e-06,
+      "loss": 0.6976,
+      "step": 24
+    },
+    {
+      "epoch": 0.0026,
+      "learning_rate": 1.8527681743037418e-06,
+      "loss": 0.1259,
+      "step": 26
+    },
+    {
+      "epoch": 0.0028,
+      "learning_rate": 1.8568182767548626e-06,
+      "loss": 0.7735,
+      "step": 28
+    },
+    {
+      "epoch": 0.003,
+      "learning_rate": 1.8608723480938196e-06,
+      "loss": 0.6747,
+      "step": 30
+    },
+    {
+      "epoch": 0.0032,
+      "learning_rate": 1.86493038634472e-06,
+      "loss": 0.1756,
+      "step": 32
+    },
+    {
+      "epoch": 0.0034,
+      "learning_rate": 1.8689923895297135e-06,
+      "loss": 0.5001,
+      "step": 34
+    },
+    {
+      "epoch": 0.0036,
+      "learning_rate": 1.8730583556690607e-06,
+      "loss": 0.0993,
+      "step": 36
+    },
+    {
+      "epoch": 0.0038,
+      "learning_rate": 1.8771282827810278e-06,
+      "loss": 0.6742,
+      "step": 38
+    },
+    {
+      "epoch": 0.004,
+      "learning_rate": 1.8812021688819914e-06,
+      "loss": 0.8416,
+      "step": 40
+    },
+    {
+      "epoch": 0.0042,
+      "learning_rate": 1.8852800119863902e-06,
+      "loss": 0.4637,
+      "step": 42
+    },
+    {
+      "epoch": 0.0044,
+      "learning_rate": 1.8893618101067323e-06,
+      "loss": 0.6535,
+      "step": 44
+    },
+    {
+      "epoch": 0.0046,
+      "learning_rate": 1.8934475612535995e-06,
+      "loss": 0.466,
+      "step": 46
+    },
+    {
+      "epoch": 0.0048,
+      "learning_rate": 1.8975372634356447e-06,
+      "loss": 0.6003,
+      "step": 48
+    },
+    {
+      "epoch": 0.005,
+      "learning_rate": 1.901630914659599e-06,
+      "loss": 0.474,
+      "step": 50
+    },
+    {
+      "epoch": 0.0052,
+      "learning_rate": 1.9057285129302648e-06,
+      "loss": 0.6772,
+      "step": 52
+    },
+    {
+      "epoch": 0.0054,
+      "learning_rate": 1.9098300562505232e-06,
+      "loss": 0.5869,
+      "step": 54
+    },
+    {
+      "epoch": 0.0056,
+      "learning_rate": 1.9139355426213325e-06,
+      "loss": 0.0233,
+      "step": 56
+    },
+    {
+      "epoch": 0.0058,
+      "learning_rate": 1.918044970041727e-06,
+      "loss": 0.7096,
+      "step": 58
+    },
+    {
+      "epoch": 0.006,
+      "learning_rate": 1.9221583365088225e-06,
+      "loss": 0.936,
+      "step": 60
+    },
+    {
+      "epoch": 0.0062,
+      "learning_rate": 1.9262756400178138e-06,
+      "loss": 0.6368,
+      "step": 62
+    },
+    {
+      "epoch": 0.0064,
+      "learning_rate": 1.9303968785619776e-06,
+      "loss": 0.0473,
+      "step": 64
+    },
+    {
+      "epoch": 0.0066,
+      "learning_rate": 1.934522050132672e-06,
+      "loss": 0.1258,
+      "step": 66
+    },
+    {
+      "epoch": 0.0068,
+      "learning_rate": 1.9386511527193385e-06,
+      "loss": 0.3714,
+      "step": 68
+    },
+    {
+      "epoch": 0.007,
+      "learning_rate": 1.942784184309503e-06,
+      "loss": 0.7465,
+      "step": 70
+    },
+    {
+      "epoch": 0.0072,
+      "learning_rate": 1.9469211428887758e-06,
+      "loss": 0.5523,
+      "step": 72
+    },
+    {
+      "epoch": 0.0074,
+      "learning_rate": 1.9510620264408543e-06,
+      "loss": 0.5647,
+      "step": 74
+    },
+    {
+      "epoch": 0.0076,
+      "learning_rate": 1.9552068329475314e-06,
+      "loss": 0.027,
+      "step": 76
+    },
+    {
+      "epoch": 0.0078,
+      "learning_rate": 1.9593555603886483e-06,
+      "loss": 0.1518,
+      "step": 78
+    },
+    {
+      "epoch": 0.008,
+      "learning_rate": 1.9635082067421964e-06,
+      "loss": 0.6175,
+      "step": 80
+    },
+    {
+      "epoch": 0.0082,
+      "learning_rate": 1.9676647699842267e-06,
+      "loss": 0.5798,
+      "step": 82
+    },
+    {
+      "epoch": 0.0084,
+      "learning_rate": 1.9718252480888487e-06,
+      "loss": 0.8177,
+      "step": 84
+    },
+    {
+      "epoch": 0.0086,
+      "learning_rate": 1.9759896390283286e-06,
+      "loss": 0.46,
+      "step": 86
+    },
+    {
+      "epoch": 0.0088,
+      "learning_rate": 1.980157940772992e-06,
+      "loss": 0.9305,
+      "step": 88
+    },
+    {
+      "epoch": 0.009,
+      "learning_rate": 1.9843301512912383e-06,
+      "loss": 0.4719,
+      "step": 90
+    },
+    {
+      "epoch": 0.0092,
+      "learning_rate": 1.9885062685495825e-06,
+      "loss": 0.5666,
+      "step": 92
+    },
+    {
+      "epoch": 0.0094,
+      "learning_rate": 1.9926862905126574e-06,
+      "loss": 0.3787,
+      "step": 94
+    },
+    {
+      "epoch": 0.0096,
+      "learning_rate": 1.996870215143172e-06,
+      "loss": 0.0162,
+      "step": 96
+    },
+    {
+      "epoch": 0.0098,
+      "learning_rate": 2.0010580404018985e-06,
+      "loss": 1.3808,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.005249764247774e-06,
+      "loss": 0.7162,
+      "step": 100
+    },
+    {
+      "epoch": 0.0102,
+      "learning_rate": 2.009445384637805e-06,
+      "loss": 0.1886,
+      "step": 102
+    },
+    {
+      "epoch": 0.0104,
+      "learning_rate": 2.013644899527073e-06,
+      "loss": 0.7329,
+      "step": 104
+    },
+    {
+      "epoch": 0.0106,
+      "learning_rate": 2.0178483068687882e-06,
+      "loss": 0.4047,
+      "step": 106
+    },
+    {
+      "epoch": 0.0108,
+      "learning_rate": 2.0220556046142916e-06,
+      "loss": 0.6338,
+      "step": 108
+    },
+    {
+      "epoch": 0.011,
+      "learning_rate": 2.026266790712965e-06,
+      "loss": 0.6028,
+      "step": 110
+    },
+    {
+      "epoch": 0.0112,
+      "learning_rate": 2.0304818631123393e-06,
+      "loss": 0.2665,
+      "step": 112
+    },
+    {
+      "epoch": 0.0114,
+      "learning_rate": 2.0347008197580376e-06,
+      "loss": 0.176,
+      "step": 114
+    },
+    {
+      "epoch": 0.0116,
+      "learning_rate": 2.038923658593793e-06,
+      "loss": 0.8253,
+      "step": 116
+    },
+    {
+      "epoch": 0.0118,
+      "learning_rate": 2.0431503775614457e-06,
+      "loss": 0.793,
+      "step": 118
+    },
+    {
+      "epoch": 0.012,
+      "learning_rate": 2.0473809746009333e-06,
+      "loss": 0.1299,
+      "step": 120
+    },
+    {
+      "epoch": 0.0122,
+      "learning_rate": 2.051615447650345e-06,
+      "loss": 0.3494,
+      "step": 122
+    },
+    {
+      "epoch": 0.0124,
+      "learning_rate": 2.0558537946458177e-06,
+      "loss": 0.4588,
+      "step": 124
+    },
+    {
+      "epoch": 0.0126,
+      "learning_rate": 2.060096013521645e-06,
+      "loss": 0.7277,
+      "step": 126
+    },
+    {
+      "epoch": 0.0128,
+      "learning_rate": 2.0643421022102216e-06,
+      "loss": 0.5025,
+      "step": 128
+    },
+    {
+      "epoch": 0.013,
+      "learning_rate": 2.068592058642053e-06,
+      "loss": 0.414,
+      "step": 130
+    },
+    {
+      "epoch": 0.0132,
+      "learning_rate": 2.072845880745763e-06,
+      "loss": 0.4479,
+      "step": 132
+    },
+    {
+      "epoch": 0.0134,
+      "learning_rate": 2.077103566448091e-06,
+      "loss": 1.309,
+      "step": 134
+    },
+    {
+      "epoch": 0.0136,
+      "learning_rate": 2.0813651136738923e-06,
+      "loss": 0.4382,
+      "step": 136
+    },
+    {
+      "epoch": 0.0138,
+      "learning_rate": 2.08563052034614e-06,
+      "loss": 0.1722,
+      "step": 138
+    },
+    {
+      "epoch": 0.014,
+      "learning_rate": 2.0898997843859283e-06,
+      "loss": 0.6185,
+      "step": 140
+    },
+    {
+      "epoch": 0.0142,
+      "learning_rate": 2.094172903712468e-06,
+      "loss": 0.5179,
+      "step": 142
+    },
+    {
+      "epoch": 0.0144,
+      "learning_rate": 2.0984498762430927e-06,
+      "loss": 0.0047,
+      "step": 144
+    },
+    {
+      "epoch": 0.0146,
+      "learning_rate": 2.1027306998932596e-06,
+      "loss": 1.0577,
+      "step": 146
+    },
+    {
+      "epoch": 0.0148,
+      "learning_rate": 2.1070153725765573e-06,
+      "loss": 0.6241,
+      "step": 148
+    },
+    {
+      "epoch": 0.015,
+      "learning_rate": 2.1113038922046547e-06,
+      "loss": 0.8945,
+      "step": 150
+    },
+    {
+      "epoch": 0.0152,
+      "learning_rate": 2.1155962566874135e-06,
+      "loss": 0.6835,
+      "step": 152
+    },
+    {
+      "epoch": 0.0154,
+      "learning_rate": 2.1198924639327757e-06,
+      "loss": 0.8412,
+      "step": 154
+    },
+    {
+      "epoch": 0.0156,
+      "learning_rate": 2.124192511846822e-06,
+      "loss": 0.0192,
+      "step": 156
+    },
+    {
+      "epoch": 0.0158,
+      "learning_rate": 2.1284963983337626e-06,
+      "loss": 0.6997,
+      "step": 158
+    },
+    {
+      "epoch": 0.016,
+      "learning_rate": 2.132804121295934e-06,
+      "loss": 0.9491,
+      "step": 160
+    },
+    {
+      "epoch": 0.0162,
+      "learning_rate": 2.1371156786338167e-06,
+      "loss": 0.9945,
+      "step": 162
+    },
+    {
+      "epoch": 0.0164,
+      "learning_rate": 2.141431068245975e-06,
+      "loss": 1.2288,
+      "step": 164
+    },
+    {
+      "epoch": 0.0166,
+      "learning_rate": 2.145750288029176e-06,
+      "loss": 0.2231,
+      "step": 166
+    },
+    {
+      "epoch": 0.0168,
+      "learning_rate": 2.150073335878282e-06,
+      "loss": 0.9383,
+      "step": 168
+    },
+    {
+      "epoch": 0.017,
+      "learning_rate": 2.1544002096862605e-06,
+      "loss": 0.0852,
+      "step": 170
+    },
+    {
+      "epoch": 0.0172,
+      "learning_rate": 2.1587309073442776e-06,
+      "loss": 0.631,
+      "step": 172
+    },
+    {
+      "epoch": 0.0174,
+      "learning_rate": 2.1630654267416064e-06,
+      "loss": 0.7887,
+      "step": 174
+    },
+    {
+      "epoch": 0.0176,
+      "learning_rate": 2.1674037657656286e-06,
+      "loss": 0.6414,
+      "step": 176
+    },
+    {
+      "epoch": 0.0178,
+      "learning_rate": 2.1717459223018955e-06,
+      "loss": 0.8751,
+      "step": 178
+    },
+    {
+      "epoch": 0.018,
+      "learning_rate": 2.1760918942341104e-06,
+      "loss": 0.7902,
+      "step": 180
+    },
+    {
+      "epoch": 0.0182,
+      "learning_rate": 2.180441679444102e-06,
+      "loss": 1.209,
+      "step": 182
+    },
+    {
+      "epoch": 0.0184,
+      "learning_rate": 2.184795275811815e-06,
+      "loss": 0.0324,
+      "step": 184
+    },
+    {
+      "epoch": 0.0186,
+      "learning_rate": 2.1891526812153585e-06,
+      "loss": 0.4772,
+      "step": 186
+    },
+    {
+      "epoch": 0.0188,
+      "learning_rate": 2.1935138935310208e-06,
+      "loss": 0.7601,
+      "step": 188
+    },
+    {
+      "epoch": 0.019,
+      "learning_rate": 2.1978789106331657e-06,
+      "loss": 1.0063,
+      "step": 190
+    },
+    {
+      "epoch": 0.0192,
+      "learning_rate": 2.202247730394338e-06,
+      "loss": 0.7586,
+      "step": 192
+    },
+    {
+      "epoch": 0.0194,
+      "learning_rate": 2.2066203506852603e-06,
+      "loss": 0.5659,
+      "step": 194
+    },
+    {
+      "epoch": 0.0196,
+      "learning_rate": 2.210996769374737e-06,
+      "loss": 0.0713,
+      "step": 196
+    },
+    {
+      "epoch": 0.0198,
+      "learning_rate": 2.2153769843297655e-06,
+      "loss": 1.2086,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.219760993415485e-06,
+      "loss": 0.9433,
+      "step": 200
+    },
+    {
+      "epoch": 0.0202,
+      "learning_rate": 2.22414879449518e-06,
+      "loss": 0.8883,
+      "step": 202
+    },
+    {
+      "epoch": 0.0204,
+      "learning_rate": 2.2285403854302912e-06,
+      "loss": 0.8672,
+      "step": 204
+    },
+    {
+      "epoch": 0.0206,
+      "learning_rate": 2.2329357640804084e-06,
+      "loss": 0.628,
+      "step": 206
+    },
+    {
+      "epoch": 0.0208,
+      "learning_rate": 2.2373349283032795e-06,
+      "loss": 0.4245,
+      "step": 208
+    },
+    {
+      "epoch": 0.021,
+      "learning_rate": 2.241737875954806e-06,
+      "loss": 0.8283,
+      "step": 210
+    },
+    {
+      "epoch": 0.0212,
+      "learning_rate": 2.246144604889041e-06,
+      "loss": 0.5997,
+      "step": 212
+    },
+    {
+      "epoch": 0.0214,
+      "learning_rate": 2.2505551129582025e-06,
+      "loss": 1.1574,
+      "step": 214
+    },
+    {
+      "epoch": 0.0216,
+      "learning_rate": 2.2549693980126595e-06,
+      "loss": 1.1181,
+      "step": 216
+    },
+    {
+      "epoch": 0.0218,
+      "learning_rate": 2.2593874579009445e-06,
+      "loss": 0.5658,
+      "step": 218
+    },
+    {
+      "epoch": 0.022,
+      "learning_rate": 2.2638092904697495e-06,
+      "loss": 0.7431,
+      "step": 220
+    },
+    {
+      "epoch": 0.0222,
+      "learning_rate": 2.268234893563924e-06,
+      "loss": 1.0879,
+      "step": 222
+    },
+    {
+      "epoch": 0.0224,
+      "learning_rate": 2.2726642650264864e-06,
+      "loss": 0.9625,
+      "step": 224
+    },
+    {
+      "epoch": 0.0226,
+      "learning_rate": 2.2770974026986127e-06,
+      "loss": 0.9548,
+      "step": 226
+    },
+    {
+      "epoch": 0.0228,
+      "learning_rate": 2.281534304419647e-06,
+      "loss": 0.5743,
+      "step": 228
+    },
+    {
+      "epoch": 0.023,
+      "learning_rate": 2.285974968027095e-06,
+      "loss": 0.6904,
+      "step": 230
+    },
+    {
+      "epoch": 0.0232,
+      "learning_rate": 2.290419391356633e-06,
+      "loss": 0.0037,
+      "step": 232
+    },
+    {
+      "epoch": 0.0234,
+      "learning_rate": 2.294867572242114e-06,
+      "loss": 0.0655,
+      "step": 234
+    },
+    {
+      "epoch": 0.0236,
+      "learning_rate": 2.299319508515515e-06,
+      "loss": 0.3359,
+      "step": 236
+    },
+    {
+      "epoch": 0.0238,
+      "learning_rate": 2.30377519800705e-06,
+      "loss": 0.8998,
+      "step": 238
+    },
+    {
+      "epoch": 0.024,
+      "learning_rate": 2.3082346385450583e-06,
+      "loss": 0.8506,
+      "step": 240
+    },
+    {
+      "epoch": 0.0242,
+      "learning_rate": 2.312697827956063e-06,
+      "loss": 1.0988,
+      "step": 242
+    },
+    {
+      "epoch": 0.0244,
+      "learning_rate": 2.31716476406476e-06,
+      "loss": 0.5812,
+      "step": 244
+    },
+    {
+      "epoch": 0.0246,
+      "learning_rate": 2.321635444694019e-06,
+      "loss": 0.8598,
+      "step": 246
+    },
+    {
+      "epoch": 0.0248,
+      "learning_rate": 2.3261098676648963e-06,
+      "loss": 0.2181,
+      "step": 248
+    },
+    {
+      "epoch": 0.025,
+      "learning_rate": 2.3305880307965766e-06,
+      "loss": 0.4306,
+      "step": 250
+    },
+    {
+      "epoch": 0.0252,
+      "learning_rate": 2.335069931906494e-06,
+      "loss": 0.6693,
+      "step": 252
+    },
+    {
+      "epoch": 0.0254,
+      "learning_rate": 2.339555568810223e-06,
+      "loss": 1.3518,
+      "step": 254
+    },
+    {
+      "epoch": 0.0256,
+      "learning_rate": 2.344044939321486e-06,
+      "loss": 0.3753,
+      "step": 256
+    },
+    {
+      "epoch": 0.0258,
+      "learning_rate": 2.3485380412522497e-06,
+      "loss": 0.7767,
+      "step": 258
+    },
+    {
+      "epoch": 0.026,
+      "learning_rate": 2.353034872412634e-06,
+      "loss": 1.1884,
+      "step": 260
+    },
+    {
+      "epoch": 0.0262,
+      "learning_rate": 2.357535430610912e-06,
+      "loss": 0.6935,
+      "step": 262
+    },
+    {
+      "epoch": 0.0264,
+      "learning_rate": 2.362039713653569e-06,
+      "loss": 0.1074,
+      "step": 264
+    },
+    {
+      "epoch": 0.0266,
+      "learning_rate": 2.366547719345306e-06,
+      "loss": 0.8132,
+      "step": 266
+    },
+    {
+      "epoch": 0.0268,
+      "learning_rate": 2.3710594454889402e-06,
+      "loss": 0.1838,
+      "step": 268
+    },
+    {
+      "epoch": 0.027,
+      "learning_rate": 2.3755748898855234e-06,
+      "loss": 0.5063,
+      "step": 270
+    },
+    {
+      "epoch": 0.0272,
+      "learning_rate": 2.3800940503342706e-06,
+      "loss": 0.66,
+      "step": 272
+    },
+    {
+      "epoch": 0.0274,
+      "learning_rate": 2.3846169246326332e-06,
+      "loss": 0.3386,
+      "step": 274
+    },
+    {
+      "epoch": 0.0276,
+      "learning_rate": 2.3891435105761838e-06,
+      "loss": 0.2036,
+      "step": 276
+    },
+    {
+      "epoch": 0.0278,
+      "learning_rate": 2.3936738059587174e-06,
+      "loss": 0.4163,
+      "step": 278
+    },
+    {
+      "epoch": 0.028,
+      "learning_rate": 2.3982078085722573e-06,
+      "loss": 0.849,
+      "step": 280
+    },
+    {
+      "epoch": 0.0282,
+      "learning_rate": 2.4027455162069537e-06,
+      "loss": 0.3537,
+      "step": 282
+    },
+    {
+      "epoch": 0.0284,
+      "learning_rate": 2.407286926651192e-06,
+      "loss": 0.805,
+      "step": 284
+    },
+    {
+      "epoch": 0.0286,
+      "learning_rate": 2.411832037691545e-06,
+      "loss": 0.5155,
+      "step": 286
+    },
+    {
+      "epoch": 0.0288,
+      "learning_rate": 2.4163808471127803e-06,
+      "loss": 0.8773,
+      "step": 288
+    },
+    {
+      "epoch": 0.029,
+      "learning_rate": 2.420933352697865e-06,
+      "loss": 0.0888,
+      "step": 290
+    },
+    {
+      "epoch": 0.0292,
+      "learning_rate": 2.425489552227962e-06,
+      "loss": 0.4811,
+      "step": 292
+    },
+    {
+      "epoch": 0.0294,
+      "learning_rate": 2.430049443482434e-06,
+      "loss": 1.4083,
+      "step": 294
+    },
+    {
+      "epoch": 0.0296,
+      "learning_rate": 2.434613024238848e-06,
+      "loss": 0.4756,
+      "step": 296
+    },
+    {
+      "epoch": 0.0298,
+      "learning_rate": 2.439180292272967e-06,
+      "loss": 0.6196,
+      "step": 298
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.443751245358762e-06,
+      "loss": 1.1845,
+      "step": 300
+    },
+    {
+      "epoch": 0.0302,
+      "learning_rate": 2.448325881268406e-06,
+      "loss": 0.5865,
+      "step": 302
+    },
+    {
+      "epoch": 0.0304,
+      "learning_rate": 2.452904197772277e-06,
+      "loss": 0.5422,
+      "step": 304
+    },
+    {
+      "epoch": 0.0306,
+      "learning_rate": 2.457486192638958e-06,
+      "loss": 0.8283,
+      "step": 306
+    },
+    {
+      "epoch": 0.0308,
+      "learning_rate": 2.4620718636352427e-06,
+      "loss": 0.9512,
+      "step": 308
+    },
+    {
+      "epoch": 0.031,
+      "learning_rate": 2.4666612085261277e-06,
+      "loss": 0.3493,
+      "step": 310
+    },
+    {
+      "epoch": 0.0312,
+      "learning_rate": 2.471254225074825e-06,
+      "loss": 0.7551,
+      "step": 312
+    },
+    {
+      "epoch": 0.0314,
+      "learning_rate": 2.475850911042752e-06,
+      "loss": 0.8462,
+      "step": 314
+    },
+    {
+      "epoch": 0.0316,
+      "learning_rate": 2.4804512641895406e-06,
+      "loss": 0.1541,
+      "step": 316
+    },
+    {
+      "epoch": 0.0318,
+      "learning_rate": 2.4850552822730346e-06,
+      "loss": 0.4945,
+      "step": 318
+    },
+    {
+      "epoch": 0.032,
+      "learning_rate": 2.489662963049303e-06,
+      "loss": 1.1292,
+      "step": 320
+    },
+    {
+      "epoch": 0.0322,
+      "learning_rate": 2.4942743042725836e-06,
+      "loss": 1.0973,
+      "step": 322
+    },
+    {
+      "epoch": 0.0324,
+      "learning_rate": 2.4988893036953986e-06,
+      "loss": 0.5071,
+      "step": 324
+    },
+    {
+      "epoch": 0.0326,
+      "learning_rate": 2.503507959068455e-06,
+      "loss": 1.0047,
+      "step": 326
+    },
+    {
+      "epoch": 0.0328,
+      "learning_rate": 2.5081302681406395e-06,
+      "loss": 0.4808,
+      "step": 328
+    },
+    {
+      "epoch": 0.033,
+      "learning_rate": 2.5127562286591313e-06,
+      "loss": 0.2838,
+      "step": 330
+    },
+    {
+      "epoch": 0.0332,
+      "learning_rate": 2.5173858383692817e-06,
+      "loss": 0.5256,
+      "step": 332
+    },
+    {
+      "epoch": 0.0334,
+      "learning_rate": 2.522019095014686e-06,
+      "loss": 0.5425,
+      "step": 334
+    },
+    {
+      "epoch": 0.0336,
+      "learning_rate": 2.526655996337116e-06,
+      "loss": 0.6775,
+      "step": 336
+    },
+    {
+      "epoch": 0.0338,
+      "learning_rate": 2.531296540076638e-06,
+      "loss": 1.1435,
+      "step": 338
+    },
+    {
+      "epoch": 0.034,
+      "learning_rate": 2.535940723971507e-06,
+      "loss": 0.4584,
+      "step": 340
+    },
+    {
+      "epoch": 0.0342,
+      "learning_rate": 2.5405885457581814e-06,
+      "loss": 1.0321,
+      "step": 342
+    },
+    {
+      "epoch": 0.0344,
+      "learning_rate": 2.5452400031713698e-06,
+      "loss": 0.4395,
+      "step": 344
+    },
+    {
+      "epoch": 0.0346,
+      "learning_rate": 2.5498950939440413e-06,
+      "loss": 0.4751,
+      "step": 346
+    },
+    {
+      "epoch": 0.0348,
+      "learning_rate": 2.5545538158073278e-06,
+      "loss": 1.1209,
+      "step": 348
+    },
+    {
+      "epoch": 0.035,
+      "learning_rate": 2.5592161664906243e-06,
+      "loss": 0.8873,
+      "step": 350
+    },
+    {
+      "epoch": 0.0352,
+      "learning_rate": 2.563882143721593e-06,
+      "loss": 0.2287,
+      "step": 352
+    },
+    {
+      "epoch": 0.0354,
+      "learning_rate": 2.5685517452260587e-06,
+      "loss": 0.4561,
+      "step": 354
+    },
+    {
+      "epoch": 0.0356,
+      "learning_rate": 2.5732249687281253e-06,
+      "loss": 0.0294,
+      "step": 356
+    },
+    {
+      "epoch": 0.0358,
+      "learning_rate": 2.5779018119501086e-06,
+      "loss": 0.8749,
+      "step": 358
+    },
+    {
+      "epoch": 0.036,
+      "learning_rate": 2.5825822726126082e-06,
+      "loss": 0.6143,
+      "step": 360
+    },
+    {
+      "epoch": 0.0362,
+      "learning_rate": 2.5872663484343887e-06,
+      "loss": 0.105,
+      "step": 362
+    },
+    {
+      "epoch": 0.0364,
+      "learning_rate": 2.591954037132488e-06,
+      "loss": 0.9742,
+      "step": 364
+    },
+    {
+      "epoch": 0.0366,
+      "learning_rate": 2.596645336422219e-06,
+      "loss": 0.4256,
+      "step": 366
+    },
+    {
+      "epoch": 0.0368,
+      "learning_rate": 2.601340244017064e-06,
+      "loss": 0.5821,
+      "step": 368
+    },
+    {
+      "epoch": 0.037,
+      "learning_rate": 2.606038757628795e-06,
+      "loss": 0.4262,
+      "step": 370
+    },
+    {
+      "epoch": 0.0372,
+      "learning_rate": 2.6107408749674112e-06,
+      "loss": 0.106,
+      "step": 372
+    },
+    {
+      "epoch": 0.0374,
+      "learning_rate": 2.615446593741161e-06,
+      "loss": 0.7671,
+      "step": 374
+    },
+    {
+      "epoch": 0.0376,
+      "learning_rate": 2.620155911656531e-06,
+      "loss": 0.4516,
+      "step": 376
+    },
+    {
+      "epoch": 0.0378,
+      "learning_rate": 2.6248688264182588e-06,
+      "loss": 0.077,
+      "step": 378
+    },
+    {
+      "epoch": 0.038,
+      "learning_rate": 2.6295853357293268e-06,
+      "loss": 0.5485,
+      "step": 380
+    },
+    {
+      "epoch": 0.0382,
+      "learning_rate": 2.6343054372909648e-06,
+      "loss": 0.6279,
+      "step": 382
+    },
+    {
+      "epoch": 0.0384,
+      "learning_rate": 2.6390291288026538e-06,
+      "loss": 0.7155,
+      "step": 384
+    },
+    {
+      "epoch": 0.0386,
+      "learning_rate": 2.6437564079621235e-06,
+      "loss": 0.5349,
+      "step": 386
+    },
+    {
+      "epoch": 0.0388,
+      "learning_rate": 2.648487272465354e-06,
+      "loss": 0.6749,
+      "step": 388
+    },
+    {
+      "epoch": 0.039,
+      "learning_rate": 2.6532217200065826e-06,
+      "loss": 0.7175,
+      "step": 390
+    },
+    {
+      "epoch": 0.0392,
+      "learning_rate": 2.657959748278294e-06,
+      "loss": 0.8628,
+      "step": 392
+    },
+    {
+      "epoch": 0.0394,
+      "learning_rate": 2.662701354971232e-06,
+      "loss": 0.0827,
+      "step": 394
+    },
+    {
+      "epoch": 0.0396,
+      "learning_rate": 2.667446537774395e-06,
+      "loss": 0.9572,
+      "step": 396
+    },
+    {
+      "epoch": 0.0398,
+      "learning_rate": 2.6721952943750396e-06,
+      "loss": 0.6057,
+      "step": 398
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6769476224586776e-06,
+      "loss": 0.1875,
+      "step": 400
+    },
+    {
+      "epoch": 0.0402,
+      "learning_rate": 2.6817035197090825e-06,
+      "loss": 0.8199,
+      "step": 402
+    },
+    {
+      "epoch": 0.0404,
+      "learning_rate": 2.686462983808289e-06,
+      "loss": 0.4394,
+      "step": 404
+    },
+    {
+      "epoch": 0.0406,
+      "learning_rate": 2.691226012436604e-06,
+      "loss": 0.415,
+      "step": 406
+    },
+    {
+      "epoch": 0.0408,
+      "learning_rate": 2.695992603272547e-06,
+      "loss": 0.4515,
+      "step": 408
+    },
+    {
+      "epoch": 0.041,
+      "learning_rate": 2.7007627539929783e-06,
+      "loss": 1.0465,
+      "step": 410
+    },
+    {
+      "epoch": 0.0412,
+      "learning_rate": 2.705536462272983e-06,
+      "loss": 0.2188,
+      "step": 412
+    },
+    {
+      "epoch": 0.0414,
+      "learning_rate": 2.7103137257858893e-06,
+      "loss": 0.2608,
+      "step": 414
+    },
+    {
+      "epoch": 0.0416,
+      "learning_rate": 2.715094542203318e-06,
+      "loss": 0.5616,
+      "step": 416
+    },
+    {
+      "epoch": 0.0418,
+      "learning_rate": 2.7198789091951806e-06,
+      "loss": 0.5226,
+      "step": 418
+    },
+    {
+      "epoch": 0.042,
+      "learning_rate": 2.724666824429635e-06,
+      "loss": 0.375,
+      "step": 420
+    },
+    {
+      "epoch": 0.0422,
+      "learning_rate": 2.7294582855730733e-06,
+      "loss": 1.2244,
+      "step": 422
+    },
+    {
+      "epoch": 0.0424,
+      "learning_rate": 2.734253290290232e-06,
+      "loss": 1.3735,
+      "step": 424
+    },
+    {
+      "epoch": 0.0426,
+      "learning_rate": 2.7390518362440843e-06,
+      "loss": 0.4655,
+      "step": 426
+    },
+    {
+      "epoch": 0.0428,
+      "learning_rate": 2.7438539210958517e-06,
+      "loss": 0.3845,
+      "step": 428
+    },
+    {
+      "epoch": 0.043,
+      "learning_rate": 2.7486595425050566e-06,
+      "loss": 0.5812,
+      "step": 430
+    },
+    {
+      "epoch": 0.0432,
+      "learning_rate": 2.7534686981295356e-06,
+      "loss": 0.1103,
+      "step": 432
+    },
+    {
+      "epoch": 0.0434,
+      "learning_rate": 2.7582813856253264e-06,
+      "loss": 0.319,
+      "step": 434
+    },
+    {
+      "epoch": 0.0436,
+      "learning_rate": 2.763097602646785e-06,
+      "loss": 0.9752,
+      "step": 436
+    },
+    {
+      "epoch": 0.0438,
+      "learning_rate": 2.7679173468465813e-06,
+      "loss": 0.7609,
+      "step": 438
+    },
+    {
+      "epoch": 0.044,
+      "learning_rate": 2.7727406158755943e-06,
+      "loss": 0.5718,
+      "step": 440
+    },
+    {
+      "epoch": 0.0442,
+      "learning_rate": 2.777567407383033e-06,
+      "loss": 0.669,
+      "step": 442
+    },
+    {
+      "epoch": 0.0444,
+      "learning_rate": 2.7823977190163665e-06,
+      "loss": 0.1296,
+      "step": 444
+    },
+    {
+      "epoch": 0.0446,
+      "learning_rate": 2.7872315484213954e-06,
+      "loss": 0.7953,
+      "step": 446
+    },
+    {
+      "epoch": 0.0448,
+      "learning_rate": 2.7920688932421337e-06,
+      "loss": 1.3948,
+      "step": 448
+    },
+    {
+      "epoch": 0.045,
+      "learning_rate": 2.796909751120931e-06,
+      "loss": 0.109,
+      "step": 450
+    },
+    {
+      "epoch": 0.0452,
+      "learning_rate": 2.801754119698411e-06,
+      "loss": 0.9909,
+      "step": 452
+    },
+    {
+      "epoch": 0.0454,
+      "learning_rate": 2.8066019966134873e-06,
+      "loss": 0.3756,
+      "step": 454
+    },
+    {
+      "epoch": 0.0456,
+      "learning_rate": 2.811453379503365e-06,
+      "loss": 0.3321,
+      "step": 456
+    },
+    {
+      "epoch": 0.0458,
+      "learning_rate": 2.816308266003538e-06,
+      "loss": 1.252,
+      "step": 458
+    },
+    {
+      "epoch": 0.046,
+      "learning_rate": 2.821166653747793e-06,
+      "loss": 1.1686,
+      "step": 460
+    },
+    {
+      "epoch": 0.0462,
+      "learning_rate": 2.826028540368212e-06,
+      "loss": 0.1396,
+      "step": 462
+    },
+    {
+      "epoch": 0.0464,
+      "learning_rate": 2.8308939234951695e-06,
+      "loss": 0.7137,
+      "step": 464
+    },
+    {
+      "epoch": 0.0466,
+      "learning_rate": 2.835762800757338e-06,
+      "loss": 0.4571,
+      "step": 466
+    },
+    {
+      "epoch": 0.0468,
+      "learning_rate": 2.8406351697816847e-06,
+      "loss": 0.7305,
+      "step": 468
+    },
+    {
+      "epoch": 0.047,
+      "learning_rate": 2.845511028193477e-06,
+      "loss": 0.4148,
+      "step": 470
+    },
+    {
+      "epoch": 0.0472,
+      "learning_rate": 2.850390373616281e-06,
+      "loss": 0.4394,
+      "step": 472
+    },
+    {
+      "epoch": 0.0474,
+      "learning_rate": 2.855273203671962e-06,
+      "loss": 0.5602,
+      "step": 474
+    },
+    {
+      "epoch": 0.0476,
+      "learning_rate": 2.8601595159806883e-06,
+      "loss": 0.9407,
+      "step": 476
+    },
+    {
+      "epoch": 0.0478,
+      "learning_rate": 2.865049308160931e-06,
+      "loss": 0.8256,
+      "step": 478
+    },
+    {
+      "epoch": 0.048,
+      "learning_rate": 2.8699425778294642e-06,
+      "loss": 0.3262,
+      "step": 480
+    },
+    {
+      "epoch": 0.0482,
+      "learning_rate": 2.874839322601368e-06,
+      "loss": 0.7832,
+      "step": 482
+    },
+    {
+      "epoch": 0.0484,
+      "learning_rate": 2.8797395400900294e-06,
+      "loss": 0.539,
+      "step": 484
+    },
+    {
+      "epoch": 0.0486,
+      "learning_rate": 2.8846432279071533e-06,
+      "loss": 0.3636,
+      "step": 486
+    },
+    {
+      "epoch": 0.0488,
+      "learning_rate": 2.8895503836627037e-06,
+      "loss": 0.5335,
+      "step": 488
+    },
+    {
+      "epoch": 0.049,
+      "learning_rate": 2.8944610049650314e-06,
+      "loss": 0.0939,
+      "step": 490
+    },
+    {
+      "epoch": 0.0492,
+      "learning_rate": 2.8993750894207597e-06,
+      "loss": 1.0111,
+      "step": 492
+    },
+    {
+      "epoch": 0.0494,
+      "learning_rate": 2.9042926346347835e-06,
+      "loss": 1.0884,
+      "step": 494
+    },
+    {
+      "epoch": 0.0496,
+      "learning_rate": 2.909213638210391e-06,
+      "loss": 0.0577,
+      "step": 496
+    },
+    {
+      "epoch": 0.0498,
+      "learning_rate": 2.914138097749143e-06,
+      "loss": 1.3655,
+      "step": 498
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.919066010850895e-06,
+      "loss": 0.9283,
+      "step": 500
+    },
+    {
+      "epoch": 0.0502,
+      "learning_rate": 2.9239973751138397e-06,
+      "loss": 0.6629,
+      "step": 502
+    },
+    {
+      "epoch": 0.0504,
+      "learning_rate": 2.928932188134529e-06,
+      "loss": 0.8002,
+      "step": 504
+    },
+    {
+      "epoch": 0.0506,
+      "learning_rate": 2.933870447507756e-06,
+      "loss": 0.4669,
+      "step": 506
+    },
+    {
+      "epoch": 0.0508,
+      "learning_rate": 2.9388121508266753e-06,
+      "loss": 0.2522,
+      "step": 508
+    },
+    {
+      "epoch": 0.051,
+      "learning_rate": 2.943757295682783e-06,
+      "loss": 0.542,
+      "step": 510
+    },
+    {
+      "epoch": 0.0512,
+      "learning_rate": 2.9487058796658785e-06,
+      "loss": 0.4526,
+      "step": 512
+    },
+    {
+      "epoch": 0.0514,
+      "learning_rate": 2.953657900364055e-06,
+      "loss": 0.1345,
+      "step": 514
+    },
+    {
+      "epoch": 0.0516,
+      "learning_rate": 2.9586133553637585e-06,
+      "loss": 0.1981,
+      "step": 516
+    },
+    {
+      "epoch": 0.0518,
+      "learning_rate": 2.9635722422497983e-06,
+      "loss": 0.0995,
+      "step": 518
+    },
+    {
+      "epoch": 0.052,
+      "learning_rate": 2.968534558605235e-06,
+      "loss": 0.3936,
+      "step": 520
+    },
+    {
+      "epoch": 0.0522,
+      "learning_rate": 2.973500302011496e-06,
+      "loss": 1.0543,
+      "step": 522
+    },
+    {
+      "epoch": 0.0524,
+      "learning_rate": 2.978469470048376e-06,
+      "loss": 1.2548,
+      "step": 524
+    },
+    {
+      "epoch": 0.0526,
+      "learning_rate": 2.983442060293926e-06,
+      "loss": 0.4494,
+      "step": 526
+    },
+    {
+      "epoch": 0.0528,
+      "learning_rate": 2.988418070324577e-06,
+      "loss": 0.5287,
+      "step": 528
+    },
+    {
+      "epoch": 0.053,
+      "learning_rate": 2.9933974977150827e-06,
+      "loss": 0.5844,
+      "step": 530
+    },
+    {
+      "epoch": 0.0532,
+      "learning_rate": 2.9983803400385313e-06,
+      "loss": 0.7371,
+      "step": 532
+    },
+    {
+      "epoch": 0.0534,
+      "learning_rate": 3.003366594866345e-06,
+      "loss": 1.1209,
+      "step": 534
+    },
+    {
+      "epoch": 0.0536,
+      "learning_rate": 3.008356259768285e-06,
+      "loss": 0.7095,
+      "step": 536
+    },
+    {
+      "epoch": 0.0538,
+      "learning_rate": 3.0133493323124474e-06,
+      "loss": 1.063,
+      "step": 538
+    },
+    {
+      "epoch": 0.054,
+      "learning_rate": 3.018345810065272e-06,
+      "loss": 0.4735,
+      "step": 540
+    },
+    {
+      "epoch": 0.0542,
+      "learning_rate": 3.0233456905915338e-06,
+      "loss": 0.4762,
+      "step": 542
+    },
+    {
+      "epoch": 0.0544,
+      "learning_rate": 3.0283489714543525e-06,
+      "loss": 0.3691,
+      "step": 544
+    },
+    {
+      "epoch": 0.0546,
+      "learning_rate": 3.0333556502151895e-06,
+      "loss": 0.5703,
+      "step": 546
+    },
+    {
+      "epoch": 0.0548,
+      "learning_rate": 3.0383657244338517e-06,
+      "loss": 0.317,
+      "step": 548
+    },
+    {
+      "epoch": 0.055,
+      "learning_rate": 3.0433791916684885e-06,
+      "loss": 0.6457,
+      "step": 550
+    },
+    {
+      "epoch": 0.0552,
+      "learning_rate": 3.0483960494755983e-06,
+      "loss": 0.5465,
+      "step": 552
+    },
+    {
+      "epoch": 0.0554,
+      "learning_rate": 3.0534162954100234e-06,
+      "loss": 0.6484,
+      "step": 554
+    },
+    {
+      "epoch": 0.0556,
+      "learning_rate": 3.0584399270249577e-06,
+      "loss": 0.7475,
+      "step": 556
+    },
+    {
+      "epoch": 0.0558,
+      "learning_rate": 3.0634669418719453e-06,
+      "loss": 0.4125,
+      "step": 558
+    },
+    {
+      "epoch": 0.056,
+      "learning_rate": 3.0684973375008797e-06,
+      "loss": 1.0152,
+      "step": 560
+    },
+    {
+      "epoch": 0.0562,
+      "learning_rate": 3.0735311114600064e-06,
+      "loss": 0.2235,
+      "step": 562
+    },
+    {
+      "epoch": 0.0564,
+      "learning_rate": 3.078568261295939e-06,
+      "loss": 0.6649,
+      "step": 564
+    },
+    {
+      "epoch": 0.0566,
+      "learning_rate": 3.0836087845535933e-06,
+      "loss": 0.9138,
+      "step": 566
+    },
+    {
+      "epoch": 0.0568,
+      "learning_rate": 3.088652678776316e-06,
+      "loss": 0.5789,
+      "step": 568
+    },
+    {
+      "epoch": 0.057,
+      "learning_rate": 3.0936999415057645e-06,
+      "loss": 0.4468,
+      "step": 570
+    },
+    {
+      "epoch": 0.0572,
+      "learning_rate": 3.0987505702819755e-06,
+      "loss": 1.0135,
+      "step": 572
+    },
+    {
+      "epoch": 0.0574,
+      "learning_rate": 3.1038045626432945e-06,
+      "loss": 0.0226,
+      "step": 574
+    },
+    {
+      "epoch": 0.0576,
+      "learning_rate": 3.1088619161265077e-06,
+      "loss": 1.28,
+      "step": 576
+    },
+    {
+      "epoch": 0.0578,
+      "learning_rate": 3.1139226282667212e-06,
+      "loss": 0.0478,
+      "step": 578
+    },
+    {
+      "epoch": 0.058,
+      "learning_rate": 3.118986696597367e-06,
+      "loss": 0.3813,
+      "step": 580
+    },
+    {
+      "epoch": 0.0582,
+      "learning_rate": 3.1240541186503173e-06,
+      "loss": 0.8472,
+      "step": 582
+    },
+    {
+      "epoch": 0.0584,
+      "learning_rate": 3.1291248919557738e-06,
+      "loss": 0.4013,
+      "step": 584
+    },
+    {
+      "epoch": 0.0586,
+      "learning_rate": 3.134199014042277e-06,
+      "loss": 0.8053,
+      "step": 586
+    },
+    {
+      "epoch": 0.0588,
+      "learning_rate": 3.1392764824367604e-06,
+      "loss": 0.0894,
+      "step": 588
+    },
+    {
+      "epoch": 0.059,
+      "learning_rate": 3.1443572946645683e-06,
+      "loss": 0.9445,
+      "step": 590
+    },
+    {
+      "epoch": 0.0592,
+      "learning_rate": 3.149441448249334e-06,
+      "loss": 0.6918,
+      "step": 592
+    },
+    {
+      "epoch": 0.0594,
+      "learning_rate": 3.154528940713103e-06,
+      "loss": 0.4439,
+      "step": 594
+    },
+    {
+      "epoch": 0.0596,
+      "learning_rate": 3.1596197695763198e-06,
+      "loss": 0.0581,
+      "step": 596
+    },
+    {
+      "epoch": 0.0598,
+      "learning_rate": 3.164713932357776e-06,
+      "loss": 1.007,
+      "step": 598
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.1698114265746126e-06,
+      "loss": 0.9652,
+      "step": 600
+    },
+    {
+      "epoch": 0.0602,
+      "learning_rate": 3.1749122497423724e-06,
+      "loss": 0.522,
+      "step": 602
+    },
+    {
+      "epoch": 0.0604,
+      "learning_rate": 3.1800163993750153e-06,
+      "loss": 0.6916,
+      "step": 604
+    },
+    {
+      "epoch": 0.0606,
+      "learning_rate": 3.1851238729848033e-06,
+      "loss": 0.4203,
+      "step": 606
+    },
+    {
+      "epoch": 0.0608,
+      "learning_rate": 3.190234668082427e-06,
+      "loss": 0.6603,
+      "step": 608
+    },
+    {
+      "epoch": 0.061,
+      "learning_rate": 3.195348782176948e-06,
+      "loss": 0.7432,
+      "step": 610
+    },
+    {
+      "epoch": 0.0612,
+      "learning_rate": 3.200466212775808e-06,
+      "loss": 0.3786,
+      "step": 612
+    },
+    {
+      "epoch": 0.0614,
+      "learning_rate": 3.205586957384834e-06,
+      "loss": 0.2522,
+      "step": 614
+    },
+    {
+      "epoch": 0.0616,
+      "learning_rate": 3.2107110135082376e-06,
+      "loss": 0.622,
+      "step": 616
+    },
+    {
+      "epoch": 0.0618,
+      "learning_rate": 3.215838378648617e-06,
+      "loss": 0.4707,
+      "step": 618
+    },
+    {
+      "epoch": 0.062,
+      "learning_rate": 3.2209690503069545e-06,
+      "loss": 0.4694,
+      "step": 620
+    },
+    {
+      "epoch": 0.0622,
+      "learning_rate": 3.2261030259826253e-06,
+      "loss": 0.4232,
+      "step": 622
+    },
+    {
+      "epoch": 0.0624,
+      "learning_rate": 3.231240303173391e-06,
+      "loss": 0.6249,
+      "step": 624
+    },
+    {
+      "epoch": 0.0626,
+      "learning_rate": 3.2363808793754036e-06,
+      "loss": 0.0042,
+      "step": 626
+    },
+    {
+      "epoch": 0.0628,
+      "learning_rate": 3.2415247520832115e-06,
+      "loss": 0.5909,
+      "step": 628
+    },
+    {
+      "epoch": 0.063,
+      "learning_rate": 3.246671918789752e-06,
+      "loss": 0.29,
+      "step": 630
+    },
+    {
+      "epoch": 0.0632,
+      "learning_rate": 3.25182237698636e-06,
+      "loss": 0.4668,
+      "step": 632
+    },
+    {
+      "epoch": 0.0634,
+      "learning_rate": 3.2569761241627617e-06,
+      "loss": 0.1413,
+      "step": 634
+    },
+    {
+      "epoch": 0.0636,
+      "learning_rate": 3.262133157807087e-06,
+      "loss": 0.649,
+      "step": 636
+    },
+    {
+      "epoch": 0.0638,
+      "learning_rate": 3.267293475405858e-06,
+      "loss": 1.1692,
+      "step": 638
+    },
+    {
+      "epoch": 0.064,
+      "learning_rate": 3.272457074443999e-06,
+      "loss": 0.586,
+      "step": 640
+    },
+    {
+      "epoch": 0.0642,
+      "learning_rate": 3.277623952404835e-06,
+      "loss": 0.496,
+      "step": 642
+    },
+    {
+      "epoch": 0.0644,
+      "learning_rate": 3.282794106770105e-06,
+      "loss": 0.5535,
+      "step": 644
+    },
+    {
+      "epoch": 0.0646,
+      "learning_rate": 3.2879675350199004e-06,
+      "loss": 0.6782,
+      "step": 646
+    },
+    {
+      "epoch": 0.0648,
+      "learning_rate": 3.2931442346327933e-06,
+      "loss": 0.5946,
+      "step": 648
+    },
+    {
+      "epoch": 0.065,
+      "learning_rate": 3.298324203085723e-06,
+      "loss": 0.8147,
+      "step": 650
+    },
+    {
+      "epoch": 0.0652,
+      "learning_rate": 3.3035074378539977e-06,
+      "loss": 0.9427,
+      "step": 652
+    },
+    {
+      "epoch": 0.0654,
+      "learning_rate": 3.3086939364114113e-06,
+      "loss": 0.0846,
+      "step": 654
+    },
+    {
+      "epoch": 0.0656,
+      "learning_rate": 3.313883696230111e-06,
+      "loss": 0.9309,
+      "step": 656
+    },
+    {
+      "epoch": 0.0658,
+      "learning_rate": 3.3190767147806892e-06,
+      "loss": 0.4872,
+      "step": 658
+    },
+    {
+      "epoch": 0.066,
+      "learning_rate": 3.324272989532088e-06,
+      "loss": 0.3411,
+      "step": 660
+    },
+    {
+      "epoch": 0.0662,
+      "learning_rate": 3.329472517951747e-06,
+      "loss": 0.8343,
+      "step": 662
+    },
+    {
+      "epoch": 0.0664,
+      "learning_rate": 3.3346752975054797e-06,
+      "loss": 0.3049,
+      "step": 664
+    },
+    {
+      "epoch": 0.0666,
+      "learning_rate": 3.3398813256574745e-06,
+      "loss": 0.4391,
+      "step": 666
+    },
+    {
+      "epoch": 0.0668,
+      "learning_rate": 3.3450905998704164e-06,
+      "loss": 0.4116,
+      "step": 668
+    },
+    {
+      "epoch": 0.067,
+      "learning_rate": 3.350303117605369e-06,
+      "loss": 0.0568,
+      "step": 670
+    },
+    {
+      "epoch": 0.0672,
+      "learning_rate": 3.355518876321787e-06,
+      "loss": 0.3581,
+      "step": 672
+    },
+    {
+      "epoch": 0.0674,
+      "learning_rate": 3.360737873477574e-06,
+      "loss": 0.3294,
+      "step": 674
+    },
+    {
+      "epoch": 0.0676,
+      "learning_rate": 3.3659601065290927e-06,
+      "loss": 0.4895,
+      "step": 676
+    },
+    {
+      "epoch": 0.0678,
+      "learning_rate": 3.3711855729310503e-06,
+      "loss": 0.9103,
+      "step": 678
+    },
+    {
+      "epoch": 0.068,
+      "learning_rate": 3.376414270136633e-06,
+      "loss": 0.5842,
+      "step": 680
+    },
+    {
+      "epoch": 0.0682,
+      "learning_rate": 3.3816461955974224e-06,
+      "loss": 0.4727,
+      "step": 682
+    },
+    {
+      "epoch": 0.0684,
+      "learning_rate": 3.386881346763482e-06,
+      "loss": 1.1475,
+      "step": 684
+    },
+    {
+      "epoch": 0.0686,
+      "learning_rate": 3.3921197210832235e-06,
+      "loss": 0.8573,
+      "step": 686
+    },
+    {
+      "epoch": 0.0688,
+      "learning_rate": 3.3973613160035255e-06,
+      "loss": 0.5311,
+      "step": 688
+    },
+    {
+      "epoch": 0.069,
+      "learning_rate": 3.4026061289697397e-06,
+      "loss": 0.6416,
+      "step": 690
+    },
+    {
+      "epoch": 0.0692,
+      "learning_rate": 3.4078541574255664e-06,
+      "loss": 0.2338,
+      "step": 692
+    },
+    {
+      "epoch": 0.0694,
+      "learning_rate": 3.4131053988131947e-06,
+      "loss": 0.2823,
+      "step": 694
+    },
+    {
+      "epoch": 0.0696,
+      "learning_rate": 3.418359850573233e-06,
+      "loss": 0.6163,
+      "step": 696
+    },
+    {
+      "epoch": 0.0698,
+      "learning_rate": 3.4236175101447257e-06,
+      "loss": 0.4697,
+      "step": 698
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.4288783749651534e-06,
+      "loss": 0.298,
+      "step": 700
+    },
+    {
+      "epoch": 0.0702,
+      "learning_rate": 3.434142442470434e-06,
+      "loss": 0.611,
+      "step": 702
+    },
+    {
+      "epoch": 0.0704,
+      "learning_rate": 3.439409710094925e-06,
+      "loss": 0.5332,
+      "step": 704
+    },
+    {
+      "epoch": 0.0706,
+      "learning_rate": 3.444680175271424e-06,
+      "loss": 0.2288,
+      "step": 706
+    },
+    {
+      "epoch": 0.0708,
+      "learning_rate": 3.4499538354311723e-06,
+      "loss": 0.2452,
+      "step": 708
+    },
+    {
+      "epoch": 0.071,
+      "learning_rate": 3.455230688003849e-06,
+      "loss": 0.9721,
+      "step": 710
+    },
+    {
+      "epoch": 0.0712,
+      "learning_rate": 3.460510730417582e-06,
+      "loss": 0.689,
+      "step": 712
+    },
+    {
+      "epoch": 0.0714,
+      "learning_rate": 3.465793960098942e-06,
+      "loss": 1.1275,
+      "step": 714
+    },
+    {
+      "epoch": 0.0716,
+      "learning_rate": 3.4710803744729483e-06,
+      "loss": 0.1863,
+      "step": 716
+    },
+    {
+      "epoch": 0.0718,
+      "learning_rate": 3.476369970963065e-06,
+      "loss": 0.5488,
+      "step": 718
+    },
+    {
+      "epoch": 0.072,
+      "learning_rate": 3.481662746991208e-06,
+      "loss": 0.5212,
+      "step": 720
+    },
+    {
+      "epoch": 0.0722,
+      "learning_rate": 3.486958699977743e-06,
+      "loss": 0.9406,
+      "step": 722
+    },
+    {
+      "epoch": 0.0724,
+      "learning_rate": 3.492257827341485e-06,
+      "loss": 0.6212,
+      "step": 724
+    },
+    {
+      "epoch": 0.0726,
+      "learning_rate": 3.497560126499706e-06,
+      "loss": 0.7326,
+      "step": 726
+    },
+    {
+      "epoch": 0.0728,
+      "learning_rate": 3.502865594868129e-06,
+      "loss": 1.1103,
+      "step": 728
+    },
+    {
+      "epoch": 0.073,
+      "learning_rate": 3.508174229860947e-06,
+      "loss": 0.175,
+      "step": 730
+    },
+    {
+      "epoch": 0.0732,
+      "learning_rate": 3.5134860288907534e-06,
+      "loss": 0.0901,
+      "step": 732
+    },
+    {
+      "epoch": 0.0734,
+      "learning_rate": 3.5188009893686836e-06,
+      "loss": 0.8669,
+      "step": 734
+    },
+    {
+      "epoch": 0.0736,
+      "learning_rate": 3.5241191087042915e-06,
+      "loss": 0.0849,
+      "step": 736
+    },
+    {
+      "epoch": 0.0738,
+      "learning_rate": 3.5294403843055493e-06,
+      "loss": 0.7977,
+      "step": 738
+    },
+    {
+      "epoch": 0.074,
+      "learning_rate": 3.5347648135789713e-06,
+      "loss": 0.2983,
+      "step": 740
+    },
+    {
+      "epoch": 0.0742,
+      "learning_rate": 3.5400923939294827e-06,
+      "loss": 1.0833,
+      "step": 742
+    },
+    {
+      "epoch": 0.0744,
+      "learning_rate": 3.5454231227604994e-06,
+      "loss": 0.9097,
+      "step": 744
+    },
+    {
+      "epoch": 0.0746,
+      "learning_rate": 3.5507569974738477e-06,
+      "loss": 0.9903,
+      "step": 746
+    },
+    {
+      "epoch": 0.0748,
+      "learning_rate": 3.556094015469903e-06,
+      "loss": 0.3188,
+      "step": 748
+    },
+    {
+      "epoch": 0.075,
+      "learning_rate": 3.5614341741474667e-06,
+      "loss": 0.3739,
+      "step": 750
+    },
+    {
+      "epoch": 0.0752,
+      "learning_rate": 3.566777470903784e-06,
+      "loss": 0.0303,
+      "step": 752
+    },
+    {
+      "epoch": 0.0754,
+      "learning_rate": 3.5721239031345966e-06,
+      "loss": 2.2915,
+      "step": 754
+    },
+    {
+      "epoch": 0.0756,
+      "learning_rate": 3.5774734682341592e-06,
+      "loss": 0.9681,
+      "step": 756
+    },
+    {
+      "epoch": 0.0758,
+      "learning_rate": 3.5828261635951177e-06,
+      "loss": 0.4996,
+      "step": 758
+    },
+    {
+      "epoch": 0.076,
+      "learning_rate": 3.588181986608634e-06,
+      "loss": 0.8069,
+      "step": 760
+    },
+    {
+      "epoch": 0.0762,
+      "learning_rate": 3.593540934664387e-06,
+      "loss": 0.7165,
+      "step": 762
+    },
+    {
+      "epoch": 0.0764,
+      "learning_rate": 3.598903005150447e-06,
+      "loss": 0.447,
+      "step": 764
+    },
+    {
+      "epoch": 0.0766,
+      "learning_rate": 3.604268195453421e-06,
+      "loss": 0.6212,
+      "step": 766
+    },
+    {
+      "epoch": 0.0768,
+      "learning_rate": 3.6096365029583803e-06,
+      "loss": 0.0863,
+      "step": 768
+    },
+    {
+      "epoch": 0.077,
+      "learning_rate": 3.6150079250488767e-06,
+      "loss": 1.3278,
+      "step": 770
+    },
+    {
+      "epoch": 0.0772,
+      "learning_rate": 3.620382459106946e-06,
+      "loss": 0.9827,
+      "step": 772
+    },
+    {
+      "epoch": 0.0774,
+      "learning_rate": 3.6257601025130893e-06,
+      "loss": 0.4101,
+      "step": 774
+    },
+    {
+      "epoch": 0.0776,
+      "learning_rate": 3.631140852646351e-06,
+      "loss": 0.0129,
+      "step": 776
+    },
+    {
+      "epoch": 0.0778,
+      "learning_rate": 3.636524706884178e-06,
+      "loss": 0.5927,
+      "step": 778
+    },
+    {
+      "epoch": 0.078,
+      "learning_rate": 3.6419116626025585e-06,
+      "loss": 0.0772,
+      "step": 780
+    },
+    {
+      "epoch": 0.0782,
+      "learning_rate": 3.647301717175955e-06,
+      "loss": 0.9892,
+      "step": 782
+    },
+    {
+      "epoch": 0.0784,
+      "learning_rate": 3.652694867977322e-06,
+      "loss": 0.5814,
+      "step": 784
+    },
+    {
+      "epoch": 0.0786,
+      "learning_rate": 3.6580911123781025e-06,
+      "loss": 0.4663,
+      "step": 786
+    },
+    {
+      "epoch": 0.0788,
+      "learning_rate": 3.6634904477482316e-06,
+      "loss": 1.0349,
+      "step": 788
+    },
+    {
+      "epoch": 0.079,
+      "learning_rate": 3.66889287145614e-06,
+      "loss": 1.1048,
+      "step": 790
+    },
+    {
+      "epoch": 0.0792,
+      "learning_rate": 3.6742983808687527e-06,
+      "loss": 0.6464,
+      "step": 792
+    },
+    {
+      "epoch": 0.0794,
+      "learning_rate": 3.679706973351488e-06,
+      "loss": 0.5394,
+      "step": 794
+    },
+    {
+      "epoch": 0.0796,
+      "learning_rate": 3.6851186462682653e-06,
+      "loss": 0.7495,
+      "step": 796
+    },
+    {
+      "epoch": 0.0798,
+      "learning_rate": 3.6905333969814995e-06,
+      "loss": 0.0116,
+      "step": 798
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.6959512228521077e-06,
+      "loss": 0.2478,
+      "step": 800
+    },
+    {
+      "epoch": 0.0802,
+      "learning_rate": 3.701372121239508e-06,
+      "loss": 0.5183,
+      "step": 802
+    },
+    {
+      "epoch": 0.0804,
+      "learning_rate": 3.706796089501621e-06,
+      "loss": 0.3346,
+      "step": 804
+    },
+    {
+      "epoch": 0.0806,
+      "learning_rate": 3.712223124994867e-06,
+      "loss": 0.5559,
+      "step": 806
+    },
+    {
+      "epoch": 0.0808,
+      "learning_rate": 3.717653225074178e-06,
+      "loss": 0.7566,
+      "step": 808
+    },
+    {
+      "epoch": 0.081,
+      "learning_rate": 3.723086387092989e-06,
+      "loss": 0.3403,
+      "step": 810
+    },
+    {
+      "epoch": 0.0812,
+      "learning_rate": 3.728522608403242e-06,
+      "loss": 0.474,
+      "step": 812
+    },
+    {
+      "epoch": 0.0814,
+      "learning_rate": 3.7339618863553885e-06,
+      "loss": 0.1862,
+      "step": 814
+    },
+    {
+      "epoch": 0.0816,
+      "learning_rate": 3.7394042182984046e-06,
+      "loss": 0.4992,
+      "step": 816
+    },
+    {
+      "epoch": 0.0818,
+      "learning_rate": 3.744849601579722e-06,
+      "loss": 0.914,
+      "step": 818
+    },
+    {
+      "epoch": 0.082,
+      "learning_rate": 3.7502980335453697e-06,
+      "loss": 0.925,
+      "step": 820
+    },
+    {
+      "epoch": 0.0822,
+      "learning_rate": 3.755749511539848e-06,
+      "loss": 0.72,
+      "step": 822
+    },
+    {
+      "epoch": 0.0824,
+      "learning_rate": 3.7612040329061295e-06,
+      "loss": 0.9931,
+      "step": 824
+    },
+    {
+      "epoch": 0.0826,
+      "learning_rate": 3.7666615949857897e-06,
+      "loss": 0.3798,
+      "step": 826
+    },
+    {
+      "epoch": 0.0828,
+      "learning_rate": 3.772122195118879e-06,
+      "loss": 0.2477,
+      "step": 828
+    },
+    {
+      "epoch": 0.083,
+      "learning_rate": 3.7775858306439404e-06,
+      "loss": 0.6303,
+      "step": 830
+    },
+    {
+      "epoch": 0.0832,
+      "learning_rate": 3.783052498898062e-06,
+      "loss": 0.9479,
+      "step": 832
+    },
+    {
+      "epoch": 0.0834,
+      "learning_rate": 3.7885221972168864e-06,
+      "loss": 0.6588,
+      "step": 834
+    },
+    {
+      "epoch": 0.0836,
+      "learning_rate": 3.7939949229345476e-06,
+      "loss": 0.1458,
+      "step": 836
+    },
+    {
+      "epoch": 0.0838,
+      "learning_rate": 3.799470673383677e-06,
+      "loss": 0.4872,
+      "step": 838
+    },
+    {
+      "epoch": 0.084,
+      "learning_rate": 3.8049494458954615e-06,
+      "loss": 0.5012,
+      "step": 840
+    },
+    {
+      "epoch": 0.0842,
+      "learning_rate": 3.810431237799657e-06,
+      "loss": 0.0508,
+      "step": 842
+    },
+    {
+      "epoch": 0.0844,
+      "learning_rate": 3.81591604642446e-06,
+      "loss": 0.6277,
+      "step": 844
+    },
+    {
+      "epoch": 0.0846,
+      "learning_rate": 3.821403869096644e-06,
+      "loss": 0.6583,
+      "step": 846
+    },
+    {
+      "epoch": 0.0848,
+      "learning_rate": 3.826894703141552e-06,
+      "loss": 0.0053,
+      "step": 848
+    },
+    {
+      "epoch": 0.085,
+      "learning_rate": 3.8323885458829745e-06,
+      "loss": 0.2335,
+      "step": 850
+    },
+    {
+      "epoch": 0.0852,
+      "learning_rate": 3.8378853946432956e-06,
+      "loss": 0.0052,
+      "step": 852
+    },
+    {
+      "epoch": 0.0854,
+      "learning_rate": 3.8433852467434175e-06,
+      "loss": 0.8769,
+      "step": 854
+    },
+    {
+      "epoch": 0.0856,
+      "learning_rate": 3.848888099502779e-06,
+      "loss": 0.9255,
+      "step": 856
+    },
+    {
+      "epoch": 0.0858,
+      "learning_rate": 3.854393950239355e-06,
+      "loss": 1.0684,
+      "step": 858
+    },
+    {
+      "epoch": 0.086,
+      "learning_rate": 3.859902796269649e-06,
+      "loss": 0.0625,
+      "step": 860
+    },
+    {
+      "epoch": 0.0862,
+      "learning_rate": 3.865414634908756e-06,
+      "loss": 0.3215,
+      "step": 862
+    },
+    {
+      "epoch": 0.0864,
+      "learning_rate": 3.870929463470233e-06,
+      "loss": 0.9766,
+      "step": 864
+    },
+    {
+      "epoch": 0.0866,
+      "learning_rate": 3.876447279266233e-06,
+      "loss": 0.3749,
+      "step": 866
+    },
+    {
+      "epoch": 0.0868,
+      "learning_rate": 3.88196807960744e-06,
+      "loss": 0.61,
+      "step": 868
+    },
+    {
+      "epoch": 0.087,
+      "learning_rate": 3.887491861803081e-06,
+      "loss": 0.5331,
+      "step": 870
+    },
+    {
+      "epoch": 0.0872,
+      "learning_rate": 3.893018623160934e-06,
+      "loss": 1.1132,
+      "step": 872
+    },
+    {
+      "epoch": 0.0874,
+      "learning_rate": 3.898548360987321e-06,
+      "loss": 0.4455,
+      "step": 874
+    },
+    {
+      "epoch": 0.0876,
+      "learning_rate": 3.9040810725871145e-06,
+      "loss": 0.0467,
+      "step": 876
+    },
+    {
+      "epoch": 0.0878,
+      "learning_rate": 3.909616755263741e-06,
+      "loss": 0.4355,
+      "step": 878
+    },
+    {
+      "epoch": 0.088,
+      "learning_rate": 3.9151554063191735e-06,
+      "loss": 0.5926,
+      "step": 880
+    },
+    {
+      "epoch": 0.0882,
+      "learning_rate": 3.920697023053941e-06,
+      "loss": 0.5583,
+      "step": 882
+    },
+    {
+      "epoch": 0.0884,
+      "learning_rate": 3.926241602767128e-06,
+      "loss": 0.0534,
+      "step": 884
+    },
+    {
+      "epoch": 0.0886,
+      "learning_rate": 3.9317891427563725e-06,
+      "loss": 0.8236,
+      "step": 886
+    },
+    {
+      "epoch": 0.0888,
+      "learning_rate": 3.937339640317887e-06,
+      "loss": 0.0593,
+      "step": 888
+    },
+    {
+      "epoch": 0.089,
+      "learning_rate": 3.942893092746381e-06,
+      "loss": 0.7405,
+      "step": 890
+    },
+    {
+      "epoch": 0.0892,
+      "learning_rate": 3.948449497335213e-06,
+      "loss": 0.7641,
+      "step": 892
+    },
+    {
+      "epoch": 0.0894,
+      "learning_rate": 3.954008851376244e-06,
+      "loss": 0.5605,
+      "step": 894
+    },
+    {
+      "epoch": 0.0896,
+      "learning_rate": 3.959571152159915e-06,
+      "loss": 0.4541,
+      "step": 896
+    },
+    {
+      "epoch": 0.0898,
+      "learning_rate": 3.965136396975227e-06,
+      "loss": 0.3771,
+      "step": 898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.970704583109746e-06,
+      "loss": 0.453,
+      "step": 900
+    },
+    {
+      "epoch": 0.0902,
+      "learning_rate": 3.976275707849619e-06,
+      "loss": 0.8942,
+      "step": 902
+    },
+    {
+      "epoch": 0.0904,
+      "learning_rate": 3.981849768479509e-06,
+      "loss": 0.1271,
+      "step": 904
+    },
+    {
+      "epoch": 0.0906,
+      "learning_rate": 3.987426762282726e-06,
+      "loss": 0.1962,
+      "step": 906
+    },
+    {
+      "epoch": 0.0908,
+      "learning_rate": 3.993006686541111e-06,
+      "loss": 0.8731,
+      "step": 908
+    },
+    {
+      "epoch": 0.091,
+      "learning_rate": 3.99858953853505e-06,
+      "loss": 1.0847,
+      "step": 910
+    },
+    {
+      "epoch": 0.0912,
+      "learning_rate": 4.004175315543527e-06,
+      "loss": 0.7916,
+      "step": 912
+    },
+    {
+      "epoch": 0.0914,
+      "learning_rate": 4.009764014844146e-06,
+      "loss": 0.3963,
+      "step": 914
+    },
+    {
+      "epoch": 0.0916,
+      "learning_rate": 4.015355633712999e-06,
+      "loss": 0.4799,
+      "step": 916
+    },
+    {
+      "epoch": 0.0918,
+      "learning_rate": 4.0209501694248e-06,
+      "loss": 0.0204,
+      "step": 918
+    },
+    {
+      "epoch": 0.092,
+      "learning_rate": 4.0265476192528685e-06,
+      "loss": 0.8358,
+      "step": 920
+    },
+    {
+      "epoch": 0.0922,
+      "learning_rate": 4.032147980469076e-06,
+      "loss": 0.9036,
+      "step": 922
+    },
+    {
+      "epoch": 0.0924,
+      "learning_rate": 4.037751250343844e-06,
+      "loss": 0.8317,
+      "step": 924
+    },
+    {
+      "epoch": 0.0926,
+      "learning_rate": 4.043357426146209e-06,
+      "loss": 0.9473,
+      "step": 926
+    },
+    {
+      "epoch": 0.0928,
+      "learning_rate": 4.048966505143831e-06,
+      "loss": 0.1746,
+      "step": 928
+    },
+    {
+      "epoch": 0.093,
+      "learning_rate": 4.054578484602869e-06,
+      "loss": 0.017,
+      "step": 930
+    },
+    {
+      "epoch": 0.0932,
+      "learning_rate": 4.060193361788115e-06,
+      "loss": 0.9029,
+      "step": 932
+    },
+    {
+      "epoch": 0.0934,
+      "learning_rate": 4.065811133962987e-06,
+      "loss": 0.0491,
+      "step": 934
+    },
+    {
+      "epoch": 0.0936,
+      "learning_rate": 4.071431798389408e-06,
+      "loss": 0.9337,
+      "step": 936
+    },
+    {
+      "epoch": 0.0938,
+      "learning_rate": 4.07705535232795e-06,
+      "loss": 0.696,
+      "step": 938
+    },
+    {
+      "epoch": 0.094,
+      "learning_rate": 4.08268179303776e-06,
+      "loss": 1.0118,
+      "step": 940
+    },
+    {
+      "epoch": 0.0942,
+      "learning_rate": 4.08831111777658e-06,
+      "loss": 0.5408,
+      "step": 942
+    },
+    {
+      "epoch": 0.0944,
+      "learning_rate": 4.093943323800746e-06,
+      "loss": 0.9339,
+      "step": 944
+    },
+    {
+      "epoch": 0.0946,
+      "learning_rate": 4.0995784083651865e-06,
+      "loss": 0.9793,
+      "step": 946
+    },
+    {
+      "epoch": 0.0948,
+      "learning_rate": 4.105216368723433e-06,
+      "loss": 0.0128,
+      "step": 948
+    },
+    {
+      "epoch": 0.095,
+      "learning_rate": 4.110857202127611e-06,
+      "loss": 0.3617,
+      "step": 950
+    },
+    {
+      "epoch": 0.0952,
+      "learning_rate": 4.116500905828445e-06,
+      "loss": 0.1626,
+      "step": 952
+    },
+    {
+      "epoch": 0.0954,
+      "learning_rate": 4.122147477075266e-06,
+      "loss": 0.109,
+      "step": 954
+    },
+    {
+      "epoch": 0.0956,
+      "learning_rate": 4.127796913116e-06,
+      "loss": 0.5961,
+      "step": 956
+    },
+    {
+      "epoch": 0.0958,
+      "learning_rate": 4.133449211197183e-06,
+      "loss": 0.2665,
+      "step": 958
+    },
+    {
+      "epoch": 0.096,
+      "learning_rate": 4.139104368563953e-06,
+      "loss": 0.2083,
+      "step": 960
+    },
+    {
+      "epoch": 0.0962,
+      "learning_rate": 4.144762382460055e-06,
+      "loss": 0.574,
+      "step": 962
+    },
+    {
+      "epoch": 0.0964,
+      "learning_rate": 4.1504232501278405e-06,
+      "loss": 0.3612,
+      "step": 964
+    },
+    {
+      "epoch": 0.0966,
+      "learning_rate": 4.156086968808274e-06,
+      "loss": 0.7134,
+      "step": 966
+    },
+    {
+      "epoch": 0.0968,
+      "learning_rate": 4.1617535357409245e-06,
+      "loss": 0.5934,
+      "step": 968
+    },
+    {
+      "epoch": 0.097,
+      "learning_rate": 4.1674229481639796e-06,
+      "loss": 0.9432,
+      "step": 970
+    },
+    {
+      "epoch": 0.0972,
+      "learning_rate": 4.1730952033142335e-06,
+      "loss": 0.2004,
+      "step": 972
+    },
+    {
+      "epoch": 0.0974,
+      "learning_rate": 4.178770298427114e-06,
+      "loss": 0.7655,
+      "step": 974
+    },
+    {
+      "epoch": 0.0976,
+      "learning_rate": 4.184448230736605e-06,
+      "loss": 0.3076,
+      "step": 976
+    },
+    {
+      "epoch": 0.0978,
+      "learning_rate": 4.190128997475395e-06,
+      "loss": 0.1405,
+      "step": 978
+    },
+    {
+      "epoch": 0.098,
+      "learning_rate": 4.195812595874732e-06,
+      "loss": 0.3993,
+      "step": 980
+    },
+    {
+      "epoch": 0.0982,
+      "learning_rate": 4.201499023164515e-06,
+      "loss": 0.0718,
+      "step": 982
+    },
+    {
+      "epoch": 0.0984,
+      "learning_rate": 4.207188276573203e-06,
+      "loss": 0.4838,
+      "step": 984
+    },
+    {
+      "epoch": 0.0986,
+      "learning_rate": 4.212880353327968e-06,
+      "loss": 0.7306,
+      "step": 986
+    },
+    {
+      "epoch": 0.0988,
+      "learning_rate": 4.218575250654562e-06,
+      "loss": 0.4279,
+      "step": 988
+    },
+    {
+      "epoch": 0.099,
+      "learning_rate": 4.224272965777315e-06,
+      "loss": 0.9562,
+      "step": 990
+    },
+    {
+      "epoch": 0.0992,
+      "learning_rate": 4.229973495919275e-06,
+      "loss": 0.8616,
+      "step": 992
+    },
+    {
+      "epoch": 0.0994,
+      "learning_rate": 4.235676838302072e-06,
+      "loss": 0.8299,
+      "step": 994
+    },
+    {
+      "epoch": 0.0996,
+      "learning_rate": 4.241382990145938e-06,
+      "loss": 1.0855,
+      "step": 996
+    },
+    {
+      "epoch": 0.0998,
+      "learning_rate": 4.247091948669764e-06,
+      "loss": 0.2342,
+      "step": 998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.252803711091116e-06,
+      "loss": 0.5816,
+      "step": 1000
+    },
+    {
+      "epoch": 0.1002,
+      "learning_rate": 4.258518274626106e-06,
+      "loss": 0.3919,
+      "step": 1002
+    },
+    {
+      "epoch": 0.1004,
+      "learning_rate": 4.264235636489527e-06,
+      "loss": 0.0257,
+      "step": 1004
+    },
+    {
+      "epoch": 0.1006,
+      "learning_rate": 4.269955793894849e-06,
+      "loss": 0.6289,
+      "step": 1006
+    },
+    {
+      "epoch": 0.1008,
+      "learning_rate": 4.275678744054094e-06,
+      "loss": 0.3056,
+      "step": 1008
+    },
+    {
+      "epoch": 0.101,
+      "learning_rate": 4.281404484177978e-06,
+      "loss": 0.0279,
+      "step": 1010
+    },
+    {
+      "epoch": 0.1012,
+      "learning_rate": 4.287133011475831e-06,
+      "loss": 0.0803,
+      "step": 1012
+    },
+    {
+      "epoch": 0.1014,
+      "learning_rate": 4.292864323155684e-06,
+      "loss": 0.9656,
+      "step": 1014
+    },
+    {
+      "epoch": 0.1016,
+      "learning_rate": 4.29859841642412e-06,
+      "loss": 0.1677,
+      "step": 1016
+    },
+    {
+      "epoch": 0.1018,
+      "learning_rate": 4.304335288486412e-06,
+      "loss": 0.4161,
+      "step": 1018
+    },
+    {
+      "epoch": 0.102,
+      "learning_rate": 4.310074936546521e-06,
+      "loss": 0.8779,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1022,
+      "learning_rate": 4.3158173578069696e-06,
+      "loss": 0.9593,
+      "step": 1022
+    },
+    {
+      "epoch": 0.1024,
+      "learning_rate": 4.321562549468986e-06,
+      "loss": 0.758,
+      "step": 1024
+    },
+    {
+      "epoch": 0.1026,
+      "learning_rate": 4.327310508732434e-06,
+      "loss": 0.9659,
+      "step": 1026
+    },
+    {
+      "epoch": 0.1028,
+      "learning_rate": 4.3330612327958265e-06,
+      "loss": 0.5595,
+      "step": 1028
+    },
+    {
+      "epoch": 0.103,
+      "learning_rate": 4.338814718856333e-06,
+      "loss": 0.1003,
+      "step": 1030
+    },
+    {
+      "epoch": 0.1032,
+      "learning_rate": 4.34457096410977e-06,
+      "loss": 0.6302,
+      "step": 1032
+    },
+    {
+      "epoch": 0.1034,
+      "learning_rate": 4.350329965750618e-06,
+      "loss": 0.3749,
+      "step": 1034
+    },
+    {
+      "epoch": 0.1036,
+      "learning_rate": 4.356091720972008e-06,
+      "loss": 0.3598,
+      "step": 1036
+    },
+    {
+      "epoch": 0.1038,
+      "learning_rate": 4.3618562269657285e-06,
+      "loss": 0.0326,
+      "step": 1038
+    },
+    {
+      "epoch": 0.104,
+      "learning_rate": 4.367623480922232e-06,
+      "loss": 0.5896,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1042,
+      "learning_rate": 4.373393480030629e-06,
+      "loss": 0.0659,
+      "step": 1042
+    },
+    {
+      "epoch": 0.1044,
+      "learning_rate": 4.37916622147869e-06,
+      "loss": 0.0276,
+      "step": 1044
+    },
+    {
+      "epoch": 0.1046,
+      "learning_rate": 4.384941702452852e-06,
+      "loss": 0.9573,
+      "step": 1046
+    },
+    {
+      "epoch": 0.1048,
+      "learning_rate": 4.390719920138216e-06,
+      "loss": 0.8881,
+      "step": 1048
+    },
+    {
+      "epoch": 0.105,
+      "learning_rate": 4.396500871718548e-06,
+      "loss": 0.5192,
+      "step": 1050
+    },
+    {
+      "epoch": 0.1052,
+      "learning_rate": 4.402284554376284e-06,
+      "loss": 0.4415,
+      "step": 1052
+    },
+    {
+      "epoch": 0.1054,
+      "learning_rate": 4.408070965292526e-06,
+      "loss": 0.3758,
+      "step": 1054
+    },
+    {
+      "epoch": 0.1056,
+      "learning_rate": 4.413860101647047e-06,
+      "loss": 1.1926,
+      "step": 1056
+    },
+    {
+      "epoch": 0.1058,
+      "learning_rate": 4.419651960618294e-06,
+      "loss": 0.6099,
+      "step": 1058
+    },
+    {
+      "epoch": 0.106,
+      "learning_rate": 4.4254465393834e-06,
+      "loss": 0.9338,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1062,
+      "learning_rate": 4.431243835118112e-06,
+      "loss": 0.5815,
+      "step": 1062
+    },
+    {
+      "epoch": 0.1064,
+      "learning_rate": 4.4370438449969445e-06,
+      "loss": 1.2615,
+      "step": 1064
+    },
+    {
+      "epoch": 0.1066,
+      "learning_rate": 4.442846566193041e-06,
+      "loss": 0.4088,
+      "step": 1066
+    },
+    {
+      "epoch": 0.1068,
+      "learning_rate": 4.448651995878197e-06,
+      "loss": 0.0427,
+      "step": 1068
+    },
+    {
+      "epoch": 0.107,
+      "learning_rate": 4.4544601312229185e-06,
+      "loss": 0.6214,
+      "step": 1070
+    },
+    {
+      "epoch": 0.1072,
+      "learning_rate": 4.460270969396418e-06,
+      "loss": 0.827,
+      "step": 1072
+    },
+    {
+      "epoch": 0.1074,
+      "learning_rate": 4.4660845075665635e-06,
+      "loss": 0.0257,
+      "step": 1074
+    },
+    {
+      "epoch": 0.1076,
+      "learning_rate": 4.471900742899865e-06,
+      "loss": 0.4853,
+      "step": 1076
+    },
+    {
+      "epoch": 0.1078,
+      "learning_rate": 4.477719672561602e-06,
+      "loss": 0.1204,
+      "step": 1078
+    },
+    {
+      "epoch": 0.108,
+      "learning_rate": 4.483541293715702e-06,
+      "loss": 0.6071,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1082,
+      "learning_rate": 4.489365603524743e-06,
+      "loss": 0.5578,
+      "step": 1082
+    },
+    {
+      "epoch": 0.1084,
+      "learning_rate": 4.495192599150032e-06,
+      "loss": 0.6612,
+      "step": 1084
+    },
+    {
+      "epoch": 0.1086,
+      "learning_rate": 4.501022277751605e-06,
+      "loss": 0.2386,
+      "step": 1086
+    },
+    {
+      "epoch": 0.1088,
+      "learning_rate": 4.506854636488101e-06,
+      "loss": 0.7253,
+      "step": 1088
+    },
+    {
+      "epoch": 0.109,
+      "learning_rate": 4.5126896725169025e-06,
+      "loss": 0.4165,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1092,
+      "learning_rate": 4.518527382994127e-06,
+      "loss": 0.0665,
+      "step": 1092
+    },
+    {
+      "epoch": 0.1094,
+      "learning_rate": 4.524367765074499e-06,
+      "loss": 0.3614,
+      "step": 1094
+    },
+    {
+      "epoch": 0.1096,
+      "learning_rate": 4.530210815911504e-06,
+      "loss": 0.1531,
+      "step": 1096
+    },
+    {
+      "epoch": 0.1098,
+      "learning_rate": 4.536056532657295e-06,
+      "loss": 0.0004,
+      "step": 1098
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.541904912462785e-06,
+      "loss": 0.6418,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1102,
+      "learning_rate": 4.5477559524775e-06,
+      "loss": 0.1593,
+      "step": 1102
+    },
+    {
+      "epoch": 0.1104,
+      "learning_rate": 4.553609649849729e-06,
+      "loss": 0.7098,
+      "step": 1104
+    },
+    {
+      "epoch": 0.1106,
+      "learning_rate": 4.559466001726451e-06,
+      "loss": 0.3389,
+      "step": 1106
+    },
+    {
+      "epoch": 0.1108,
+      "learning_rate": 4.565325005253352e-06,
+      "loss": 0.1018,
+      "step": 1108
+    },
+    {
+      "epoch": 0.111,
+      "learning_rate": 4.571186657574823e-06,
+      "loss": 0.3894,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1112,
+      "learning_rate": 4.577050955833968e-06,
+      "loss": 0.5579,
+      "step": 1112
+    },
+    {
+      "epoch": 0.1114,
+      "learning_rate": 4.582917897172599e-06,
+      "loss": 0.7133,
+      "step": 1114
+    },
+    {
+      "epoch": 0.1116,
+      "learning_rate": 4.588787478731239e-06,
+      "loss": 0.709,
+      "step": 1116
+    },
+    {
+      "epoch": 0.1118,
+      "learning_rate": 4.5946596976491254e-06,
+      "loss": 0.4973,
+      "step": 1118
+    },
+    {
+      "epoch": 0.112,
+      "learning_rate": 4.600534551064211e-06,
+      "loss": 0.5036,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1122,
+      "learning_rate": 4.6064120361131624e-06,
+      "loss": 2.5019,
+      "step": 1122
+    },
+    {
+      "epoch": 0.1124,
+      "learning_rate": 4.612292149931365e-06,
+      "loss": 0.479,
+      "step": 1124
+    },
+    {
+      "epoch": 0.1126,
+      "learning_rate": 4.618174889652924e-06,
+      "loss": 0.1447,
+      "step": 1126
+    },
+    {
+      "epoch": 0.1128,
+      "learning_rate": 4.6240602524106625e-06,
+      "loss": 0.8194,
+      "step": 1128
+    },
+    {
+      "epoch": 0.113,
+      "learning_rate": 4.629948235336126e-06,
+      "loss": 0.9221,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1132,
+      "learning_rate": 4.635838835559583e-06,
+      "loss": 0.0687,
+      "step": 1132
+    },
+    {
+      "epoch": 0.1134,
+      "learning_rate": 4.6417320502100286e-06,
+      "loss": 0.1404,
+      "step": 1134
+    },
+    {
+      "epoch": 0.1136,
+      "learning_rate": 4.647627876415178e-06,
+      "loss": 0.5761,
+      "step": 1136
+    },
+    {
+      "epoch": 0.1138,
+      "learning_rate": 4.653526311301479e-06,
+      "loss": 0.6115,
+      "step": 1138
+    },
+    {
+      "epoch": 0.114,
+      "learning_rate": 4.659427351994123e-06,
+      "loss": 0.8348,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1142,
+      "learning_rate": 4.665330995616967e-06,
+      "loss": 1.163,
+      "step": 1142
+    },
+    {
+      "epoch": 0.1144,
+      "learning_rate": 4.671237239292692e-06,
+      "loss": 0.8207,
+      "step": 1144
+    },
+    {
+      "epoch": 0.1146,
+      "learning_rate": 4.677146080142667e-06,
+      "loss": 0.0113,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1148,
+      "learning_rate": 4.6830575152869505e-06,
+      "loss": 0.0555,
+      "step": 1148
+    },
+    {
+      "epoch": 0.115,
+      "learning_rate": 4.688971541844424e-06,
+      "loss": 0.0038,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1152,
+      "learning_rate": 4.694888156932666e-06,
+      "loss": 1.0354,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1154,
+      "learning_rate": 4.700807357667956e-06,
+      "loss": 0.9315,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1156,
+      "learning_rate": 4.70672914116535e-06,
+      "loss": 0.0435,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1158,
+      "learning_rate": 4.712653504538672e-06,
+      "loss": 0.7449,
+      "step": 1158
+    },
+    {
+      "epoch": 0.116,
+      "learning_rate": 4.71858044490046e-06,
+      "loss": 0.1125,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1162,
+      "learning_rate": 4.7245099593619495e-06,
+      "loss": 1.2507,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1164,
+      "learning_rate": 4.7304420450332134e-06,
+      "loss": 1.296,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1166,
+      "learning_rate": 4.736376699023023e-06,
+      "loss": 0.4292,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1168,
+      "learning_rate": 4.742313918438871e-06,
+      "loss": 0.2873,
+      "step": 1168
+    },
+    {
+      "epoch": 0.117,
+      "learning_rate": 4.74825370038703e-06,
+      "loss": 0.9469,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1172,
+      "learning_rate": 4.754196041972566e-06,
+      "loss": 0.8364,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1174,
+      "learning_rate": 4.76014094029921e-06,
+      "loss": 0.6775,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1176,
+      "learning_rate": 4.766088392469505e-06,
+      "loss": 0.546,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1178,
+      "learning_rate": 4.772038395584735e-06,
+      "loss": 0.6121,
+      "step": 1178
+    },
+    {
+      "epoch": 0.118,
+      "learning_rate": 4.7779909467449416e-06,
+      "loss": 0.0293,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1182,
+      "learning_rate": 4.7839460430489216e-06,
+      "loss": 0.4648,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1184,
+      "learning_rate": 4.789903681594221e-06,
+      "loss": 1.4081,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1186,
+      "learning_rate": 4.7958638594772035e-06,
+      "loss": 0.745,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1188,
+      "learning_rate": 4.801826573792903e-06,
+      "loss": 0.5329,
+      "step": 1188
+    },
+    {
+      "epoch": 0.119,
+      "learning_rate": 4.807791821635185e-06,
+      "loss": 0.8145,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1192,
+      "learning_rate": 4.813759600096657e-06,
+      "loss": 0.6169,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1194,
+      "learning_rate": 4.8197299062686954e-06,
+      "loss": 0.7143,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1196,
+      "learning_rate": 4.825702737241448e-06,
+      "loss": 0.3842,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1198,
+      "learning_rate": 4.831678090103828e-06,
+      "loss": 0.8108,
+      "step": 1198
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.837655961943521e-06,
+      "loss": 1.0677,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1202,
+      "learning_rate": 4.8436363498469865e-06,
+      "loss": 0.7445,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1204,
+      "learning_rate": 4.849619250899454e-06,
+      "loss": 0.5388,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1206,
+      "learning_rate": 4.855604662184931e-06,
+      "loss": 1.178,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1208,
+      "learning_rate": 4.861592580786201e-06,
+      "loss": 0.4867,
+      "step": 1208
+    },
+    {
+      "epoch": 0.121,
+      "learning_rate": 4.867583003784825e-06,
+      "loss": 1.0766,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1212,
+      "learning_rate": 4.873575928261159e-06,
+      "loss": 0.1136,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1214,
+      "learning_rate": 4.8795713512942785e-06,
+      "loss": 0.1229,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1216,
+      "learning_rate": 4.885569269962133e-06,
+      "loss": 0.7161,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1218,
+      "learning_rate": 4.891569681341395e-06,
+      "loss": 0.5169,
+      "step": 1218
+    },
+    {
+      "epoch": 0.122,
+      "learning_rate": 4.8975725825075356e-06,
+      "loss": 0.0126,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1222,
+      "learning_rate": 4.903577970534815e-06,
+      "loss": 0.9696,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1224,
+      "learning_rate": 4.90958584249628e-06,
+      "loss": 0.7897,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1226,
+      "learning_rate": 4.91559619546378e-06,
+      "loss": 0.9631,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1228,
+      "learning_rate": 4.921609026507899e-06,
+      "loss": 0.6589,
+      "step": 1228
+    },
+    {
+      "epoch": 0.123,
+      "learning_rate": 4.9276243326981e-06,
+      "loss": 0.5349,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1232,
+      "learning_rate": 4.933642111102598e-06,
+      "loss": 0.04,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1234,
+      "learning_rate": 4.939662358788352e-06,
+      "loss": 0.1037,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1236,
+      "learning_rate": 4.9456850728212145e-06,
+      "loss": 0.2466,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1238,
+      "learning_rate": 4.951710250265788e-06,
+      "loss": 0.4235,
+      "step": 1238
+    },
+    {
+      "epoch": 0.124,
+      "learning_rate": 4.957737888185442e-06,
+      "loss": 0.5749,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1242,
+      "learning_rate": 4.96376798364238e-06,
+      "loss": 0.7125,
+      "step": 1242
+    },
+    {
+      "epoch": 0.1244,
+      "learning_rate": 4.969800533697639e-06,
+      "loss": 0.5044,
+      "step": 1244
+    },
+    {
+      "epoch": 0.1246,
+      "learning_rate": 4.975835535411023e-06,
+      "loss": 0.0354,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1248,
+      "learning_rate": 4.981872985841118e-06,
+      "loss": 0.6203,
+      "step": 1248
+    },
+    {
+      "epoch": 0.125,
+      "learning_rate": 4.987912882045345e-06,
+      "loss": 0.0862,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1252,
+      "learning_rate": 4.993955221079976e-06,
+      "loss": 0.0145,
+      "step": 1252
+    },
+    {
+      "epoch": 0.1254,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.1055,
+      "step": 1254
+    },
+    {
+      "epoch": 0.1256,
+      "learning_rate": 5.0060472158592724e-06,
+      "loss": 1.0853,
+      "step": 1256
+    },
+    {
+      "epoch": 0.1258,
+      "learning_rate": 5.012096865710493e-06,
+      "loss": 0.5271,
+      "step": 1258
+    },
+    {
+      "epoch": 0.126,
+      "learning_rate": 5.018148946605092e-06,
+      "loss": 0.6021,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1262,
+      "learning_rate": 5.024203455593375e-06,
+      "loss": 0.5064,
+      "step": 1262
+    },
+    {
+      "epoch": 0.1264,
+      "learning_rate": 5.030260389724446e-06,
+      "loss": 0.6263,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1266,
+      "learning_rate": 5.036319746046232e-06,
+      "loss": 0.8112,
+      "step": 1266
+    },
+    {
+      "epoch": 0.1268,
+      "learning_rate": 5.042381521605473e-06,
+      "loss": 0.9785,
+      "step": 1268
+    },
+    {
+      "epoch": 0.127,
+      "learning_rate": 5.048445713447734e-06,
+      "loss": 0.1772,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1272,
+      "learning_rate": 5.054512318617402e-06,
+      "loss": 0.5534,
+      "step": 1272
+    },
+    {
+      "epoch": 0.1274,
+      "learning_rate": 5.0605813341576885e-06,
+      "loss": 0.4529,
+      "step": 1274
+    },
+    {
+      "epoch": 0.1276,
+      "learning_rate": 5.066652757110628e-06,
+      "loss": 0.4775,
+      "step": 1276
+    },
+    {
+      "epoch": 0.1278,
+      "learning_rate": 5.072726584517083e-06,
+      "loss": 0.4274,
+      "step": 1278
+    },
+    {
+      "epoch": 0.128,
+      "learning_rate": 5.078802813416741e-06,
+      "loss": 0.742,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1282,
+      "learning_rate": 5.084881440848126e-06,
+      "loss": 0.9233,
+      "step": 1282
+    },
+    {
+      "epoch": 0.1284,
+      "learning_rate": 5.090962463848588e-06,
+      "loss": 0.1187,
+      "step": 1284
+    },
+    {
+      "epoch": 0.1286,
+      "learning_rate": 5.097045879454308e-06,
+      "loss": 1.2247,
+      "step": 1286
+    },
+    {
+      "epoch": 0.1288,
+      "learning_rate": 5.103131684700306e-06,
+      "loss": 0.364,
+      "step": 1288
+    },
+    {
+      "epoch": 0.129,
+      "learning_rate": 5.109219876620433e-06,
+      "loss": 1.0716,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1292,
+      "learning_rate": 5.115310452247377e-06,
+      "loss": 0.0063,
+      "step": 1292
+    },
+    {
+      "epoch": 0.1294,
+      "learning_rate": 5.1214034086126685e-06,
+      "loss": 0.7507,
+      "step": 1294
+    },
+    {
+      "epoch": 0.1296,
+      "learning_rate": 5.127498742746671e-06,
+      "loss": 0.611,
+      "step": 1296
+    },
+    {
+      "epoch": 0.1298,
+      "learning_rate": 5.133596451678611e-06,
+      "loss": 0.0249,
+      "step": 1298
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.139696532436491e-06,
+      "loss": 0.7807,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1302,
+      "learning_rate": 5.145798982047253e-06,
+      "loss": 0.4715,
+      "step": 1302
+    },
+    {
+      "epoch": 0.1304,
+      "learning_rate": 5.151903797536637e-06,
+      "loss": 0.409,
+      "step": 1304
+    },
+    {
+      "epoch": 0.1306,
+      "learning_rate": 5.158010975929185e-06,
+      "loss": 0.8903,
+      "step": 1306
+    },
+    {
+      "epoch": 0.1308,
+      "learning_rate": 5.164120514248376e-06,
+      "loss": 0.8635,
+      "step": 1308
+    },
+    {
+      "epoch": 0.131,
+      "learning_rate": 5.170232409516483e-06,
+      "loss": 0.5848,
+      "step": 1310
+    },
+    {
+      "epoch": 0.1312,
+      "learning_rate": 5.176346658754656e-06,
+      "loss": 1.3437,
+      "step": 1312
+    },
+    {
+      "epoch": 0.1314,
+      "learning_rate": 5.182463258982837e-06,
+      "loss": 0.8974,
+      "step": 1314
+    },
+    {
+      "epoch": 0.1316,
+      "learning_rate": 5.188582207219919e-06,
+      "loss": 0.3384,
+      "step": 1316
+    },
+    {
+      "epoch": 0.1318,
+      "learning_rate": 5.194703500483597e-06,
+      "loss": 0.1355,
+      "step": 1318
+    },
+    {
+      "epoch": 0.132,
+      "learning_rate": 5.2008271357903835e-06,
+      "loss": 0.4312,
+      "step": 1320
+    },
+    {
+      "epoch": 0.1322,
+      "learning_rate": 5.2069531101557395e-06,
+      "loss": 0.3205,
+      "step": 1322
+    },
+    {
+      "epoch": 0.1324,
+      "learning_rate": 5.213081420593936e-06,
+      "loss": 0.6328,
+      "step": 1324
+    },
+    {
+      "epoch": 0.1326,
+      "learning_rate": 5.219212064118082e-06,
+      "loss": 0.4965,
+      "step": 1326
+    },
+    {
+      "epoch": 0.1328,
+      "learning_rate": 5.225345037740175e-06,
+      "loss": 0.329,
+      "step": 1328
+    },
+    {
+      "epoch": 0.133,
+      "learning_rate": 5.231480338471124e-06,
+      "loss": 1.2944,
+      "step": 1330
+    },
+    {
+      "epoch": 0.1332,
+      "learning_rate": 5.237617963320612e-06,
+      "loss": 0.5443,
+      "step": 1332
+    },
+    {
+      "epoch": 0.1334,
+      "learning_rate": 5.24375790929725e-06,
+      "loss": 1.1617,
+      "step": 1334
+    },
+    {
+      "epoch": 0.1336,
+      "learning_rate": 5.2499001734084884e-06,
+      "loss": 2.8705,
+      "step": 1336
+    },
+    {
+      "epoch": 0.1338,
+      "learning_rate": 5.256044752660709e-06,
+      "loss": 0.0114,
+      "step": 1338
+    },
+    {
+      "epoch": 0.134,
+      "learning_rate": 5.262191644059071e-06,
+      "loss": 0.772,
+      "step": 1340
+    },
+    {
+      "epoch": 0.1342,
+      "learning_rate": 5.268340844607653e-06,
+      "loss": 0.2271,
+      "step": 1342
+    },
+    {
+      "epoch": 0.1344,
+      "learning_rate": 5.27449235130946e-06,
+      "loss": 0.9173,
+      "step": 1344
+    },
+    {
+      "epoch": 0.1346,
+      "learning_rate": 5.2806461611662725e-06,
+      "loss": 0.274,
+      "step": 1346
+    },
+    {
+      "epoch": 0.1348,
+      "learning_rate": 5.286802271178815e-06,
+      "loss": 1.1225,
+      "step": 1348
+    },
+    {
+      "epoch": 0.135,
+      "learning_rate": 5.2929606783466735e-06,
+      "loss": 0.9891,
+      "step": 1350
+    },
+    {
+      "epoch": 0.1352,
+      "learning_rate": 5.299121379668316e-06,
+      "loss": 1.0171,
+      "step": 1352
+    },
+    {
+      "epoch": 0.1354,
+      "learning_rate": 5.305284372141091e-06,
+      "loss": 0.3997,
+      "step": 1354
+    },
+    {
+      "epoch": 0.1356,
+      "learning_rate": 5.311449652761232e-06,
+      "loss": 0.6876,
+      "step": 1356
+    },
+    {
+      "epoch": 0.1358,
+      "learning_rate": 5.317617218523853e-06,
+      "loss": 0.6878,
+      "step": 1358
+    },
+    {
+      "epoch": 0.136,
+      "learning_rate": 5.323787066422961e-06,
+      "loss": 0.6978,
+      "step": 1360
+    },
+    {
+      "epoch": 0.1362,
+      "learning_rate": 5.3299591934514435e-06,
+      "loss": 0.4185,
+      "step": 1362
+    },
+    {
+      "epoch": 0.1364,
+      "learning_rate": 5.336133596601085e-06,
+      "loss": 1.1262,
+      "step": 1364
+    },
+    {
+      "epoch": 0.1366,
+      "learning_rate": 5.342310272862553e-06,
+      "loss": 0.0274,
+      "step": 1366
+    },
+    {
+      "epoch": 0.1368,
+      "learning_rate": 5.348489219225412e-06,
+      "loss": 0.6422,
+      "step": 1368
+    },
+    {
+      "epoch": 0.137,
+      "learning_rate": 5.354670432678119e-06,
+      "loss": 0.4035,
+      "step": 1370
+    },
+    {
+      "epoch": 0.1372,
+      "learning_rate": 5.360853910208025e-06,
+      "loss": 0.8256,
+      "step": 1372
+    },
+    {
+      "epoch": 0.1374,
+      "learning_rate": 5.367039648801377e-06,
+      "loss": 0.0733,
+      "step": 1374
+    },
+    {
+      "epoch": 0.1376,
+      "learning_rate": 5.373227645443324e-06,
+      "loss": 0.8352,
+      "step": 1376
+    },
+    {
+      "epoch": 0.1378,
+      "learning_rate": 5.379417897117909e-06,
+      "loss": 0.8463,
+      "step": 1378
+    },
+    {
+      "epoch": 0.138,
+      "learning_rate": 5.385610400808079e-06,
+      "loss": 0.8759,
+      "step": 1380
+    },
+    {
+      "epoch": 0.1382,
+      "learning_rate": 5.391805153495684e-06,
+      "loss": 0.6101,
+      "step": 1382
+    },
+    {
+      "epoch": 0.1384,
+      "learning_rate": 5.398002152161492e-06,
+      "loss": 0.1938,
+      "step": 1384
+    },
+    {
+      "epoch": 0.1386,
+      "learning_rate": 5.404201393785113e-06,
+      "loss": 0.0218,
+      "step": 1386
+    },
+    {
+      "epoch": 0.1388,
+      "learning_rate": 5.410402875345162e-06,
+      "loss": 0.5709,
+      "step": 1388
+    },
+    {
+      "epoch": 0.139,
+      "learning_rate": 5.416606593819109e-06,
+      "loss": 1.2985,
+      "step": 1390
+    },
+    {
+      "epoch": 0.1392,
+      "learning_rate": 5.4228125461832915e-06,
+      "loss": 0.4577,
+      "step": 1392
+    },
+    {
+      "epoch": 0.1394,
+      "learning_rate": 5.429020729413049e-06,
+      "loss": 1.124,
+      "step": 1394
+    },
+    {
+      "epoch": 0.1396,
+      "learning_rate": 5.435231140482575e-06,
+      "loss": 1.0066,
+      "step": 1396
+    },
+    {
+      "epoch": 0.1398,
+      "learning_rate": 5.441443776365005e-06,
+      "loss": 0.1512,
+      "step": 1398
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 5.4476586340323304e-06,
+      "loss": 0.0211,
+      "step": 1400
+    },
+    {
+      "epoch": 0.1402,
+      "learning_rate": 5.453875710455549e-06,
+      "loss": 0.2193,
+      "step": 1402
+    },
+    {
+      "epoch": 0.1404,
+      "learning_rate": 5.4600950026045375e-06,
+      "loss": 0.8811,
+      "step": 1404
+    },
+    {
+      "epoch": 0.1406,
+      "learning_rate": 5.466316507448053e-06,
+      "loss": 0.2323,
+      "step": 1406
+    },
+    {
+      "epoch": 0.1408,
+      "learning_rate": 5.4725402219538106e-06,
+      "loss": 0.404,
+      "step": 1408
+    },
+    {
+      "epoch": 0.141,
+      "learning_rate": 5.478766143088497e-06,
+      "loss": 0.7116,
+      "step": 1410
+    },
+    {
+      "epoch": 0.1412,
+      "learning_rate": 5.484994267817623e-06,
+      "loss": 0.9061,
+      "step": 1412
+    },
+    {
+      "epoch": 0.1414,
+      "learning_rate": 5.49122459310568e-06,
+      "loss": 0.5825,
+      "step": 1414
+    },
+    {
+      "epoch": 0.1416,
+      "learning_rate": 5.497457115916127e-06,
+      "loss": 0.4318,
+      "step": 1416
+    },
+    {
+      "epoch": 0.1418,
+      "learning_rate": 5.503691833211264e-06,
+      "loss": 0.8841,
+      "step": 1418
+    },
+    {
+      "epoch": 0.142,
+      "learning_rate": 5.509928741952379e-06,
+      "loss": 0.9672,
+      "step": 1420
+    },
+    {
+      "epoch": 0.1422,
+      "learning_rate": 5.516167839099662e-06,
+      "loss": 0.6836,
+      "step": 1422
+    },
+    {
+      "epoch": 0.1424,
+      "learning_rate": 5.522409121612304e-06,
+      "loss": 1.1271,
+      "step": 1424
+    },
+    {
+      "epoch": 0.1426,
+      "learning_rate": 5.5286525864483285e-06,
+      "loss": 0.4845,
+      "step": 1426
+    },
+    {
+      "epoch": 0.1428,
+      "learning_rate": 5.534898230564749e-06,
+      "loss": 0.8096,
+      "step": 1428
+    },
+    {
+      "epoch": 0.143,
+      "learning_rate": 5.5411460509175605e-06,
+      "loss": 1.2247,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1432,
+      "learning_rate": 5.547396044461603e-06,
+      "loss": 0.4647,
+      "step": 1432
+    },
+    {
+      "epoch": 0.1434,
+      "learning_rate": 5.553648208150724e-06,
+      "loss": 0.5394,
+      "step": 1434
+    },
+    {
+      "epoch": 0.1436,
+      "learning_rate": 5.559902538937694e-06,
+      "loss": 1.1221,
+      "step": 1436
+    },
+    {
+      "epoch": 0.1438,
+      "learning_rate": 5.5661590337742255e-06,
+      "loss": 0.098,
+      "step": 1438
+    },
+    {
+      "epoch": 0.144,
+      "learning_rate": 5.572417689610983e-06,
+      "loss": 0.949,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1442,
+      "learning_rate": 5.57867850339757e-06,
+      "loss": 0.1312,
+      "step": 1442
+    },
+    {
+      "epoch": 0.1444,
+      "learning_rate": 5.584941472082546e-06,
+      "loss": 0.7753,
+      "step": 1444
+    },
+    {
+      "epoch": 0.1446,
+      "learning_rate": 5.591206592613412e-06,
+      "loss": 0.909,
+      "step": 1446
+    },
+    {
+      "epoch": 0.1448,
+      "learning_rate": 5.5974738619366255e-06,
+      "loss": 0.6128,
+      "step": 1448
+    },
+    {
+      "epoch": 0.145,
+      "learning_rate": 5.603743276997597e-06,
+      "loss": 0.5822,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1452,
+      "learning_rate": 5.610014834740689e-06,
+      "loss": 0.2957,
+      "step": 1452
+    },
+    {
+      "epoch": 0.1454,
+      "learning_rate": 5.616288532109221e-06,
+      "loss": 0.5766,
+      "step": 1454
+    },
+    {
+      "epoch": 0.1456,
+      "learning_rate": 5.622564366045468e-06,
+      "loss": 0.2261,
+      "step": 1456
+    },
+    {
+      "epoch": 0.1458,
+      "learning_rate": 5.628842333490665e-06,
+      "loss": 0.7677,
+      "step": 1458
+    },
+    {
+      "epoch": 0.146,
+      "learning_rate": 5.635122431385007e-06,
+      "loss": 1.1474,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1462,
+      "learning_rate": 5.641404656667652e-06,
+      "loss": 0.8027,
+      "step": 1462
+    },
+    {
+      "epoch": 0.1464,
+      "learning_rate": 5.6476890062767175e-06,
+      "loss": 0.4367,
+      "step": 1464
+    },
+    {
+      "epoch": 0.1466,
+      "learning_rate": 5.653975477149289e-06,
+      "loss": 0.6891,
+      "step": 1466
+    },
+    {
+      "epoch": 0.1468,
+      "learning_rate": 5.660264066221418e-06,
+      "loss": 0.3153,
+      "step": 1468
+    },
+    {
+      "epoch": 0.147,
+      "learning_rate": 5.666554770428136e-06,
+      "loss": 1.38,
+      "step": 1470
+    },
+    {
+      "epoch": 0.1472,
+      "learning_rate": 5.672847586703385e-06,
+      "loss": 0.0496,
+      "step": 1472
+    },
+    {
+      "epoch": 0.1474,
+      "learning_rate": 5.679142511980168e-06,
+      "loss": 0.7877,
+      "step": 1474
+    },
+    {
+      "epoch": 0.1476,
+      "learning_rate": 5.685439543190416e-06,
+      "loss": 0.372,
+      "step": 1476
+    },
+    {
+      "epoch": 0.1478,
+      "learning_rate": 5.6917386772650015e-06,
+      "loss": 1.3439,
+      "step": 1478
+    },
+    {
+      "epoch": 0.148,
+      "learning_rate": 5.698039911133803e-06,
+      "loss": 0.7369,
+      "step": 1480
+    },
+    {
+      "epoch": 0.1482,
+      "learning_rate": 5.7043432417257076e-06,
+      "loss": 0.0278,
+      "step": 1482
+    },
+    {
+      "epoch": 0.1484,
+      "learning_rate": 5.710648665968546e-06,
+      "loss": 1.0408,
+      "step": 1484
+    },
+    {
+      "epoch": 0.1486,
+      "learning_rate": 5.716956180789086e-06,
+      "loss": 0.8032,
+      "step": 1486
+    },
+    {
+      "epoch": 0.1488,
+      "learning_rate": 5.723265783113168e-06,
+      "loss": 0.0741,
+      "step": 1488
+    },
+    {
+      "epoch": 0.149,
+      "learning_rate": 5.729577469865569e-06,
+      "loss": 0.1515,
+      "step": 1490
+    },
+    {
+      "epoch": 0.1492,
+      "learning_rate": 5.735891237970019e-06,
+      "loss": 0.3621,
+      "step": 1492
+    },
+    {
+      "epoch": 0.1494,
+      "learning_rate": 5.74220708434926e-06,
+      "loss": 0.0334,
+      "step": 1494
+    },
+    {
+      "epoch": 0.1496,
+      "learning_rate": 5.748525005925079e-06,
+      "loss": 1.2072,
+      "step": 1496
+    },
+    {
+      "epoch": 0.1498,
+      "learning_rate": 5.754844999618143e-06,
+      "loss": 0.3968,
+      "step": 1498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 5.76116706234817e-06,
+      "loss": 0.4237,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1502,
+      "learning_rate": 5.767491191033909e-06,
+      "loss": 0.9232,
+      "step": 1502
+    },
+    {
+      "epoch": 0.1504,
+      "learning_rate": 5.773817382593006e-06,
+      "loss": 0.6997,
+      "step": 1504
+    },
+    {
+      "epoch": 0.1506,
+      "learning_rate": 5.780145633942173e-06,
+      "loss": 0.8909,
+      "step": 1506
+    },
+    {
+      "epoch": 0.1508,
+      "learning_rate": 5.786475941997094e-06,
+      "loss": 1.1049,
+      "step": 1508
+    },
+    {
+      "epoch": 0.151,
+      "learning_rate": 5.7928083036724535e-06,
+      "loss": 0.5563,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1512,
+      "learning_rate": 5.799142715881937e-06,
+      "loss": 0.5835,
+      "step": 1512
+    },
+    {
+      "epoch": 0.1514,
+      "learning_rate": 5.8054791755382125e-06,
+      "loss": 0.6132,
+      "step": 1514
+    },
+    {
+      "epoch": 0.1516,
+      "learning_rate": 5.811817679553014e-06,
+      "loss": 0.9181,
+      "step": 1516
+    },
+    {
+      "epoch": 0.1518,
+      "learning_rate": 5.818158224836983e-06,
+      "loss": 1.0884,
+      "step": 1518
+    },
+    {
+      "epoch": 0.152,
+      "learning_rate": 5.824500808299832e-06,
+      "loss": 0.96,
+      "step": 1520
+    },
+    {
+      "epoch": 0.1522,
+      "learning_rate": 5.830845426850263e-06,
+      "loss": 0.5264,
+      "step": 1522
+    },
+    {
+      "epoch": 0.1524,
+      "learning_rate": 5.837192077395986e-06,
+      "loss": 0.14,
+      "step": 1524
+    },
+    {
+      "epoch": 0.1526,
+      "learning_rate": 5.8435407568437194e-06,
+      "loss": 1.1115,
+      "step": 1526
+    },
+    {
+      "epoch": 0.1528,
+      "learning_rate": 5.849891462099195e-06,
+      "loss": 0.182,
+      "step": 1528
+    },
+    {
+      "epoch": 0.153,
+      "learning_rate": 5.856244190067155e-06,
+      "loss": 0.035,
+      "step": 1530
+    },
+    {
+      "epoch": 0.1532,
+      "learning_rate": 5.8625989376513584e-06,
+      "loss": 0.6246,
+      "step": 1532
+    },
+    {
+      "epoch": 0.1534,
+      "learning_rate": 5.868955701754577e-06,
+      "loss": 0.4644,
+      "step": 1534
+    },
+    {
+      "epoch": 0.1536,
+      "learning_rate": 5.875314479278604e-06,
+      "loss": 0.9221,
+      "step": 1536
+    },
+    {
+      "epoch": 0.1538,
+      "learning_rate": 5.881675267124245e-06,
+      "loss": 0.4775,
+      "step": 1538
+    },
+    {
+      "epoch": 0.154,
+      "learning_rate": 5.888038062191332e-06,
+      "loss": 1.2459,
+      "step": 1540
+    },
+    {
+      "epoch": 0.1542,
+      "learning_rate": 5.894402861378714e-06,
+      "loss": 0.1769,
+      "step": 1542
+    },
+    {
+      "epoch": 0.1544,
+      "learning_rate": 5.900769661584266e-06,
+      "loss": 0.9127,
+      "step": 1544
+    },
+    {
+      "epoch": 0.1546,
+      "learning_rate": 5.907138459704886e-06,
+      "loss": 0.5619,
+      "step": 1546
+    },
+    {
+      "epoch": 0.1548,
+      "learning_rate": 5.913509252636502e-06,
+      "loss": 0.8335,
+      "step": 1548
+    },
+    {
+      "epoch": 0.155,
+      "learning_rate": 5.919882037274065e-06,
+      "loss": 1.0706,
+      "step": 1550
+    },
+    {
+      "epoch": 0.1552,
+      "learning_rate": 5.9262568105115555e-06,
+      "loss": 0.5141,
+      "step": 1552
+    },
+    {
+      "epoch": 0.1554,
+      "learning_rate": 5.932633569241989e-06,
+      "loss": 1.0706,
+      "step": 1554
+    },
+    {
+      "epoch": 0.1556,
+      "learning_rate": 5.939012310357426e-06,
+      "loss": 0.0516,
+      "step": 1556
+    },
+    {
+      "epoch": 0.1558,
+      "learning_rate": 5.9453930307488985e-06,
+      "loss": 0.0881,
+      "step": 1558
+    },
+    {
+      "epoch": 0.156,
+      "learning_rate": 5.951775727306566e-06,
+      "loss": 0.2888,
+      "step": 1560
+    },
+    {
+      "epoch": 0.1562,
+      "learning_rate": 5.958160396919584e-06,
+      "loss": 0.7471,
+      "step": 1562
+    },
+    {
+      "epoch": 0.1564,
+      "learning_rate": 5.964547036476105e-06,
+      "loss": 0.1904,
+      "step": 1564
+    },
+    {
+      "epoch": 0.1566,
+      "learning_rate": 5.970935642863362e-06,
+      "loss": 0.4448,
+      "step": 1566
+    },
+    {
+      "epoch": 0.1568,
+      "learning_rate": 5.9773262129676735e-06,
+      "loss": 0.0597,
+      "step": 1568
+    },
+    {
+      "epoch": 0.157,
+      "learning_rate": 5.983718743674305e-06,
+      "loss": 0.7885,
+      "step": 1570
+    },
+    {
+      "epoch": 0.1572,
+      "learning_rate": 5.990113231867614e-06,
+      "loss": 0.6787,
+      "step": 1572
+    },
+    {
+      "epoch": 0.1574,
+      "learning_rate": 5.996509674431038e-06,
+      "loss": 0.7433,
+      "step": 1574
+    },
+    {
+      "epoch": 0.1576,
+      "learning_rate": 6.002908068247027e-06,
+      "loss": 0.6793,
+      "step": 1576
+    },
+    {
+      "epoch": 0.1578,
+      "learning_rate": 6.00930841019705e-06,
+      "loss": 0.689,
+      "step": 1578
+    },
+    {
+      "epoch": 0.158,
+      "learning_rate": 6.015710697161659e-06,
+      "loss": 0.641,
+      "step": 1580
+    },
+    {
+      "epoch": 0.1582,
+      "learning_rate": 6.022114926020505e-06,
+      "loss": 0.7696,
+      "step": 1582
+    },
+    {
+      "epoch": 0.1584,
+      "learning_rate": 6.028521093652195e-06,
+      "loss": 0.0225,
+      "step": 1584
+    },
+    {
+      "epoch": 0.1586,
+      "learning_rate": 6.0349291969344426e-06,
+      "loss": 0.5692,
+      "step": 1586
+    },
+    {
+      "epoch": 0.1588,
+      "learning_rate": 6.0413392327440635e-06,
+      "loss": 0.3909,
+      "step": 1588
+    },
+    {
+      "epoch": 0.159,
+      "learning_rate": 6.047751197956836e-06,
+      "loss": 0.2661,
+      "step": 1590
+    },
+    {
+      "epoch": 0.1592,
+      "learning_rate": 6.0541650894476614e-06,
+      "loss": 0.6324,
+      "step": 1592
+    },
+    {
+      "epoch": 0.1594,
+      "learning_rate": 6.060580904090489e-06,
+      "loss": 0.9539,
+      "step": 1594
+    },
+    {
+      "epoch": 0.1596,
+      "learning_rate": 6.066998638758326e-06,
+      "loss": 0.2518,
+      "step": 1596
+    },
+    {
+      "epoch": 0.1598,
+      "learning_rate": 6.0734182903232475e-06,
+      "loss": 0.1483,
+      "step": 1598
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.0798398556563955e-06,
+      "loss": 0.5771,
+      "step": 1600
+    },
+    {
+      "epoch": 0.1602,
+      "learning_rate": 6.086263331627974e-06,
+      "loss": 0.9874,
+      "step": 1602
+    },
+    {
+      "epoch": 0.1604,
+      "learning_rate": 6.09268871510726e-06,
+      "loss": 0.752,
+      "step": 1604
+    },
+    {
+      "epoch": 0.1606,
+      "learning_rate": 6.0991160029626e-06,
+      "loss": 0.9625,
+      "step": 1606
+    },
+    {
+      "epoch": 0.1608,
+      "learning_rate": 6.1055451920614114e-06,
+      "loss": 0.9635,
+      "step": 1608
+    },
+    {
+      "epoch": 0.161,
+      "learning_rate": 6.111976279270187e-06,
+      "loss": 0.353,
+      "step": 1610
+    },
+    {
+      "epoch": 0.1612,
+      "learning_rate": 6.11840926145449e-06,
+      "loss": 0.6463,
+      "step": 1612
+    },
+    {
+      "epoch": 0.1614,
+      "learning_rate": 6.124844135478966e-06,
+      "loss": 0.4619,
+      "step": 1614
+    },
+    {
+      "epoch": 0.1616,
+      "learning_rate": 6.131280898207335e-06,
+      "loss": 0.7865,
+      "step": 1616
+    },
+    {
+      "epoch": 0.1618,
+      "learning_rate": 6.137719546502394e-06,
+      "loss": 0.5555,
+      "step": 1618
+    },
+    {
+      "epoch": 0.162,
+      "learning_rate": 6.144160077226029e-06,
+      "loss": 0.979,
+      "step": 1620
+    },
+    {
+      "epoch": 0.1622,
+      "learning_rate": 6.1506024872392e-06,
+      "loss": 0.314,
+      "step": 1622
+    },
+    {
+      "epoch": 0.1624,
+      "learning_rate": 6.157046773401956e-06,
+      "loss": 0.0029,
+      "step": 1624
+    },
+    {
+      "epoch": 0.1626,
+      "learning_rate": 6.163492932573429e-06,
+      "loss": 0.0331,
+      "step": 1626
+    },
+    {
+      "epoch": 0.1628,
+      "learning_rate": 6.169940961611859e-06,
+      "loss": 0.4195,
+      "step": 1628
+    },
+    {
+      "epoch": 0.163,
+      "learning_rate": 6.176390857374501e-06,
+      "loss": 0.1998,
+      "step": 1630
+    },
+    {
+      "epoch": 0.1632,
+      "learning_rate": 6.182842616717808e-06,
+      "loss": 0.6594,
+      "step": 1632
+    },
+    {
+      "epoch": 0.1634,
+      "learning_rate": 6.189296236497251e-06,
+      "loss": 0.9515,
+      "step": 1634
+    },
+    {
+      "epoch": 0.1636,
+      "learning_rate": 6.195751713567431e-06,
+      "loss": 0.7136,
+      "step": 1636
+    },
+    {
+      "epoch": 0.1638,
+      "learning_rate": 6.202209044781979e-06,
+      "loss": 0.5117,
+      "step": 1638
+    },
+    {
+      "epoch": 0.164,
+      "learning_rate": 6.20866822699372e-06,
+      "loss": 0.6183,
+      "step": 1640
+    },
+    {
+      "epoch": 0.1642,
+      "learning_rate": 6.215129257054525e-06,
+      "loss": 0.4295,
+      "step": 1642
+    },
+    {
+      "epoch": 0.1644,
+      "learning_rate": 6.2215921318153186e-06,
+      "loss": 0.6207,
+      "step": 1644
+    },
+    {
+      "epoch": 0.1646,
+      "learning_rate": 6.228056848126223e-06,
+      "loss": 0.5284,
+      "step": 1646
+    },
+    {
+      "epoch": 0.1648,
+      "learning_rate": 6.234523402836413e-06,
+      "loss": 0.454,
+      "step": 1648
+    },
+    {
+      "epoch": 0.165,
+      "learning_rate": 6.240991792794137e-06,
+      "loss": 1.1207,
+      "step": 1650
+    },
+    {
+      "epoch": 0.1652,
+      "learning_rate": 6.24746201484678e-06,
+      "loss": 0.4678,
+      "step": 1652
+    },
+    {
+      "epoch": 0.1654,
+      "learning_rate": 6.253934065840883e-06,
+      "loss": 0.7602,
+      "step": 1654
+    },
+    {
+      "epoch": 0.1656,
+      "learning_rate": 6.260407942622e-06,
+      "loss": 0.53,
+      "step": 1656
+    },
+    {
+      "epoch": 0.1658,
+      "learning_rate": 6.2668836420348374e-06,
+      "loss": 0.0249,
+      "step": 1658
+    },
+    {
+      "epoch": 0.166,
+      "learning_rate": 6.273361160923256e-06,
+      "loss": 0.9495,
+      "step": 1660
+    },
+    {
+      "epoch": 0.1662,
+      "learning_rate": 6.279840496130188e-06,
+      "loss": 0.5108,
+      "step": 1662
+    },
+    {
+      "epoch": 0.1664,
+      "learning_rate": 6.2863216444976535e-06,
+      "loss": 0.7681,
+      "step": 1664
+    },
+    {
+      "epoch": 0.1666,
+      "learning_rate": 6.2928046028668185e-06,
+      "loss": 0.8013,
+      "step": 1666
+    },
+    {
+      "epoch": 0.1668,
+      "learning_rate": 6.299289368078019e-06,
+      "loss": 0.0963,
+      "step": 1668
+    },
+    {
+      "epoch": 0.167,
+      "learning_rate": 6.305775936970606e-06,
+      "loss": 0.0457,
+      "step": 1670
+    },
+    {
+      "epoch": 0.1672,
+      "learning_rate": 6.3122643063831245e-06,
+      "loss": 0.8038,
+      "step": 1672
+    },
+    {
+      "epoch": 0.1674,
+      "learning_rate": 6.3187544731532205e-06,
+      "loss": 0.1024,
+      "step": 1674
+    },
+    {
+      "epoch": 0.1676,
+      "learning_rate": 6.325246434117667e-06,
+      "loss": 0.0919,
+      "step": 1676
+    },
+    {
+      "epoch": 0.1678,
+      "learning_rate": 6.331740186112359e-06,
+      "loss": 0.1816,
+      "step": 1678
+    },
+    {
+      "epoch": 0.168,
+      "learning_rate": 6.3382357259723214e-06,
+      "loss": 0.5851,
+      "step": 1680
+    },
+    {
+      "epoch": 0.1682,
+      "learning_rate": 6.344733050531709e-06,
+      "loss": 0.3207,
+      "step": 1682
+    },
+    {
+      "epoch": 0.1684,
+      "learning_rate": 6.351232156623801e-06,
+      "loss": 0.6342,
+      "step": 1684
+    },
+    {
+      "epoch": 0.1686,
+      "learning_rate": 6.357733041081015e-06,
+      "loss": 0.7796,
+      "step": 1686
+    },
+    {
+      "epoch": 0.1688,
+      "learning_rate": 6.364235700734899e-06,
+      "loss": 0.7628,
+      "step": 1688
+    },
+    {
+      "epoch": 0.169,
+      "learning_rate": 6.370740132416133e-06,
+      "loss": 0.6815,
+      "step": 1690
+    },
+    {
+      "epoch": 0.1692,
+      "learning_rate": 6.377246332954539e-06,
+      "loss": 0.6747,
+      "step": 1692
+    },
+    {
+      "epoch": 0.1694,
+      "learning_rate": 6.383754299179072e-06,
+      "loss": 0.9745,
+      "step": 1694
+    },
+    {
+      "epoch": 0.1696,
+      "learning_rate": 6.390264027917829e-06,
+      "loss": 0.5274,
+      "step": 1696
+    },
+    {
+      "epoch": 0.1698,
+      "learning_rate": 6.3967755159980485e-06,
+      "loss": 0.0465,
+      "step": 1698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 6.403288760246108e-06,
+      "loss": 0.0024,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1702,
+      "learning_rate": 6.409803757487532e-06,
+      "loss": 0.5663,
+      "step": 1702
+    },
+    {
+      "epoch": 0.1704,
+      "learning_rate": 6.4163205045469916e-06,
+      "loss": 1.1925,
+      "step": 1704
+    },
+    {
+      "epoch": 0.1706,
+      "learning_rate": 6.422838998248301e-06,
+      "loss": 0.4389,
+      "step": 1706
+    },
+    {
+      "epoch": 0.1708,
+      "learning_rate": 6.429359235414444e-06,
+      "loss": 0.2898,
+      "step": 1708
+    },
+    {
+      "epoch": 0.171,
+      "learning_rate": 6.435881212867485e-06,
+      "loss": 0.4384,
+      "step": 1710
+    },
+    {
+      "epoch": 0.1712,
+      "learning_rate": 6.442404927428743e-06,
+      "loss": 1.2891,
+      "step": 1712
+    },
+    {
+      "epoch": 0.1714,
+      "learning_rate": 6.4489303759186385e-06,
+      "loss": 0.0072,
+      "step": 1714
+    },
+    {
+      "epoch": 0.1716,
+      "learning_rate": 6.455457555156695e-06,
+      "loss": 0.5563,
+      "step": 1716
+    },
+    {
+      "epoch": 0.1718,
+      "learning_rate": 6.4619864619616975e-06,
+      "loss": 0.2891,
+      "step": 1718
+    },
+    {
+      "epoch": 0.172,
+      "learning_rate": 6.468517093151517e-06,
+      "loss": 0.5899,
+      "step": 1720
+    },
+    {
+      "epoch": 0.1722,
+      "learning_rate": 6.475049445543222e-06,
+      "loss": 0.4164,
+      "step": 1722
+    },
+    {
+      "epoch": 0.1724,
+      "learning_rate": 6.481583515952972e-06,
+      "loss": 0.4955,
+      "step": 1724
+    },
+    {
+      "epoch": 0.1726,
+      "learning_rate": 6.48811930119619e-06,
+      "loss": 0.4041,
+      "step": 1726
+    },
+    {
+      "epoch": 0.1728,
+      "learning_rate": 6.4946567980874155e-06,
+      "loss": 1.137,
+      "step": 1728
+    },
+    {
+      "epoch": 0.173,
+      "learning_rate": 6.5011960034403e-06,
+      "loss": 0.4897,
+      "step": 1730
+    },
+    {
+      "epoch": 0.1732,
+      "learning_rate": 6.507736914067766e-06,
+      "loss": 0.7879,
+      "step": 1732
+    },
+    {
+      "epoch": 0.1734,
+      "learning_rate": 6.514279526781853e-06,
+      "loss": 0.8214,
+      "step": 1734
+    },
+    {
+      "epoch": 0.1736,
+      "learning_rate": 6.520823838393735e-06,
+      "loss": 0.5218,
+      "step": 1736
+    },
+    {
+      "epoch": 0.1738,
+      "learning_rate": 6.5273698457137965e-06,
+      "loss": 0.5681,
+      "step": 1738
+    },
+    {
+      "epoch": 0.174,
+      "learning_rate": 6.533917545551644e-06,
+      "loss": 0.4953,
+      "step": 1740
+    },
+    {
+      "epoch": 0.1742,
+      "learning_rate": 6.540466934715955e-06,
+      "loss": 0.456,
+      "step": 1742
+    },
+    {
+      "epoch": 0.1744,
+      "learning_rate": 6.547018010014656e-06,
+      "loss": 0.6907,
+      "step": 1744
+    },
+    {
+      "epoch": 0.1746,
+      "learning_rate": 6.553570768254831e-06,
+      "loss": 0.8607,
+      "step": 1746
+    },
+    {
+      "epoch": 0.1748,
+      "learning_rate": 6.560125206242746e-06,
+      "loss": 1.0177,
+      "step": 1748
+    },
+    {
+      "epoch": 0.175,
+      "learning_rate": 6.566681320783848e-06,
+      "loss": 1.0527,
+      "step": 1750
+    },
+    {
+      "epoch": 0.1752,
+      "learning_rate": 6.573239108682749e-06,
+      "loss": 0.4315,
+      "step": 1752
+    },
+    {
+      "epoch": 0.1754,
+      "learning_rate": 6.579798566743313e-06,
+      "loss": 0.6811,
+      "step": 1754
+    },
+    {
+      "epoch": 0.1756,
+      "learning_rate": 6.586359691768489e-06,
+      "loss": 0.2933,
+      "step": 1756
+    },
+    {
+      "epoch": 0.1758,
+      "learning_rate": 6.592922480560483e-06,
+      "loss": 0.8273,
+      "step": 1758
+    },
+    {
+      "epoch": 0.176,
+      "learning_rate": 6.59948692992067e-06,
+      "loss": 0.0459,
+      "step": 1760
+    },
+    {
+      "epoch": 0.1762,
+      "learning_rate": 6.606053036649618e-06,
+      "loss": 0.8313,
+      "step": 1762
+    },
+    {
+      "epoch": 0.1764,
+      "learning_rate": 6.612620797547085e-06,
+      "loss": 0.4796,
+      "step": 1764
+    },
+    {
+      "epoch": 0.1766,
+      "learning_rate": 6.619190209412025e-06,
+      "loss": 0.334,
+      "step": 1766
+    },
+    {
+      "epoch": 0.1768,
+      "learning_rate": 6.625761269042586e-06,
+      "loss": 0.5898,
+      "step": 1768
+    },
+    {
+      "epoch": 0.177,
+      "learning_rate": 6.632333973236113e-06,
+      "loss": 0.5792,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1772,
+      "learning_rate": 6.638908318789153e-06,
+      "loss": 0.4086,
+      "step": 1772
+    },
+    {
+      "epoch": 0.1774,
+      "learning_rate": 6.6454843024974465e-06,
+      "loss": 1.1671,
+      "step": 1774
+    },
+    {
+      "epoch": 0.1776,
+      "learning_rate": 6.652061921155939e-06,
+      "loss": 0.9475,
+      "step": 1776
+    },
+    {
+      "epoch": 0.1778,
+      "learning_rate": 6.6586411715587805e-06,
+      "loss": 0.721,
+      "step": 1778
+    },
+    {
+      "epoch": 0.178,
+      "learning_rate": 6.665222050499324e-06,
+      "loss": 0.9761,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1782,
+      "learning_rate": 6.671804554770128e-06,
+      "loss": 0.6245,
+      "step": 1782
+    },
+    {
+      "epoch": 0.1784,
+      "learning_rate": 6.678388681162962e-06,
+      "loss": 0.4354,
+      "step": 1784
+    },
+    {
+      "epoch": 0.1786,
+      "learning_rate": 6.6849744264688e-06,
+      "loss": 0.6417,
+      "step": 1786
+    },
+    {
+      "epoch": 0.1788,
+      "learning_rate": 6.691561787477832e-06,
+      "loss": 0.4533,
+      "step": 1788
+    },
+    {
+      "epoch": 0.179,
+      "learning_rate": 6.698150760979456e-06,
+      "loss": 0.4025,
+      "step": 1790
+    },
+    {
+      "epoch": 0.1792,
+      "learning_rate": 6.704741343762289e-06,
+      "loss": 1.0558,
+      "step": 1792
+    },
+    {
+      "epoch": 0.1794,
+      "learning_rate": 6.711333532614177e-06,
+      "loss": 0.6159,
+      "step": 1794
+    },
+    {
+      "epoch": 0.1796,
+      "learning_rate": 6.717927324322116e-06,
+      "loss": 0.4105,
+      "step": 1796
+    },
+    {
+      "epoch": 0.1798,
+      "learning_rate": 6.724522715672421e-06,
+      "loss": 0.8161,
+      "step": 1798
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 6.731119703450584e-06,
+      "loss": 0.4918,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1802,
+      "learning_rate": 6.737718284441256e-06,
+      "loss": 0.729,
+      "step": 1802
+    },
+    {
+      "epoch": 0.1804,
+      "learning_rate": 6.744318455428424e-06,
+      "loss": 1.17,
+      "step": 1804
+    },
+    {
+      "epoch": 0.1806,
+      "learning_rate": 6.750920213195242e-06,
+      "loss": 0.397,
+      "step": 1806
+    },
+    {
+      "epoch": 0.1808,
+      "learning_rate": 6.757523554524062e-06,
+      "loss": 1.1147,
+      "step": 1808
+    },
+    {
+      "epoch": 0.181,
+      "learning_rate": 6.764128476196494e-06,
+      "loss": 0.0459,
+      "step": 1810
+    },
+    {
+      "epoch": 0.1812,
+      "learning_rate": 6.770734974993414e-06,
+      "loss": 0.4741,
+      "step": 1812
+    },
+    {
+      "epoch": 0.1814,
+      "learning_rate": 6.777343047694894e-06,
+      "loss": 0.8585,
+      "step": 1814
+    },
+    {
+      "epoch": 0.1816,
+      "learning_rate": 6.78395269108019e-06,
+      "loss": 0.5914,
+      "step": 1816
+    },
+    {
+      "epoch": 0.1818,
+      "learning_rate": 6.7905639019278925e-06,
+      "loss": 0.7222,
+      "step": 1818
+    },
+    {
+      "epoch": 0.182,
+      "learning_rate": 6.797176677015778e-06,
+      "loss": 0.2016,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1822,
+      "learning_rate": 6.803791013120824e-06,
+      "loss": 0.1404,
+      "step": 1822
+    },
+    {
+      "epoch": 0.1824,
+      "learning_rate": 6.810406907019287e-06,
+      "loss": 0.2666,
+      "step": 1824
+    },
+    {
+      "epoch": 0.1826,
+      "learning_rate": 6.817024355486707e-06,
+      "loss": 0.4165,
+      "step": 1826
+    },
+    {
+      "epoch": 0.1828,
+      "learning_rate": 6.823643355297775e-06,
+      "loss": 0.6512,
+      "step": 1828
+    },
+    {
+      "epoch": 0.183,
+      "learning_rate": 6.8302639032264836e-06,
+      "loss": 0.4092,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1832,
+      "learning_rate": 6.836885996046061e-06,
+      "loss": 1.4267,
+      "step": 1832
+    },
+    {
+      "epoch": 0.1834,
+      "learning_rate": 6.8435096305289765e-06,
+      "loss": 0.5425,
+      "step": 1834
+    },
+    {
+      "epoch": 0.1836,
+      "learning_rate": 6.850134803446953e-06,
+      "loss": 0.6099,
+      "step": 1836
+    },
+    {
+      "epoch": 0.1838,
+      "learning_rate": 6.856761511570944e-06,
+      "loss": 0.6668,
+      "step": 1838
+    },
+    {
+      "epoch": 0.184,
+      "learning_rate": 6.863389751671224e-06,
+      "loss": 0.653,
+      "step": 1840
+    },
+    {
+      "epoch": 0.1842,
+      "learning_rate": 6.870019520517217e-06,
+      "loss": 1.2419,
+      "step": 1842
+    },
+    {
+      "epoch": 0.1844,
+      "learning_rate": 6.876650814877674e-06,
+      "loss": 0.8471,
+      "step": 1844
+    },
+    {
+      "epoch": 0.1846,
+      "learning_rate": 6.883283631520579e-06,
+      "loss": 0.5134,
+      "step": 1846
+    },
+    {
+      "epoch": 0.1848,
+      "learning_rate": 6.889917967213181e-06,
+      "loss": 0.5058,
+      "step": 1848
+    },
+    {
+      "epoch": 0.185,
+      "learning_rate": 6.896553818721985e-06,
+      "loss": 0.0265,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1852,
+      "learning_rate": 6.903191182812754e-06,
+      "loss": 1.126,
+      "step": 1852
+    },
+    {
+      "epoch": 0.1854,
+      "learning_rate": 6.909830056250522e-06,
+      "loss": 0.1193,
+      "step": 1854
+    },
+    {
+      "epoch": 0.1856,
+      "learning_rate": 6.916470435799581e-06,
+      "loss": 0.0843,
+      "step": 1856
+    },
+    {
+      "epoch": 0.1858,
+      "learning_rate": 6.9231123182234895e-06,
+      "loss": 0.5982,
+      "step": 1858
+    },
+    {
+      "epoch": 0.186,
+      "learning_rate": 6.9297557002850765e-06,
+      "loss": 0.3603,
+      "step": 1860
+    },
+    {
+      "epoch": 0.1862,
+      "learning_rate": 6.936400578746436e-06,
+      "loss": 0.4718,
+      "step": 1862
+    },
+    {
+      "epoch": 0.1864,
+      "learning_rate": 6.943046950368938e-06,
+      "loss": 0.9386,
+      "step": 1864
+    },
+    {
+      "epoch": 0.1866,
+      "learning_rate": 6.949694811913237e-06,
+      "loss": 1.1185,
+      "step": 1866
+    },
+    {
+      "epoch": 0.1868,
+      "learning_rate": 6.9563441601391945e-06,
+      "loss": 0.5484,
+      "step": 1868
+    },
+    {
+      "epoch": 0.187,
+      "learning_rate": 6.96299499180605e-06,
+      "loss": 0.5534,
+      "step": 1870
+    },
+    {
+      "epoch": 0.1872,
+      "learning_rate": 6.969647303672253e-06,
+      "loss": 0.5747,
+      "step": 1872
+    },
+    {
+      "epoch": 0.1874,
+      "learning_rate": 6.976301092495548e-06,
+      "loss": 0.4591,
+      "step": 1874
+    },
+    {
+      "epoch": 0.1876,
+      "learning_rate": 6.982956355032957e-06,
+      "loss": 0.6008,
+      "step": 1876
+    },
+    {
+      "epoch": 0.1878,
+      "learning_rate": 6.989613088040787e-06,
+      "loss": 1.0699,
+      "step": 1878
+    },
+    {
+      "epoch": 0.188,
+      "learning_rate": 6.996271288274646e-06,
+      "loss": 0.2374,
+      "step": 1880
+    },
+    {
+      "epoch": 0.1882,
+      "learning_rate": 7.002930952489353e-06,
+      "loss": 0.9571,
+      "step": 1882
+    },
+    {
+      "epoch": 0.1884,
+      "learning_rate": 7.009592077439124e-06,
+      "loss": 0.2224,
+      "step": 1884
+    },
+    {
+      "epoch": 0.1886,
+      "learning_rate": 7.016254659877404e-06,
+      "loss": 0.6512,
+      "step": 1886
+    },
+    {
+      "epoch": 0.1888,
+      "learning_rate": 7.022918696556883e-06,
+      "loss": 0.7142,
+      "step": 1888
+    },
+    {
+      "epoch": 0.189,
+      "learning_rate": 7.029584184229641e-06,
+      "loss": 0.6765,
+      "step": 1890
+    },
+    {
+      "epoch": 0.1892,
+      "learning_rate": 7.036251119646998e-06,
+      "loss": 0.8776,
+      "step": 1892
+    },
+    {
+      "epoch": 0.1894,
+      "learning_rate": 7.042919499559539e-06,
+      "loss": 0.5,
+      "step": 1894
+    },
+    {
+      "epoch": 0.1896,
+      "learning_rate": 7.049589320717175e-06,
+      "loss": 0.3666,
+      "step": 1896
+    },
+    {
+      "epoch": 0.1898,
+      "learning_rate": 7.056260579869152e-06,
+      "loss": 1.1619,
+      "step": 1898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.062933273763978e-06,
+      "loss": 0.4314,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1902,
+      "learning_rate": 7.06960739914943e-06,
+      "loss": 1.2965,
+      "step": 1902
+    },
+    {
+      "epoch": 0.1904,
+      "learning_rate": 7.07628295277262e-06,
+      "loss": 0.7751,
+      "step": 1904
+    },
+    {
+      "epoch": 0.1906,
+      "learning_rate": 7.082959931380013e-06,
+      "loss": 0.2593,
+      "step": 1906
+    },
+    {
+      "epoch": 0.1908,
+      "learning_rate": 7.089638331717284e-06,
+      "loss": 0.5879,
+      "step": 1908
+    },
+    {
+      "epoch": 0.191,
+      "learning_rate": 7.09631815052946e-06,
+      "loss": 0.52,
+      "step": 1910
+    },
+    {
+      "epoch": 0.1912,
+      "learning_rate": 7.102999384560927e-06,
+      "loss": 0.1887,
+      "step": 1912
+    },
+    {
+      "epoch": 0.1914,
+      "learning_rate": 7.109682030555285e-06,
+      "loss": 0.197,
+      "step": 1914
+    },
+    {
+      "epoch": 0.1916,
+      "learning_rate": 7.116366085255509e-06,
+      "loss": 0.9713,
+      "step": 1916
+    },
+    {
+      "epoch": 0.1918,
+      "learning_rate": 7.123051545403873e-06,
+      "loss": 0.8468,
+      "step": 1918
+    },
+    {
+      "epoch": 0.192,
+      "learning_rate": 7.129738407741964e-06,
+      "loss": 0.8443,
+      "step": 1920
+    },
+    {
+      "epoch": 0.1922,
+      "learning_rate": 7.136426669010686e-06,
+      "loss": 0.9938,
+      "step": 1922
+    },
+    {
+      "epoch": 0.1924,
+      "learning_rate": 7.143116325950246e-06,
+      "loss": 0.0204,
+      "step": 1924
+    },
+    {
+      "epoch": 0.1926,
+      "learning_rate": 7.1498073753002375e-06,
+      "loss": 0.0439,
+      "step": 1926
+    },
+    {
+      "epoch": 0.1928,
+      "learning_rate": 7.156499813799471e-06,
+      "loss": 0.2399,
+      "step": 1928
+    },
+    {
+      "epoch": 0.193,
+      "learning_rate": 7.1631936381861544e-06,
+      "loss": 0.8468,
+      "step": 1930
+    },
+    {
+      "epoch": 0.1932,
+      "learning_rate": 7.169888845197795e-06,
+      "loss": 0.049,
+      "step": 1932
+    },
+    {
+      "epoch": 0.1934,
+      "learning_rate": 7.1765854315712325e-06,
+      "loss": 0.6706,
+      "step": 1934
+    },
+    {
+      "epoch": 0.1936,
+      "learning_rate": 7.18328339404263e-06,
+      "loss": 0.8402,
+      "step": 1936
+    },
+    {
+      "epoch": 0.1938,
+      "learning_rate": 7.189982729347485e-06,
+      "loss": 0.3542,
+      "step": 1938
+    },
+    {
+      "epoch": 0.194,
+      "learning_rate": 7.1966834342206215e-06,
+      "loss": 0.0383,
+      "step": 1940
+    },
+    {
+      "epoch": 0.1942,
+      "learning_rate": 7.203385505396197e-06,
+      "loss": 1.3131,
+      "step": 1942
+    },
+    {
+      "epoch": 0.1944,
+      "learning_rate": 7.2100889396077025e-06,
+      "loss": 1.0287,
+      "step": 1944
+    },
+    {
+      "epoch": 0.1946,
+      "learning_rate": 7.216793733587966e-06,
+      "loss": 1.2361,
+      "step": 1946
+    },
+    {
+      "epoch": 0.1948,
+      "learning_rate": 7.223499884069155e-06,
+      "loss": 0.4656,
+      "step": 1948
+    },
+    {
+      "epoch": 0.195,
+      "learning_rate": 7.230207387782771e-06,
+      "loss": 0.5281,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1952,
+      "learning_rate": 7.236916241459674e-06,
+      "loss": 0.1729,
+      "step": 1952
+    },
+    {
+      "epoch": 0.1954,
+      "learning_rate": 7.243626441830001e-06,
+      "loss": 0.3027,
+      "step": 1954
+    },
+    {
+      "epoch": 0.1956,
+      "learning_rate": 7.250337985623332e-06,
+      "loss": 0.9592,
+      "step": 1956
+    },
+    {
+      "epoch": 0.1958,
+      "learning_rate": 7.257050869568527e-06,
+      "loss": 0.8997,
+      "step": 1958
+    },
+    {
+      "epoch": 0.196,
+      "learning_rate": 7.263765090393806e-06,
+      "loss": 0.4494,
+      "step": 1960
+    },
+    {
+      "epoch": 0.1962,
+      "learning_rate": 7.270480644826739e-06,
+      "loss": 0.3059,
+      "step": 1962
+    },
+    {
+      "epoch": 0.1964,
+      "learning_rate": 7.277197529594246e-06,
+      "loss": 0.7767,
+      "step": 1964
+    },
+    {
+      "epoch": 0.1966,
+      "learning_rate": 7.28391574142262e-06,
+      "loss": 0.6228,
+      "step": 1966
+    },
+    {
+      "epoch": 0.1968,
+      "learning_rate": 7.290635277037431e-06,
+      "loss": 0.706,
+      "step": 1968
+    },
+    {
+      "epoch": 0.197,
+      "learning_rate": 7.297356133163711e-06,
+      "loss": 0.9038,
+      "step": 1970
+    },
+    {
+      "epoch": 0.1972,
+      "learning_rate": 7.304078306525797e-06,
+      "loss": 0.3075,
+      "step": 1972
+    },
+    {
+      "epoch": 0.1974,
+      "learning_rate": 7.3108017938473485e-06,
+      "loss": 1.3941,
+      "step": 1974
+    },
+    {
+      "epoch": 0.1976,
+      "learning_rate": 7.317526591851421e-06,
+      "loss": 0.8109,
+      "step": 1976
+    },
+    {
+      "epoch": 0.1978,
+      "learning_rate": 7.324252697260479e-06,
+      "loss": 0.0903,
+      "step": 1978
+    },
+    {
+      "epoch": 0.198,
+      "learning_rate": 7.3309801067962485e-06,
+      "loss": 0.4553,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1982,
+      "learning_rate": 7.337708817179875e-06,
+      "loss": 0.5662,
+      "step": 1982
+    },
+    {
+      "epoch": 0.1984,
+      "learning_rate": 7.344438825131899e-06,
+      "loss": 0.7307,
+      "step": 1984
+    },
+    {
+      "epoch": 0.1986,
+      "learning_rate": 7.351170127372196e-06,
+      "loss": 0.2055,
+      "step": 1986
+    },
+    {
+      "epoch": 0.1988,
+      "learning_rate": 7.357902720619978e-06,
+      "loss": 1.1788,
+      "step": 1988
+    },
+    {
+      "epoch": 0.199,
+      "learning_rate": 7.36463660159386e-06,
+      "loss": 0.6543,
+      "step": 1990
+    },
+    {
+      "epoch": 0.1992,
+      "learning_rate": 7.371371767011881e-06,
+      "loss": 0.2515,
+      "step": 1992
+    },
+    {
+      "epoch": 0.1994,
+      "learning_rate": 7.378108213591355e-06,
+      "loss": 0.7394,
+      "step": 1994
+    },
+    {
+      "epoch": 0.1996,
+      "learning_rate": 7.384845938049016e-06,
+      "loss": 1.1461,
+      "step": 1996
+    },
+    {
+      "epoch": 0.1998,
+      "learning_rate": 7.39158493710103e-06,
+      "loss": 0.4205,
+      "step": 1998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 7.398325207462844e-06,
+      "loss": 0.5187,
+      "step": 2000
+    },
+    {
+      "epoch": 0.2002,
+      "learning_rate": 7.405066745849345e-06,
+      "loss": 0.36,
+      "step": 2002
+    },
+    {
+      "epoch": 0.2004,
+      "learning_rate": 7.411809548974791e-06,
+      "loss": 0.5179,
+      "step": 2004
+    },
+    {
+      "epoch": 0.2006,
+      "learning_rate": 7.418553613552822e-06,
+      "loss": 0.1826,
+      "step": 2006
+    },
+    {
+      "epoch": 0.2008,
+      "learning_rate": 7.425298936296461e-06,
+      "loss": 1.0062,
+      "step": 2008
+    },
+    {
+      "epoch": 0.201,
+      "learning_rate": 7.432045513918121e-06,
+      "loss": 1.2673,
+      "step": 2010
+    },
+    {
+      "epoch": 0.2012,
+      "learning_rate": 7.438793343129601e-06,
+      "loss": 0.932,
+      "step": 2012
+    },
+    {
+      "epoch": 0.2014,
+      "learning_rate": 7.445542420642091e-06,
+      "loss": 0.3942,
+      "step": 2014
+    },
+    {
+      "epoch": 0.2016,
+      "learning_rate": 7.452292743166175e-06,
+      "loss": 0.2742,
+      "step": 2016
+    },
+    {
+      "epoch": 0.2018,
+      "learning_rate": 7.459044307411826e-06,
+      "loss": 0.6336,
+      "step": 2018
+    },
+    {
+      "epoch": 0.202,
+      "learning_rate": 7.465797110088413e-06,
+      "loss": 0.6214,
+      "step": 2020
+    },
+    {
+      "epoch": 0.2022,
+      "learning_rate": 7.472551147904703e-06,
+      "loss": 0.9604,
+      "step": 2022
+    },
+    {
+      "epoch": 0.2024,
+      "learning_rate": 7.479306417568859e-06,
+      "loss": 0.3009,
+      "step": 2024
+    },
+    {
+      "epoch": 0.2026,
+      "learning_rate": 7.486062915788446e-06,
+      "loss": 0.4846,
+      "step": 2026
+    },
+    {
+      "epoch": 0.2028,
+      "learning_rate": 7.4928206392704285e-06,
+      "loss": 0.8351,
+      "step": 2028
+    },
+    {
+      "epoch": 0.203,
+      "learning_rate": 7.499579584721173e-06,
+      "loss": 0.1641,
+      "step": 2030
+    },
+    {
+      "epoch": 0.2032,
+      "learning_rate": 7.506339748846452e-06,
+      "loss": 0.3217,
+      "step": 2032
+    },
+    {
+      "epoch": 0.2034,
+      "learning_rate": 7.513101128351446e-06,
+      "loss": 0.089,
+      "step": 2034
+    },
+    {
+      "epoch": 0.2036,
+      "learning_rate": 7.519863719940737e-06,
+      "loss": 1.0693,
+      "step": 2036
+    },
+    {
+      "epoch": 0.2038,
+      "learning_rate": 7.5266275203183395e-06,
+      "loss": 1.004,
+      "step": 2038
+    },
+    {
+      "epoch": 0.204,
+      "learning_rate": 7.533392526187608e-06,
+      "loss": 1.3446,
+      "step": 2040
+    },
+    {
+      "epoch": 0.2042,
+      "learning_rate": 7.540158734251412e-06,
+      "loss": 0.761,
+      "step": 2042
+    },
+    {
+      "epoch": 0.2044,
+      "learning_rate": 7.546926141211982e-06,
+      "loss": 0.7984,
+      "step": 2044
+    },
+    {
+      "epoch": 0.2046,
+      "learning_rate": 7.553694743770917e-06,
+      "loss": 1.2003,
+      "step": 2046
+    },
+    {
+      "epoch": 0.2048,
+      "learning_rate": 7.560464538629334e-06,
+      "loss": 0.5543,
+      "step": 2048
+    },
+    {
+      "epoch": 0.205,
+      "learning_rate": 7.567235522487698e-06,
+      "loss": 0.6688,
+      "step": 2050
+    },
+    {
+      "epoch": 0.2052,
+      "learning_rate": 7.574007692045932e-06,
+      "loss": 0.394,
+      "step": 2052
+    },
+    {
+      "epoch": 0.2054,
+      "learning_rate": 7.580781044003312e-06,
+      "loss": 0.9886,
+      "step": 2054
+    },
+    {
+      "epoch": 0.2056,
+      "learning_rate": 7.587555575058638e-06,
+      "loss": 0.5569,
+      "step": 2056
+    },
+    {
+      "epoch": 0.2058,
+      "learning_rate": 7.5943312819100875e-06,
+      "loss": 0.7853,
+      "step": 2058
+    },
+    {
+      "epoch": 0.206,
+      "learning_rate": 7.601108161255231e-06,
+      "loss": 0.4565,
+      "step": 2060
+    },
+    {
+      "epoch": 0.2062,
+      "learning_rate": 7.607886209791095e-06,
+      "loss": 0.309,
+      "step": 2062
+    },
+    {
+      "epoch": 0.2064,
+      "learning_rate": 7.614665424214195e-06,
+      "loss": 0.5934,
+      "step": 2064
+    },
+    {
+      "epoch": 0.2066,
+      "learning_rate": 7.6214458012203726e-06,
+      "loss": 0.092,
+      "step": 2066
+    },
+    {
+      "epoch": 0.2068,
+      "learning_rate": 7.628227337504957e-06,
+      "loss": 0.2973,
+      "step": 2068
+    },
+    {
+      "epoch": 0.207,
+      "learning_rate": 7.635010029762755e-06,
+      "loss": 0.6811,
+      "step": 2070
+    },
+    {
+      "epoch": 0.2072,
+      "learning_rate": 7.64179387468792e-06,
+      "loss": 0.4744,
+      "step": 2072
+    },
+    {
+      "epoch": 0.2074,
+      "learning_rate": 7.648578868974102e-06,
+      "loss": 1.1269,
+      "step": 2074
+    },
+    {
+      "epoch": 0.2076,
+      "learning_rate": 7.65536500931436e-06,
+      "loss": 0.5663,
+      "step": 2076
+    },
+    {
+      "epoch": 0.2078,
+      "learning_rate": 7.662152292401265e-06,
+      "loss": 0.5893,
+      "step": 2078
+    },
+    {
+      "epoch": 0.208,
+      "learning_rate": 7.668940714926724e-06,
+      "loss": 0.8844,
+      "step": 2080
+    },
+    {
+      "epoch": 0.2082,
+      "learning_rate": 7.675730273582142e-06,
+      "loss": 0.3014,
+      "step": 2082
+    },
+    {
+      "epoch": 0.2084,
+      "learning_rate": 7.682520965058426e-06,
+      "loss": 1.0382,
+      "step": 2084
+    },
+    {
+      "epoch": 0.2086,
+      "learning_rate": 7.689312786045822e-06,
+      "loss": 0.6024,
+      "step": 2086
+    },
+    {
+      "epoch": 0.2088,
+      "learning_rate": 7.696105733234092e-06,
+      "loss": 0.6817,
+      "step": 2088
+    },
+    {
+      "epoch": 0.209,
+      "learning_rate": 7.702899803312443e-06,
+      "loss": 0.8235,
+      "step": 2090
+    },
+    {
+      "epoch": 0.2092,
+      "learning_rate": 7.709694992969523e-06,
+      "loss": 0.0231,
+      "step": 2092
+    },
+    {
+      "epoch": 0.2094,
+      "learning_rate": 7.716491298893441e-06,
+      "loss": 1.1488,
+      "step": 2094
+    },
+    {
+      "epoch": 0.2096,
+      "learning_rate": 7.723288717771759e-06,
+      "loss": 0.1179,
+      "step": 2096
+    },
+    {
+      "epoch": 0.2098,
+      "learning_rate": 7.730087246291498e-06,
+      "loss": 0.0381,
+      "step": 2098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 7.736886881139138e-06,
+      "loss": 0.4626,
+      "step": 2100
+    },
+    {
+      "epoch": 0.2102,
+      "learning_rate": 7.74368761900062e-06,
+      "loss": 0.6441,
+      "step": 2102
+    },
+    {
+      "epoch": 0.2104,
+      "learning_rate": 7.750489456561346e-06,
+      "loss": 0.3431,
+      "step": 2104
+    },
+    {
+      "epoch": 0.2106,
+      "learning_rate": 7.757292390506184e-06,
+      "loss": 0.9073,
+      "step": 2106
+    },
+    {
+      "epoch": 0.2108,
+      "learning_rate": 7.764096417519465e-06,
+      "loss": 0.9509,
+      "step": 2108
+    },
+    {
+      "epoch": 0.211,
+      "learning_rate": 7.770901534284991e-06,
+      "loss": 0.9929,
+      "step": 2110
+    },
+    {
+      "epoch": 0.2112,
+      "learning_rate": 7.77770773748603e-06,
+      "loss": 0.5186,
+      "step": 2112
+    },
+    {
+      "epoch": 0.2114,
+      "learning_rate": 7.78451502380532e-06,
+      "loss": 0.8202,
+      "step": 2114
+    },
+    {
+      "epoch": 0.2116,
+      "learning_rate": 7.791323389925074e-06,
+      "loss": 0.6126,
+      "step": 2116
+    },
+    {
+      "epoch": 0.2118,
+      "learning_rate": 7.798132832526976e-06,
+      "loss": 0.1882,
+      "step": 2118
+    },
+    {
+      "epoch": 0.212,
+      "learning_rate": 7.804943348292189e-06,
+      "loss": 0.4641,
+      "step": 2120
+    },
+    {
+      "epoch": 0.2122,
+      "learning_rate": 7.811754933901346e-06,
+      "loss": 0.6145,
+      "step": 2122
+    },
+    {
+      "epoch": 0.2124,
+      "learning_rate": 7.818567586034583e-06,
+      "loss": 1.9143,
+      "step": 2124
+    },
+    {
+      "epoch": 0.2126,
+      "learning_rate": 7.825381301371444e-06,
+      "loss": 0.7333,
+      "step": 2126
+    },
+    {
+      "epoch": 0.2128,
+      "learning_rate": 7.832196076591057e-06,
+      "loss": 0.0457,
+      "step": 2128
+    },
+    {
+      "epoch": 0.213,
+      "learning_rate": 7.839011908371987e-06,
+      "loss": 0.5735,
+      "step": 2130
+    },
+    {
+      "epoch": 0.2132,
+      "learning_rate": 7.845828793392243e-06,
+      "loss": 0.5765,
+      "step": 2132
+    },
+    {
+      "epoch": 0.2134,
+      "learning_rate": 7.852646728329358e-06,
+      "loss": 0.9113,
+      "step": 2134
+    },
+    {
+      "epoch": 0.2136,
+      "learning_rate": 7.859465709860387e-06,
+      "loss": 1.8094,
+      "step": 2136
+    },
+    {
+      "epoch": 0.2138,
+      "learning_rate": 7.866285734661845e-06,
+      "loss": 0.6522,
+      "step": 2138
+    },
+    {
+      "epoch": 0.214,
+      "learning_rate": 7.87310679940968e-06,
+      "loss": 1.0383,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2142,
+      "learning_rate": 7.879928900779441e-06,
+      "loss": 0.5937,
+      "step": 2142
+    },
+    {
+      "epoch": 0.2144,
+      "learning_rate": 7.886752035446117e-06,
+      "loss": 0.1322,
+      "step": 2144
+    },
+    {
+      "epoch": 0.2146,
+      "learning_rate": 7.893576200084164e-06,
+      "loss": 1.2282,
+      "step": 2146
+    },
+    {
+      "epoch": 0.2148,
+      "learning_rate": 7.900401391367563e-06,
+      "loss": 0.5155,
+      "step": 2148
+    },
+    {
+      "epoch": 0.215,
+      "learning_rate": 7.907227605969852e-06,
+      "loss": 1.0344,
+      "step": 2150
+    },
+    {
+      "epoch": 0.2152,
+      "learning_rate": 7.914054840563965e-06,
+      "loss": 0.4332,
+      "step": 2152
+    },
+    {
+      "epoch": 0.2154,
+      "learning_rate": 7.92088309182239e-06,
+      "loss": 1.1363,
+      "step": 2154
+    },
+    {
+      "epoch": 0.2156,
+      "learning_rate": 7.927712356417176e-06,
+      "loss": 0.6352,
+      "step": 2156
+    },
+    {
+      "epoch": 0.2158,
+      "learning_rate": 7.934542631019767e-06,
+      "loss": 0.5466,
+      "step": 2158
+    },
+    {
+      "epoch": 0.216,
+      "learning_rate": 7.941373912301187e-06,
+      "loss": 1.0844,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2162,
+      "learning_rate": 7.948206196931937e-06,
+      "loss": 0.5725,
+      "step": 2162
+    },
+    {
+      "epoch": 0.2164,
+      "learning_rate": 7.955039481582096e-06,
+      "loss": 0.9052,
+      "step": 2164
+    },
+    {
+      "epoch": 0.2166,
+      "learning_rate": 7.961873762921153e-06,
+      "loss": 0.5947,
+      "step": 2166
+    },
+    {
+      "epoch": 0.2168,
+      "learning_rate": 7.968709037618177e-06,
+      "loss": 0.8227,
+      "step": 2168
+    },
+    {
+      "epoch": 0.217,
+      "learning_rate": 7.97554530234174e-06,
+      "loss": 0.5077,
+      "step": 2170
+    },
+    {
+      "epoch": 0.2172,
+      "learning_rate": 7.982382553759928e-06,
+      "loss": 1.4386,
+      "step": 2172
+    },
+    {
+      "epoch": 0.2174,
+      "learning_rate": 7.989220788540351e-06,
+      "loss": 0.4695,
+      "step": 2174
+    },
+    {
+      "epoch": 0.2176,
+      "learning_rate": 7.996060003350134e-06,
+      "loss": 0.6591,
+      "step": 2176
+    },
+    {
+      "epoch": 0.2178,
+      "learning_rate": 8.002900194855927e-06,
+      "loss": 0.8362,
+      "step": 2178
+    },
+    {
+      "epoch": 0.218,
+      "learning_rate": 8.009741359723904e-06,
+      "loss": 1.485,
+      "step": 2180
+    },
+    {
+      "epoch": 0.2182,
+      "learning_rate": 8.016583494619764e-06,
+      "loss": 0.2655,
+      "step": 2182
+    },
+    {
+      "epoch": 0.2184,
+      "learning_rate": 8.023426596208734e-06,
+      "loss": 0.3507,
+      "step": 2184
+    },
+    {
+      "epoch": 0.2186,
+      "learning_rate": 8.03027066115557e-06,
+      "loss": 0.3704,
+      "step": 2186
+    },
+    {
+      "epoch": 0.2188,
+      "learning_rate": 8.037115686124557e-06,
+      "loss": 1.0936,
+      "step": 2188
+    },
+    {
+      "epoch": 0.219,
+      "learning_rate": 8.043961667779511e-06,
+      "loss": 1.1119,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2192,
+      "learning_rate": 8.05080860278379e-06,
+      "loss": 0.9106,
+      "step": 2192
+    },
+    {
+      "epoch": 0.2194,
+      "learning_rate": 8.057656487800274e-06,
+      "loss": 1.0293,
+      "step": 2194
+    },
+    {
+      "epoch": 0.2196,
+      "learning_rate": 8.064505319491391e-06,
+      "loss": 0.3617,
+      "step": 2196
+    },
+    {
+      "epoch": 0.2198,
+      "learning_rate": 8.071355094519103e-06,
+      "loss": 0.1118,
+      "step": 2198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.078205809544911e-06,
+      "loss": 0.8567,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2202,
+      "learning_rate": 8.085057461229862e-06,
+      "loss": 0.5149,
+      "step": 2202
+    },
+    {
+      "epoch": 0.2204,
+      "learning_rate": 8.091910046234562e-06,
+      "loss": 0.2543,
+      "step": 2204
+    },
+    {
+      "epoch": 0.2206,
+      "learning_rate": 8.098763561219089e-06,
+      "loss": 0.0319,
+      "step": 2206
+    },
+    {
+      "epoch": 0.2208,
+      "learning_rate": 8.105618002843178e-06,
+      "loss": 1.1794,
+      "step": 2208
+    },
+    {
+      "epoch": 0.221,
+      "learning_rate": 8.112473367766056e-06,
+      "loss": 0.4744,
+      "step": 2210
+    },
+    {
+      "epoch": 0.2212,
+      "learning_rate": 8.119329652646451e-06,
+      "loss": 0.5783,
+      "step": 2212
+    },
+    {
+      "epoch": 0.2214,
+      "learning_rate": 8.126186854142744e-06,
+      "loss": 1.2524,
+      "step": 2214
+    },
+    {
+      "epoch": 0.2216,
+      "learning_rate": 8.133044968912818e-06,
+      "loss": 0.7335,
+      "step": 2216
+    },
+    {
+      "epoch": 0.2218,
+      "learning_rate": 8.139903993614075e-06,
+      "loss": 0.5789,
+      "step": 2218
+    },
+    {
+      "epoch": 0.222,
+      "learning_rate": 8.146763924903514e-06,
+      "loss": 0.8898,
+      "step": 2220
+    },
+    {
+      "epoch": 0.2222,
+      "learning_rate": 8.153624759437718e-06,
+      "loss": 1.2918,
+      "step": 2222
+    },
+    {
+      "epoch": 0.2224,
+      "learning_rate": 8.160486493872803e-06,
+      "loss": 1.0147,
+      "step": 2224
+    },
+    {
+      "epoch": 0.2226,
+      "learning_rate": 8.167349124864389e-06,
+      "loss": 0.6983,
+      "step": 2226
+    },
+    {
+      "epoch": 0.2228,
+      "learning_rate": 8.174212649067766e-06,
+      "loss": 0.6305,
+      "step": 2228
+    },
+    {
+      "epoch": 0.223,
+      "learning_rate": 8.181077063137735e-06,
+      "loss": 0.7846,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2232,
+      "learning_rate": 8.18794236372863e-06,
+      "loss": 1.0662,
+      "step": 2232
+    },
+    {
+      "epoch": 0.2234,
+      "learning_rate": 8.194808547494386e-06,
+      "loss": 0.5797,
+      "step": 2234
+    },
+    {
+      "epoch": 0.2236,
+      "learning_rate": 8.20167561108856e-06,
+      "loss": 0.459,
+      "step": 2236
+    },
+    {
+      "epoch": 0.2238,
+      "learning_rate": 8.208543551164178e-06,
+      "loss": 0.498,
+      "step": 2238
+    },
+    {
+      "epoch": 0.224,
+      "learning_rate": 8.215412364373906e-06,
+      "loss": 0.6227,
+      "step": 2240
+    },
+    {
+      "epoch": 0.2242,
+      "learning_rate": 8.22228204736997e-06,
+      "loss": 0.6899,
+      "step": 2242
+    },
+    {
+      "epoch": 0.2244,
+      "learning_rate": 8.229152596804167e-06,
+      "loss": 0.617,
+      "step": 2244
+    },
+    {
+      "epoch": 0.2246,
+      "learning_rate": 8.236024009327877e-06,
+      "loss": 1.1118,
+      "step": 2246
+    },
+    {
+      "epoch": 0.2248,
+      "learning_rate": 8.242896281592054e-06,
+      "loss": 0.4163,
+      "step": 2248
+    },
+    {
+      "epoch": 0.225,
+      "learning_rate": 8.249769410247239e-06,
+      "loss": 0.425,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2252,
+      "learning_rate": 8.25664339194355e-06,
+      "loss": 1.0442,
+      "step": 2252
+    },
+    {
+      "epoch": 0.2254,
+      "learning_rate": 8.263518223330695e-06,
+      "loss": 0.4674,
+      "step": 2254
+    },
+    {
+      "epoch": 0.2256,
+      "learning_rate": 8.270393901057963e-06,
+      "loss": 0.1062,
+      "step": 2256
+    },
+    {
+      "epoch": 0.2258,
+      "learning_rate": 8.277270421774231e-06,
+      "loss": 0.55,
+      "step": 2258
+    },
+    {
+      "epoch": 0.226,
+      "learning_rate": 8.284147782127966e-06,
+      "loss": 0.4374,
+      "step": 2260
+    },
+    {
+      "epoch": 0.2262,
+      "learning_rate": 8.29102597876723e-06,
+      "loss": 0.6777,
+      "step": 2262
+    },
+    {
+      "epoch": 0.2264,
+      "learning_rate": 8.29790500833967e-06,
+      "loss": 0.4518,
+      "step": 2264
+    },
+    {
+      "epoch": 0.2266,
+      "learning_rate": 8.304784867492532e-06,
+      "loss": 0.2182,
+      "step": 2266
+    },
+    {
+      "epoch": 0.2268,
+      "learning_rate": 8.311665552872656e-06,
+      "loss": 0.4478,
+      "step": 2268
+    },
+    {
+      "epoch": 0.227,
+      "learning_rate": 8.31854706112648e-06,
+      "loss": 0.5967,
+      "step": 2270
+    },
+    {
+      "epoch": 0.2272,
+      "learning_rate": 8.325429388900039e-06,
+      "loss": 0.5374,
+      "step": 2272
+    },
+    {
+      "epoch": 0.2274,
+      "learning_rate": 8.332312532838972e-06,
+      "loss": 0.2089,
+      "step": 2274
+    },
+    {
+      "epoch": 0.2276,
+      "learning_rate": 8.339196489588515e-06,
+      "loss": 0.0972,
+      "step": 2276
+    },
+    {
+      "epoch": 0.2278,
+      "learning_rate": 8.346081255793516e-06,
+      "loss": 0.1782,
+      "step": 2278
+    },
+    {
+      "epoch": 0.228,
+      "learning_rate": 8.352966828098418e-06,
+      "loss": 0.5135,
+      "step": 2280
+    },
+    {
+      "epoch": 0.2282,
+      "learning_rate": 8.359853203147282e-06,
+      "loss": 1.0634,
+      "step": 2282
+    },
+    {
+      "epoch": 0.2284,
+      "learning_rate": 8.36674037758377e-06,
+      "loss": 0.7258,
+      "step": 2284
+    },
+    {
+      "epoch": 0.2286,
+      "learning_rate": 8.373628348051156e-06,
+      "loss": 0.1033,
+      "step": 2286
+    },
+    {
+      "epoch": 0.2288,
+      "learning_rate": 8.380517111192328e-06,
+      "loss": 0.6747,
+      "step": 2288
+    },
+    {
+      "epoch": 0.229,
+      "learning_rate": 8.387406663649803e-06,
+      "loss": 1.2225,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2292,
+      "learning_rate": 8.394297002065646e-06,
+      "loss": 0.0394,
+      "step": 2292
+    },
+    {
+      "epoch": 0.2294,
+      "learning_rate": 8.401188123081642e-06,
+      "loss": 0.3581,
+      "step": 2294
+    },
+    {
+      "epoch": 0.2296,
+      "learning_rate": 8.40808002333914e-06,
+      "loss": 0.2054,
+      "step": 2296
+    },
+    {
+      "epoch": 0.2298,
+      "learning_rate": 8.414972699479062e-06,
+      "loss": 0.2107,
+      "step": 2298
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.421866148142053e-06,
+      "loss": 0.5955,
+      "step": 2300
+    },
+    {
+      "epoch": 0.2302,
+      "learning_rate": 8.428760365968329e-06,
+      "loss": 0.0886,
+      "step": 2302
+    },
+    {
+      "epoch": 0.2304,
+      "learning_rate": 8.435655349597697e-06,
+      "loss": 0.4981,
+      "step": 2304
+    },
+    {
+      "epoch": 0.2306,
+      "learning_rate": 8.442551095669627e-06,
+      "loss": 0.2291,
+      "step": 2306
+    },
+    {
+      "epoch": 0.2308,
+      "learning_rate": 8.449447600823266e-06,
+      "loss": 0.5326,
+      "step": 2308
+    },
+    {
+      "epoch": 0.231,
+      "learning_rate": 8.456344861697293e-06,
+      "loss": 1.0287,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2312,
+      "learning_rate": 8.463242874930063e-06,
+      "loss": 0.6345,
+      "step": 2312
+    },
+    {
+      "epoch": 0.2314,
+      "learning_rate": 8.470141637159605e-06,
+      "loss": 0.4191,
+      "step": 2314
+    },
+    {
+      "epoch": 0.2316,
+      "learning_rate": 8.477041145023546e-06,
+      "loss": 0.5379,
+      "step": 2316
+    },
+    {
+      "epoch": 0.2318,
+      "learning_rate": 8.483941395159114e-06,
+      "loss": 0.0358,
+      "step": 2318
+    },
+    {
+      "epoch": 0.232,
+      "learning_rate": 8.490842384203211e-06,
+      "loss": 0.6049,
+      "step": 2320
+    },
+    {
+      "epoch": 0.2322,
+      "learning_rate": 8.497744108792431e-06,
+      "loss": 0.0764,
+      "step": 2322
+    },
+    {
+      "epoch": 0.2324,
+      "learning_rate": 8.504646565562907e-06,
+      "loss": 0.9736,
+      "step": 2324
+    },
+    {
+      "epoch": 0.2326,
+      "learning_rate": 8.511549751150478e-06,
+      "loss": 0.5223,
+      "step": 2326
+    },
+    {
+      "epoch": 0.2328,
+      "learning_rate": 8.518453662190622e-06,
+      "loss": 0.373,
+      "step": 2328
+    },
+    {
+      "epoch": 0.233,
+      "learning_rate": 8.52535829531845e-06,
+      "loss": 0.2623,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2332,
+      "learning_rate": 8.532263647168732e-06,
+      "loss": 0.6542,
+      "step": 2332
+    },
+    {
+      "epoch": 0.2334,
+      "learning_rate": 8.539169714375883e-06,
+      "loss": 0.5866,
+      "step": 2334
+    },
+    {
+      "epoch": 0.2336,
+      "learning_rate": 8.546076493573968e-06,
+      "loss": 0.4316,
+      "step": 2336
+    },
+    {
+      "epoch": 0.2338,
+      "learning_rate": 8.552983981396707e-06,
+      "loss": 0.6212,
+      "step": 2338
+    },
+    {
+      "epoch": 0.234,
+      "learning_rate": 8.559892174477475e-06,
+      "loss": 1.1798,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2342,
+      "learning_rate": 8.566801069449304e-06,
+      "loss": 0.1929,
+      "step": 2342
+    },
+    {
+      "epoch": 0.2344,
+      "learning_rate": 8.573710662944881e-06,
+      "loss": 0.2592,
+      "step": 2344
+    },
+    {
+      "epoch": 0.2346,
+      "learning_rate": 8.580620951596553e-06,
+      "loss": 0.5937,
+      "step": 2346
+    },
+    {
+      "epoch": 0.2348,
+      "learning_rate": 8.58753193203633e-06,
+      "loss": 0.613,
+      "step": 2348
+    },
+    {
+      "epoch": 0.235,
+      "learning_rate": 8.594443600895886e-06,
+      "loss": 0.9139,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2352,
+      "learning_rate": 8.601355954806554e-06,
+      "loss": 0.4708,
+      "step": 2352
+    },
+    {
+      "epoch": 0.2354,
+      "learning_rate": 8.60826899039934e-06,
+      "loss": 0.9781,
+      "step": 2354
+    },
+    {
+      "epoch": 0.2356,
+      "learning_rate": 8.615182704304913e-06,
+      "loss": 0.3024,
+      "step": 2356
+    },
+    {
+      "epoch": 0.2358,
+      "learning_rate": 8.622097093153612e-06,
+      "loss": 0.0655,
+      "step": 2358
+    },
+    {
+      "epoch": 0.236,
+      "learning_rate": 8.629012153575452e-06,
+      "loss": 0.6489,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2362,
+      "learning_rate": 8.635927882200128e-06,
+      "loss": 0.1051,
+      "step": 2362
+    },
+    {
+      "epoch": 0.2364,
+      "learning_rate": 8.642844275656949e-06,
+      "loss": 0.7529,
+      "step": 2364
+    },
+    {
+      "epoch": 0.2366,
+      "learning_rate": 8.649761330575e-06,
+      "loss": 0.3893,
+      "step": 2366
+    },
+    {
+      "epoch": 0.2368,
+      "learning_rate": 8.656679043582993e-06,
+      "loss": 0.8924,
+      "step": 2368
+    },
+    {
+      "epoch": 0.237,
+      "learning_rate": 8.663597411309268e-06,
+      "loss": 0.62,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2372,
+      "learning_rate": 8.670516430381947e-06,
+      "loss": 1.2648,
+      "step": 2372
+    },
+    {
+      "epoch": 0.2374,
+      "learning_rate": 8.677436097428766e-06,
+      "loss": 0.0821,
+      "step": 2374
+    },
+    {
+      "epoch": 0.2376,
+      "learning_rate": 8.684356409077183e-06,
+      "loss": 0.4689,
+      "step": 2376
+    },
+    {
+      "epoch": 0.2378,
+      "learning_rate": 8.691277361954266e-06,
+      "loss": 0.5048,
+      "step": 2378
+    },
+    {
+      "epoch": 0.238,
+      "learning_rate": 8.698198952686884e-06,
+      "loss": 0.3686,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2382,
+      "learning_rate": 8.705121177901537e-06,
+      "loss": 1.1811,
+      "step": 2382
+    },
+    {
+      "epoch": 0.2384,
+      "learning_rate": 8.71204403422436e-06,
+      "loss": 0.6216,
+      "step": 2384
+    },
+    {
+      "epoch": 0.2386,
+      "learning_rate": 8.718967518281292e-06,
+      "loss": 1.1814,
+      "step": 2386
+    },
+    {
+      "epoch": 0.2388,
+      "learning_rate": 8.725891626697913e-06,
+      "loss": 1.4119,
+      "step": 2388
+    },
+    {
+      "epoch": 0.239,
+      "learning_rate": 8.732816356099459e-06,
+      "loss": 1.1255,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2392,
+      "learning_rate": 8.7397417031109e-06,
+      "loss": 0.3436,
+      "step": 2392
+    },
+    {
+      "epoch": 0.2394,
+      "learning_rate": 8.746667664356962e-06,
+      "loss": 0.1292,
+      "step": 2394
+    },
+    {
+      "epoch": 0.2396,
+      "learning_rate": 8.75359423646196e-06,
+      "loss": 0.5808,
+      "step": 2396
+    },
+    {
+      "epoch": 0.2398,
+      "learning_rate": 8.760521416049986e-06,
+      "loss": 0.6302,
+      "step": 2398
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.767449199744798e-06,
+      "loss": 1.1342,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2402,
+      "learning_rate": 8.774377584169934e-06,
+      "loss": 0.9855,
+      "step": 2402
+    },
+    {
+      "epoch": 0.2404,
+      "learning_rate": 8.781306565948526e-06,
+      "loss": 1.2401,
+      "step": 2404
+    },
+    {
+      "epoch": 0.2406,
+      "learning_rate": 8.788236141703477e-06,
+      "loss": 1.2224,
+      "step": 2406
+    },
+    {
+      "epoch": 0.2408,
+      "learning_rate": 8.795166308057449e-06,
+      "loss": 0.5484,
+      "step": 2408
+    },
+    {
+      "epoch": 0.241,
+      "learning_rate": 8.802097061632706e-06,
+      "loss": 0.4805,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2412,
+      "learning_rate": 8.809028399051302e-06,
+      "loss": 0.9418,
+      "step": 2412
+    },
+    {
+      "epoch": 0.2414,
+      "learning_rate": 8.81596031693499e-06,
+      "loss": 0.7223,
+      "step": 2414
+    },
+    {
+      "epoch": 0.2416,
+      "learning_rate": 8.822892811905235e-06,
+      "loss": 0.637,
+      "step": 2416
+    },
+    {
+      "epoch": 0.2418,
+      "learning_rate": 8.829825880583224e-06,
+      "loss": 0.9395,
+      "step": 2418
+    },
+    {
+      "epoch": 0.242,
+      "learning_rate": 8.836759519589864e-06,
+      "loss": 0.218,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2422,
+      "learning_rate": 8.84369372554578e-06,
+      "loss": 0.799,
+      "step": 2422
+    },
+    {
+      "epoch": 0.2424,
+      "learning_rate": 8.850628495071331e-06,
+      "loss": 0.0504,
+      "step": 2424
+    },
+    {
+      "epoch": 0.2426,
+      "learning_rate": 8.85756382478659e-06,
+      "loss": 0.6481,
+      "step": 2426
+    },
+    {
+      "epoch": 0.2428,
+      "learning_rate": 8.86449971131136e-06,
+      "loss": 0.3193,
+      "step": 2428
+    },
+    {
+      "epoch": 0.243,
+      "learning_rate": 8.87143615126518e-06,
+      "loss": 0.6066,
+      "step": 2430
+    },
+    {
+      "epoch": 0.2432,
+      "learning_rate": 8.878373141267307e-06,
+      "loss": 0.6824,
+      "step": 2432
+    },
+    {
+      "epoch": 0.2434,
+      "learning_rate": 8.88531067793674e-06,
+      "loss": 0.3553,
+      "step": 2434
+    },
+    {
+      "epoch": 0.2436,
+      "learning_rate": 8.892248757892209e-06,
+      "loss": 0.3881,
+      "step": 2436
+    },
+    {
+      "epoch": 0.2438,
+      "learning_rate": 8.899187377752173e-06,
+      "loss": 1.2964,
+      "step": 2438
+    },
+    {
+      "epoch": 0.244,
+      "learning_rate": 8.906126534134838e-06,
+      "loss": 1.0246,
+      "step": 2440
+    },
+    {
+      "epoch": 0.2442,
+      "learning_rate": 8.913066223658141e-06,
+      "loss": 0.0991,
+      "step": 2442
+    },
+    {
+      "epoch": 0.2444,
+      "learning_rate": 8.920006442939763e-06,
+      "loss": 0.4986,
+      "step": 2444
+    },
+    {
+      "epoch": 0.2446,
+      "learning_rate": 8.926947188597127e-06,
+      "loss": 0.1069,
+      "step": 2446
+    },
+    {
+      "epoch": 0.2448,
+      "learning_rate": 8.93388845724741e-06,
+      "loss": 0.552,
+      "step": 2448
+    },
+    {
+      "epoch": 0.245,
+      "learning_rate": 8.940830245507473e-06,
+      "loss": 0.521,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2452,
+      "learning_rate": 8.947772549994027e-06,
+      "loss": 0.7212,
+      "step": 2452
+    },
+    {
+      "epoch": 0.2454,
+      "learning_rate": 8.954715367323473e-06,
+      "loss": 0.122,
+      "step": 2454
+    },
+    {
+      "epoch": 0.2456,
+      "learning_rate": 8.961658694111917e-06,
+      "loss": 0.6621,
+      "step": 2456
+    },
+    {
+      "epoch": 0.2458,
+      "learning_rate": 8.968602526975317e-06,
+      "loss": 0.5459,
+      "step": 2458
+    },
+    {
+      "epoch": 0.246,
+      "learning_rate": 8.975546862529313e-06,
+      "loss": 0.9801,
+      "step": 2460
+    },
+    {
+      "epoch": 0.2462,
+      "learning_rate": 8.982491697389344e-06,
+      "loss": 0.9218,
+      "step": 2462
+    },
+    {
+      "epoch": 0.2464,
+      "learning_rate": 8.989437028170524e-06,
+      "loss": 0.6905,
+      "step": 2464
+    },
+    {
+      "epoch": 0.2466,
+      "learning_rate": 8.996382851487839e-06,
+      "loss": 0.4498,
+      "step": 2466
+    },
+    {
+      "epoch": 0.2468,
+      "learning_rate": 9.003329163955976e-06,
+      "loss": 0.569,
+      "step": 2468
+    },
+    {
+      "epoch": 0.247,
+      "learning_rate": 9.010275962189356e-06,
+      "loss": 0.3475,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2472,
+      "learning_rate": 9.01722324280219e-06,
+      "loss": 0.8878,
+      "step": 2472
+    },
+    {
+      "epoch": 0.2474,
+      "learning_rate": 9.024171002408509e-06,
+      "loss": 0.4537,
+      "step": 2474
+    },
+    {
+      "epoch": 0.2476,
+      "learning_rate": 9.031119237622011e-06,
+      "loss": 0.7523,
+      "step": 2476
+    },
+    {
+      "epoch": 0.2478,
+      "learning_rate": 9.03806794505621e-06,
+      "loss": 0.743,
+      "step": 2478
+    },
+    {
+      "epoch": 0.248,
+      "learning_rate": 9.045017121324441e-06,
+      "loss": 0.5473,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2482,
+      "learning_rate": 9.051966763039708e-06,
+      "loss": 0.0981,
+      "step": 2482
+    },
+    {
+      "epoch": 0.2484,
+      "learning_rate": 9.058916866814859e-06,
+      "loss": 0.9218,
+      "step": 2484
+    },
+    {
+      "epoch": 0.2486,
+      "learning_rate": 9.065867429262497e-06,
+      "loss": 0.6254,
+      "step": 2486
+    },
+    {
+      "epoch": 0.2488,
+      "learning_rate": 9.072818446995e-06,
+      "loss": 0.0372,
+      "step": 2488
+    },
+    {
+      "epoch": 0.249,
+      "learning_rate": 9.07976991662453e-06,
+      "loss": 0.2845,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2492,
+      "learning_rate": 9.086721834763006e-06,
+      "loss": 1.1598,
+      "step": 2492
+    },
+    {
+      "epoch": 0.2494,
+      "learning_rate": 9.093674198022198e-06,
+      "loss": 0.5652,
+      "step": 2494
+    },
+    {
+      "epoch": 0.2496,
+      "learning_rate": 9.10062700301356e-06,
+      "loss": 0.6208,
+      "step": 2496
+    },
+    {
+      "epoch": 0.2498,
+      "learning_rate": 9.107580246348395e-06,
+      "loss": 0.5081,
+      "step": 2498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.114533924637776e-06,
+      "loss": 0.2605,
+      "step": 2500
+    },
+    {
+      "epoch": 0.2502,
+      "learning_rate": 9.121488034492567e-06,
+      "loss": 0.6296,
+      "step": 2502
+    },
+    {
+      "epoch": 0.2504,
+      "learning_rate": 9.128442572523417e-06,
+      "loss": 0.6617,
+      "step": 2504
+    },
+    {
+      "epoch": 0.2506,
+      "learning_rate": 9.135397535340768e-06,
+      "loss": 0.1213,
+      "step": 2506
+    },
+    {
+      "epoch": 0.2508,
+      "learning_rate": 9.142352919554857e-06,
+      "loss": 0.5687,
+      "step": 2508
+    },
+    {
+      "epoch": 0.251,
+      "learning_rate": 9.149308721775717e-06,
+      "loss": 0.6812,
+      "step": 2510
+    },
+    {
+      "epoch": 0.2512,
+      "learning_rate": 9.15626493861317e-06,
+      "loss": 0.5968,
+      "step": 2512
+    },
+    {
+      "epoch": 0.2514,
+      "learning_rate": 9.16322156667684e-06,
+      "loss": 0.8102,
+      "step": 2514
+    },
+    {
+      "epoch": 0.2516,
+      "learning_rate": 9.170178602576156e-06,
+      "loss": 0.7884,
+      "step": 2516
+    },
+    {
+      "epoch": 0.2518,
+      "learning_rate": 9.177136042920338e-06,
+      "loss": 0.6671,
+      "step": 2518
+    },
+    {
+      "epoch": 0.252,
+      "learning_rate": 9.18409388431842e-06,
+      "loss": 0.1888,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2522,
+      "learning_rate": 9.191052123379227e-06,
+      "loss": 0.8438,
+      "step": 2522
+    },
+    {
+      "epoch": 0.2524,
+      "learning_rate": 9.198010756711405e-06,
+      "loss": 0.0612,
+      "step": 2524
+    },
+    {
+      "epoch": 0.2526,
+      "learning_rate": 9.204969780923396e-06,
+      "loss": 0.154,
+      "step": 2526
+    },
+    {
+      "epoch": 0.2528,
+      "learning_rate": 9.211929192623456e-06,
+      "loss": 0.6171,
+      "step": 2528
+    },
+    {
+      "epoch": 0.253,
+      "learning_rate": 9.218888988419656e-06,
+      "loss": 0.2328,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2532,
+      "learning_rate": 9.225849164919874e-06,
+      "loss": 0.0545,
+      "step": 2532
+    },
+    {
+      "epoch": 0.2534,
+      "learning_rate": 9.232809718731822e-06,
+      "loss": 0.2465,
+      "step": 2534
+    },
+    {
+      "epoch": 0.2536,
+      "learning_rate": 9.239770646462957e-06,
+      "loss": 0.9814,
+      "step": 2536
+    },
+    {
+      "epoch": 0.2538,
+      "learning_rate": 9.246731944720663e-06,
+      "loss": 1.0083,
+      "step": 2538
+    },
+    {
+      "epoch": 0.254,
+      "learning_rate": 9.253693610112084e-06,
+      "loss": 0.5033,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2542,
+      "learning_rate": 9.26065563924414e-06,
+      "loss": 1.1215,
+      "step": 2542
+    },
+    {
+      "epoch": 0.2544,
+      "learning_rate": 9.267618028723672e-06,
+      "loss": 0.9422,
+      "step": 2544
+    },
+    {
+      "epoch": 0.2546,
+      "learning_rate": 9.274580775157299e-06,
+      "loss": 0.4788,
+      "step": 2546
+    },
+    {
+      "epoch": 0.2548,
+      "learning_rate": 9.281543875151422e-06,
+      "loss": 1.0695,
+      "step": 2548
+    },
+    {
+      "epoch": 0.255,
+      "learning_rate": 9.288507325312319e-06,
+      "loss": 0.1261,
+      "step": 2550
+    },
+    {
+      "epoch": 0.2552,
+      "learning_rate": 9.295471122246118e-06,
+      "loss": 1.3237,
+      "step": 2552
+    },
+    {
+      "epoch": 0.2554,
+      "learning_rate": 9.302435262558752e-06,
+      "loss": 0.7441,
+      "step": 2554
+    },
+    {
+      "epoch": 0.2556,
+      "learning_rate": 9.309399742855945e-06,
+      "loss": 1.056,
+      "step": 2556
+    },
+    {
+      "epoch": 0.2558,
+      "learning_rate": 9.316364559743298e-06,
+      "loss": 0.3096,
+      "step": 2558
+    },
+    {
+      "epoch": 0.256,
+      "learning_rate": 9.323329709826295e-06,
+      "loss": 0.3603,
+      "step": 2560
+    },
+    {
+      "epoch": 0.2562,
+      "learning_rate": 9.330295189710153e-06,
+      "loss": 0.3949,
+      "step": 2562
+    },
+    {
+      "epoch": 0.2564,
+      "learning_rate": 9.337260995999983e-06,
+      "loss": 0.6971,
+      "step": 2564
+    },
+    {
+      "epoch": 0.2566,
+      "learning_rate": 9.344227125300788e-06,
+      "loss": 1.0358,
+      "step": 2566
+    },
+    {
+      "epoch": 0.2568,
+      "learning_rate": 9.351193574217306e-06,
+      "loss": 0.111,
+      "step": 2568
+    },
+    {
+      "epoch": 0.257,
+      "learning_rate": 9.358160339354196e-06,
+      "loss": 0.9514,
+      "step": 2570
+    },
+    {
+      "epoch": 0.2572,
+      "learning_rate": 9.365127417315938e-06,
+      "loss": 1.2384,
+      "step": 2572
+    },
+    {
+      "epoch": 0.2574,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 0.2459,
+      "step": 2574
+    },
+    {
+      "epoch": 0.2576,
+      "learning_rate": 9.379062498131158e-06,
+      "loss": 0.3088,
+      "step": 2576
+    },
+    {
+      "epoch": 0.2578,
+      "learning_rate": 9.386030494192826e-06,
+      "loss": 0.8697,
+      "step": 2578
+    },
+    {
+      "epoch": 0.258,
+      "learning_rate": 9.39299878949581e-06,
+      "loss": 0.9074,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2582,
+      "learning_rate": 9.39996738064379e-06,
+      "loss": 0.0552,
+      "step": 2582
+    },
+    {
+      "epoch": 0.2584,
+      "learning_rate": 9.406936264240381e-06,
+      "loss": 0.0653,
+      "step": 2584
+    },
+    {
+      "epoch": 0.2586,
+      "learning_rate": 9.413905436889032e-06,
+      "loss": 0.4585,
+      "step": 2586
+    },
+    {
+      "epoch": 0.2588,
+      "learning_rate": 9.420874895193054e-06,
+      "loss": 0.2083,
+      "step": 2588
+    },
+    {
+      "epoch": 0.259,
+      "learning_rate": 9.427844635755615e-06,
+      "loss": 0.6948,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2592,
+      "learning_rate": 9.43481465517975e-06,
+      "loss": 1.0577,
+      "step": 2592
+    },
+    {
+      "epoch": 0.2594,
+      "learning_rate": 9.441784950068357e-06,
+      "loss": 0.7798,
+      "step": 2594
+    },
+    {
+      "epoch": 0.2596,
+      "learning_rate": 9.448755517024198e-06,
+      "loss": 0.7489,
+      "step": 2596
+    },
+    {
+      "epoch": 0.2598,
+      "learning_rate": 9.455726352649904e-06,
+      "loss": 0.6037,
+      "step": 2598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.462697453547973e-06,
+      "loss": 0.5209,
+      "step": 2600
+    },
+    {
+      "epoch": 0.2602,
+      "learning_rate": 9.469668816320777e-06,
+      "loss": 0.3988,
+      "step": 2602
+    },
+    {
+      "epoch": 0.2604,
+      "learning_rate": 9.476640437570556e-06,
+      "loss": 0.6512,
+      "step": 2604
+    },
+    {
+      "epoch": 0.2606,
+      "learning_rate": 9.483612313899446e-06,
+      "loss": 0.3288,
+      "step": 2606
+    },
+    {
+      "epoch": 0.2608,
+      "learning_rate": 9.490584441909386e-06,
+      "loss": 1.1642,
+      "step": 2608
+    },
+    {
+      "epoch": 0.261,
+      "learning_rate": 9.497556818202297e-06,
+      "loss": 0.0728,
+      "step": 2610
+    },
+    {
+      "epoch": 0.2612,
+      "learning_rate": 9.504529439379911e-06,
+      "loss": 0.6924,
+      "step": 2612
+    },
+    {
+      "epoch": 0.2614,
+      "learning_rate": 9.511502302043859e-06,
+      "loss": 0.0143,
+      "step": 2614
+    },
+    {
+      "epoch": 0.2616,
+      "learning_rate": 9.51847540279565e-06,
+      "loss": 0.4561,
+      "step": 2616
+    },
+    {
+      "epoch": 0.2618,
+      "learning_rate": 9.52544873823668e-06,
+      "loss": 0.0584,
+      "step": 2618
+    },
+    {
+      "epoch": 0.262,
+      "learning_rate": 9.532422304968247e-06,
+      "loss": 0.8771,
+      "step": 2620
+    },
+    {
+      "epoch": 0.2622,
+      "learning_rate": 9.539396099591469e-06,
+      "loss": 0.6973,
+      "step": 2622
+    },
+    {
+      "epoch": 0.2624,
+      "learning_rate": 9.546370118707451e-06,
+      "loss": 0.9277,
+      "step": 2624
+    },
+    {
+      "epoch": 0.2626,
+      "learning_rate": 9.553344358917146e-06,
+      "loss": 0.8414,
+      "step": 2626
+    },
+    {
+      "epoch": 0.2628,
+      "learning_rate": 9.560318816821357e-06,
+      "loss": 0.3749,
+      "step": 2628
+    },
+    {
+      "epoch": 0.263,
+      "learning_rate": 9.567293489020816e-06,
+      "loss": 0.6977,
+      "step": 2630
+    },
+    {
+      "epoch": 0.2632,
+      "learning_rate": 9.57426837211621e-06,
+      "loss": 0.9827,
+      "step": 2632
+    },
+    {
+      "epoch": 0.2634,
+      "learning_rate": 9.581243462708009e-06,
+      "loss": 1.3967,
+      "step": 2634
+    },
+    {
+      "epoch": 0.2636,
+      "learning_rate": 9.58821875739664e-06,
+      "loss": 0.9119,
+      "step": 2636
+    },
+    {
+      "epoch": 0.2638,
+      "learning_rate": 9.595194252782461e-06,
+      "loss": 0.3969,
+      "step": 2638
+    },
+    {
+      "epoch": 0.264,
+      "learning_rate": 9.602169945465705e-06,
+      "loss": 0.8051,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2642,
+      "learning_rate": 9.609145832046469e-06,
+      "loss": 0.4059,
+      "step": 2642
+    },
+    {
+      "epoch": 0.2644,
+      "learning_rate": 9.616121909124786e-06,
+      "loss": 0.711,
+      "step": 2644
+    },
+    {
+      "epoch": 0.2646,
+      "learning_rate": 9.623098173300656e-06,
+      "loss": 0.6229,
+      "step": 2646
+    },
+    {
+      "epoch": 0.2648,
+      "learning_rate": 9.630074621173882e-06,
+      "loss": 0.6614,
+      "step": 2648
+    },
+    {
+      "epoch": 0.265,
+      "learning_rate": 9.637051249344225e-06,
+      "loss": 0.7455,
+      "step": 2650
+    },
+    {
+      "epoch": 0.2652,
+      "learning_rate": 9.644028054411416e-06,
+      "loss": 0.7979,
+      "step": 2652
+    },
+    {
+      "epoch": 0.2654,
+      "learning_rate": 9.651005032974991e-06,
+      "loss": 0.0707,
+      "step": 2654
+    },
+    {
+      "epoch": 0.2656,
+      "learning_rate": 9.657982181634473e-06,
+      "loss": 0.0673,
+      "step": 2656
+    },
+    {
+      "epoch": 0.2658,
+      "learning_rate": 9.664959496989285e-06,
+      "loss": 1.1024,
+      "step": 2658
+    },
+    {
+      "epoch": 0.266,
+      "learning_rate": 9.671936975638768e-06,
+      "loss": 0.4239,
+      "step": 2660
+    },
+    {
+      "epoch": 0.2662,
+      "learning_rate": 9.678914614182184e-06,
+      "loss": 0.5091,
+      "step": 2662
+    },
+    {
+      "epoch": 0.2664,
+      "learning_rate": 9.685892409218716e-06,
+      "loss": 0.4673,
+      "step": 2664
+    },
+    {
+      "epoch": 0.2666,
+      "learning_rate": 9.69287035734747e-06,
+      "loss": 0.4289,
+      "step": 2666
+    },
+    {
+      "epoch": 0.2668,
+      "learning_rate": 9.699848455167484e-06,
+      "loss": 1.0221,
+      "step": 2668
+    },
+    {
+      "epoch": 0.267,
+      "learning_rate": 9.706826699277714e-06,
+      "loss": 0.7471,
+      "step": 2670
+    },
+    {
+      "epoch": 0.2672,
+      "learning_rate": 9.713805086277048e-06,
+      "loss": 0.4852,
+      "step": 2672
+    },
+    {
+      "epoch": 0.2674,
+      "learning_rate": 9.720783612764307e-06,
+      "loss": 1.1593,
+      "step": 2674
+    },
+    {
+      "epoch": 0.2676,
+      "learning_rate": 9.727762275338242e-06,
+      "loss": 1.1435,
+      "step": 2676
+    },
+    {
+      "epoch": 0.2678,
+      "learning_rate": 9.734741070597535e-06,
+      "loss": 0.5967,
+      "step": 2678
+    },
+    {
+      "epoch": 0.268,
+      "learning_rate": 9.741719995140809e-06,
+      "loss": 0.4029,
+      "step": 2680
+    },
+    {
+      "epoch": 0.2682,
+      "learning_rate": 9.74869904556662e-06,
+      "loss": 1.4407,
+      "step": 2682
+    },
+    {
+      "epoch": 0.2684,
+      "learning_rate": 9.755678218473462e-06,
+      "loss": 0.4413,
+      "step": 2684
+    },
+    {
+      "epoch": 0.2686,
+      "learning_rate": 9.762657510459774e-06,
+      "loss": 0.351,
+      "step": 2686
+    },
+    {
+      "epoch": 0.2688,
+      "learning_rate": 9.76963691812393e-06,
+      "loss": 0.7666,
+      "step": 2688
+    },
+    {
+      "epoch": 0.269,
+      "learning_rate": 9.776616438064255e-06,
+      "loss": 0.5353,
+      "step": 2690
+    },
+    {
+      "epoch": 0.2692,
+      "learning_rate": 9.783596066879033e-06,
+      "loss": 0.2882,
+      "step": 2692
+    },
+    {
+      "epoch": 0.2694,
+      "learning_rate": 9.790575801166422e-06,
+      "loss": 1.1417,
+      "step": 2694
+    },
+    {
+      "epoch": 0.2696,
+      "learning_rate": 9.79755563752464e-06,
+      "loss": 0.6287,
+      "step": 2696
+    },
+    {
+      "epoch": 0.2698,
+      "learning_rate": 9.804535572551782e-06,
+      "loss": 0.1073,
+      "step": 2698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.811515602845927e-06,
+      "loss": 0.8812,
+      "step": 2700
+    },
+    {
+      "epoch": 0.2702,
+      "learning_rate": 9.818495725005043e-06,
+      "loss": 0.1838,
+      "step": 2702
+    },
+    {
+      "epoch": 0.2704,
+      "learning_rate": 9.825475935627155e-06,
+      "loss": 1.3748,
+      "step": 2704
+    },
+    {
+      "epoch": 0.2706,
+      "learning_rate": 9.832456231310194e-06,
+      "loss": 0.1045,
+      "step": 2706
+    },
+    {
+      "epoch": 0.2708,
+      "learning_rate": 9.839436608651995e-06,
+      "loss": 0.851,
+      "step": 2708
+    },
+    {
+      "epoch": 0.271,
+      "learning_rate": 9.846417064250459e-06,
+      "loss": 0.9591,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2712,
+      "learning_rate": 9.853397594703399e-06,
+      "loss": 0.5353,
+      "step": 2712
+    },
+    {
+      "epoch": 0.2714,
+      "learning_rate": 9.860378196608552e-06,
+      "loss": 0.0195,
+      "step": 2714
+    },
+    {
+      "epoch": 0.2716,
+      "learning_rate": 9.867358866563658e-06,
+      "loss": 0.7037,
+      "step": 2716
+    },
+    {
+      "epoch": 0.2718,
+      "learning_rate": 9.874339601166479e-06,
+      "loss": 1.0111,
+      "step": 2718
+    },
+    {
+      "epoch": 0.272,
+      "learning_rate": 9.881320397014632e-06,
+      "loss": 0.7201,
+      "step": 2720
+    },
+    {
+      "epoch": 0.2722,
+      "learning_rate": 9.888301250705765e-06,
+      "loss": 0.0098,
+      "step": 2722
+    },
+    {
+      "epoch": 0.2724,
+      "learning_rate": 9.895282158837527e-06,
+      "loss": 2.895,
+      "step": 2724
+    },
+    {
+      "epoch": 0.2726,
+      "learning_rate": 9.902263118007513e-06,
+      "loss": 0.9113,
+      "step": 2726
+    },
+    {
+      "epoch": 0.2728,
+      "learning_rate": 9.90924412481325e-06,
+      "loss": 0.8591,
+      "step": 2728
+    },
+    {
+      "epoch": 0.273,
+      "learning_rate": 9.916225175852278e-06,
+      "loss": 0.5503,
+      "step": 2730
+    },
+    {
+      "epoch": 0.2732,
+      "learning_rate": 9.923206267722174e-06,
+      "loss": 0.0409,
+      "step": 2732
+    },
+    {
+      "epoch": 0.2734,
+      "learning_rate": 9.930187397020385e-06,
+      "loss": 1.3571,
+      "step": 2734
+    },
+    {
+      "epoch": 0.2736,
+      "learning_rate": 9.937168560344412e-06,
+      "loss": 0.3465,
+      "step": 2736
+    },
+    {
+      "epoch": 0.2738,
+      "learning_rate": 9.944149754291716e-06,
+      "loss": 0.507,
+      "step": 2738
+    },
+    {
+      "epoch": 0.274,
+      "learning_rate": 9.951130975459755e-06,
+      "loss": 0.4292,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2742,
+      "learning_rate": 9.95811222044596e-06,
+      "loss": 0.5174,
+      "step": 2742
+    },
+    {
+      "epoch": 0.2744,
+      "learning_rate": 9.965093485847761e-06,
+      "loss": 0.988,
+      "step": 2744
+    },
+    {
+      "epoch": 0.2746,
+      "learning_rate": 9.972074768262572e-06,
+      "loss": 0.425,
+      "step": 2746
+    },
+    {
+      "epoch": 0.2748,
+      "learning_rate": 9.979056064287804e-06,
+      "loss": 0.6884,
+      "step": 2748
+    },
+    {
+      "epoch": 0.275,
+      "learning_rate": 9.986037370520855e-06,
+      "loss": 0.5523,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2752,
+      "learning_rate": 9.993018683559122e-06,
+      "loss": 0.3832,
+      "step": 2752
+    },
+    {
+      "epoch": 0.2754,
+      "learning_rate": 9.999999999999996e-06,
+      "loss": 0.5644,
+      "step": 2754
+    },
+    {
+      "epoch": 0.2756,
+      "learning_rate": 1.0006981316440871e-05,
+      "loss": 0.5777,
+      "step": 2756
+    },
+    {
+      "epoch": 0.2758,
+      "learning_rate": 1.0013962629479139e-05,
+      "loss": 0.5476,
+      "step": 2758
+    },
+    {
+      "epoch": 0.276,
+      "learning_rate": 1.0020943935712188e-05,
+      "loss": 0.5584,
+      "step": 2760
+    },
+    {
+      "epoch": 0.2762,
+      "learning_rate": 1.0027925231737419e-05,
+      "loss": 0.5859,
+      "step": 2762
+    },
+    {
+      "epoch": 0.2764,
+      "learning_rate": 1.0034906514152232e-05,
+      "loss": 1.0007,
+      "step": 2764
+    },
+    {
+      "epoch": 0.2766,
+      "learning_rate": 1.0041887779554034e-05,
+      "loss": 0.4703,
+      "step": 2766
+    },
+    {
+      "epoch": 0.2768,
+      "learning_rate": 1.004886902454024e-05,
+      "loss": 1.528,
+      "step": 2768
+    },
+    {
+      "epoch": 0.277,
+      "learning_rate": 1.0055850245708276e-05,
+      "loss": 1.2099,
+      "step": 2770
+    },
+    {
+      "epoch": 0.2772,
+      "learning_rate": 1.0062831439655583e-05,
+      "loss": 0.6075,
+      "step": 2772
+    },
+    {
+      "epoch": 0.2774,
+      "learning_rate": 1.0069812602979607e-05,
+      "loss": 1.1831,
+      "step": 2774
+    },
+    {
+      "epoch": 0.2776,
+      "learning_rate": 1.0076793732277819e-05,
+      "loss": 0.526,
+      "step": 2776
+    },
+    {
+      "epoch": 0.2778,
+      "learning_rate": 1.0083774824147717e-05,
+      "loss": 0.5777,
+      "step": 2778
+    },
+    {
+      "epoch": 0.278,
+      "learning_rate": 1.0090755875186744e-05,
+      "loss": 0.4097,
+      "step": 2780
+    },
+    {
+      "epoch": 0.2782,
+      "learning_rate": 1.0097736881992482e-05,
+      "loss": 0.9878,
+      "step": 2782
+    },
+    {
+      "epoch": 0.2784,
+      "learning_rate": 1.0104717841162465e-05,
+      "loss": 0.0221,
+      "step": 2784
+    },
+    {
+      "epoch": 0.2786,
+      "learning_rate": 1.011169874929423e-05,
+      "loss": 0.5212,
+      "step": 2786
+    },
+    {
+      "epoch": 0.2788,
+      "learning_rate": 1.0118679602985362e-05,
+      "loss": 0.495,
+      "step": 2788
+    },
+    {
+      "epoch": 0.279,
+      "learning_rate": 1.0125660398833514e-05,
+      "loss": 0.5109,
+      "step": 2790
+    },
+    {
+      "epoch": 0.2792,
+      "learning_rate": 1.0132641133436333e-05,
+      "loss": 0.6767,
+      "step": 2792
+    },
+    {
+      "epoch": 0.2794,
+      "learning_rate": 1.013962180339144e-05,
+      "loss": 1.1295,
+      "step": 2794
+    },
+    {
+      "epoch": 0.2796,
+      "learning_rate": 1.0146602405296594e-05,
+      "loss": 0.4456,
+      "step": 2796
+    },
+    {
+      "epoch": 0.2798,
+      "learning_rate": 1.0153582935749533e-05,
+      "loss": 0.7029,
+      "step": 2798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.0160563391348e-05,
+      "loss": 0.5941,
+      "step": 2800
+    },
+    {
+      "epoch": 0.2802,
+      "learning_rate": 1.01675437686898e-05,
+      "loss": 0.3617,
+      "step": 2802
+    },
+    {
+      "epoch": 0.2804,
+      "learning_rate": 1.017452406437284e-05,
+      "loss": 0.7118,
+      "step": 2804
+    },
+    {
+      "epoch": 0.2806,
+      "learning_rate": 1.0181504274994952e-05,
+      "loss": 0.3687,
+      "step": 2806
+    },
+    {
+      "epoch": 0.2808,
+      "learning_rate": 1.0188484397154066e-05,
+      "loss": 0.981,
+      "step": 2808
+    },
+    {
+      "epoch": 0.281,
+      "learning_rate": 1.0195464427448212e-05,
+      "loss": 0.9095,
+      "step": 2810
+    },
+    {
+      "epoch": 0.2812,
+      "learning_rate": 1.0202444362475352e-05,
+      "loss": 0.9891,
+      "step": 2812
+    },
+    {
+      "epoch": 0.2814,
+      "learning_rate": 1.0209424198833571e-05,
+      "loss": 0.6184,
+      "step": 2814
+    },
+    {
+      "epoch": 0.2816,
+      "learning_rate": 1.021640393312096e-05,
+      "loss": 0.6959,
+      "step": 2816
+    },
+    {
+      "epoch": 0.2818,
+      "learning_rate": 1.0223383561935738e-05,
+      "loss": 0.7469,
+      "step": 2818
+    },
+    {
+      "epoch": 0.282,
+      "learning_rate": 1.0230363081876063e-05,
+      "loss": 0.3771,
+      "step": 2820
+    },
+    {
+      "epoch": 0.2822,
+      "learning_rate": 1.0237342489540218e-05,
+      "loss": 0.7681,
+      "step": 2822
+    },
+    {
+      "epoch": 0.2824,
+      "learning_rate": 1.0244321781526531e-05,
+      "loss": 0.4855,
+      "step": 2824
+    },
+    {
+      "epoch": 0.2826,
+      "learning_rate": 1.0251300954433374e-05,
+      "loss": 0.992,
+      "step": 2826
+    },
+    {
+      "epoch": 0.2828,
+      "learning_rate": 1.0258280004859184e-05,
+      "loss": 0.5122,
+      "step": 2828
+    },
+    {
+      "epoch": 0.283,
+      "learning_rate": 1.0265258929402458e-05,
+      "loss": 0.664,
+      "step": 2830
+    },
+    {
+      "epoch": 0.2832,
+      "learning_rate": 1.0272237724661751e-05,
+      "loss": 0.5409,
+      "step": 2832
+    },
+    {
+      "epoch": 0.2834,
+      "learning_rate": 1.0279216387235686e-05,
+      "loss": 0.7089,
+      "step": 2834
+    },
+    {
+      "epoch": 0.2836,
+      "learning_rate": 1.0286194913722945e-05,
+      "loss": 0.357,
+      "step": 2836
+    },
+    {
+      "epoch": 0.2838,
+      "learning_rate": 1.029317330072228e-05,
+      "loss": 0.7401,
+      "step": 2838
+    },
+    {
+      "epoch": 0.284,
+      "learning_rate": 1.0300151544832508e-05,
+      "loss": 0.5213,
+      "step": 2840
+    },
+    {
+      "epoch": 0.2842,
+      "learning_rate": 1.0307129642652523e-05,
+      "loss": 0.2034,
+      "step": 2842
+    },
+    {
+      "epoch": 0.2844,
+      "learning_rate": 1.0314107590781277e-05,
+      "loss": 0.5998,
+      "step": 2844
+    },
+    {
+      "epoch": 0.2846,
+      "learning_rate": 1.0321085385817811e-05,
+      "loss": 0.0021,
+      "step": 2846
+    },
+    {
+      "epoch": 0.2848,
+      "learning_rate": 1.0328063024361226e-05,
+      "loss": 0.7595,
+      "step": 2848
+    },
+    {
+      "epoch": 0.285,
+      "learning_rate": 1.033504050301071e-05,
+      "loss": 0.4529,
+      "step": 2850
+    },
+    {
+      "epoch": 0.2852,
+      "learning_rate": 1.0342017818365519e-05,
+      "loss": 0.2549,
+      "step": 2852
+    },
+    {
+      "epoch": 0.2854,
+      "learning_rate": 1.0348994967025004e-05,
+      "loss": 0.7015,
+      "step": 2854
+    },
+    {
+      "epoch": 0.2856,
+      "learning_rate": 1.0355971945588576e-05,
+      "loss": 0.5476,
+      "step": 2856
+    },
+    {
+      "epoch": 0.2858,
+      "learning_rate": 1.0362948750655768e-05,
+      "loss": 0.4541,
+      "step": 2858
+    },
+    {
+      "epoch": 0.286,
+      "learning_rate": 1.0369925378826111e-05,
+      "loss": 0.6803,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2862,
+      "learning_rate": 1.0376901826699337e-05,
+      "loss": 0.3441,
+      "step": 2862
+    },
+    {
+      "epoch": 0.2864,
+      "learning_rate": 1.0383878090875208e-05,
+      "loss": 1.3166,
+      "step": 2864
+    },
+    {
+      "epoch": 0.2866,
+      "learning_rate": 1.0390854167953526e-05,
+      "loss": 0.4949,
+      "step": 2866
+    },
+    {
+      "epoch": 0.2868,
+      "learning_rate": 1.0397830054534288e-05,
+      "loss": 0.943,
+      "step": 2868
+    },
+    {
+      "epoch": 0.287,
+      "learning_rate": 1.0404805747217532e-05,
+      "loss": 0.4313,
+      "step": 2870
+    },
+    {
+      "epoch": 0.2872,
+      "learning_rate": 1.0411781242603354e-05,
+      "loss": 0.8738,
+      "step": 2872
+    },
+    {
+      "epoch": 0.2874,
+      "learning_rate": 1.0418756537291984e-05,
+      "loss": 0.4531,
+      "step": 2874
+    },
+    {
+      "epoch": 0.2876,
+      "learning_rate": 1.0425731627883784e-05,
+      "loss": 1.0837,
+      "step": 2876
+    },
+    {
+      "epoch": 0.2878,
+      "learning_rate": 1.0432706510979175e-05,
+      "loss": 0.6087,
+      "step": 2878
+    },
+    {
+      "epoch": 0.288,
+      "learning_rate": 1.0439681183178636e-05,
+      "loss": 0.4407,
+      "step": 2880
+    },
+    {
+      "epoch": 0.2882,
+      "learning_rate": 1.0446655641082846e-05,
+      "loss": 0.6514,
+      "step": 2882
+    },
+    {
+      "epoch": 0.2884,
+      "learning_rate": 1.0453629881292542e-05,
+      "loss": 0.5113,
+      "step": 2884
+    },
+    {
+      "epoch": 0.2886,
+      "learning_rate": 1.0460603900408526e-05,
+      "loss": 0.2122,
+      "step": 2886
+    },
+    {
+      "epoch": 0.2888,
+      "learning_rate": 1.0467577695031745e-05,
+      "loss": 0.1334,
+      "step": 2888
+    },
+    {
+      "epoch": 0.289,
+      "learning_rate": 1.0474551261763312e-05,
+      "loss": 0.5581,
+      "step": 2890
+    },
+    {
+      "epoch": 0.2892,
+      "learning_rate": 1.0481524597204344e-05,
+      "loss": 1.2284,
+      "step": 2892
+    },
+    {
+      "epoch": 0.2894,
+      "learning_rate": 1.0488497697956134e-05,
+      "loss": 0.0548,
+      "step": 2894
+    },
+    {
+      "epoch": 0.2896,
+      "learning_rate": 1.0495470560620082e-05,
+      "loss": 0.4181,
+      "step": 2896
+    },
+    {
+      "epoch": 0.2898,
+      "learning_rate": 1.0502443181797696e-05,
+      "loss": 0.3446,
+      "step": 2898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.0509415558090608e-05,
+      "loss": 1.2107,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2902,
+      "learning_rate": 1.0516387686100549e-05,
+      "loss": 0.1991,
+      "step": 2902
+    },
+    {
+      "epoch": 0.2904,
+      "learning_rate": 1.0523359562429438e-05,
+      "loss": 0.459,
+      "step": 2904
+    },
+    {
+      "epoch": 0.2906,
+      "learning_rate": 1.0530331183679216e-05,
+      "loss": 0.5008,
+      "step": 2906
+    },
+    {
+      "epoch": 0.2908,
+      "learning_rate": 1.0537302546452019e-05,
+      "loss": 0.8078,
+      "step": 2908
+    },
+    {
+      "epoch": 0.291,
+      "learning_rate": 1.054427364735009e-05,
+      "loss": 0.4523,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2912,
+      "learning_rate": 1.0551244482975795e-05,
+      "loss": 0.6894,
+      "step": 2912
+    },
+    {
+      "epoch": 0.2914,
+      "learning_rate": 1.0558215049931634e-05,
+      "loss": 0.1781,
+      "step": 2914
+    },
+    {
+      "epoch": 0.2916,
+      "learning_rate": 1.056518534482024e-05,
+      "loss": 0.8507,
+      "step": 2916
+    },
+    {
+      "epoch": 0.2918,
+      "learning_rate": 1.0572155364244378e-05,
+      "loss": 0.6128,
+      "step": 2918
+    },
+    {
+      "epoch": 0.292,
+      "learning_rate": 1.0579125104806939e-05,
+      "loss": 1.1331,
+      "step": 2920
+    },
+    {
+      "epoch": 0.2922,
+      "learning_rate": 1.058609456311096e-05,
+      "loss": 0.9547,
+      "step": 2922
+    },
+    {
+      "epoch": 0.2924,
+      "learning_rate": 1.0593063735759612e-05,
+      "loss": 0.6112,
+      "step": 2924
+    },
+    {
+      "epoch": 0.2926,
+      "learning_rate": 1.0600032619356203e-05,
+      "loss": 0.1533,
+      "step": 2926
+    },
+    {
+      "epoch": 0.2928,
+      "learning_rate": 1.0607001210504184e-05,
+      "loss": 1.0945,
+      "step": 2928
+    },
+    {
+      "epoch": 0.293,
+      "learning_rate": 1.0613969505807167e-05,
+      "loss": 0.5445,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2932,
+      "learning_rate": 1.0620937501868835e-05,
+      "loss": 0.492,
+      "step": 2932
+    },
+    {
+      "epoch": 0.2934,
+      "learning_rate": 1.0627905195293127e-05,
+      "loss": 1.1081,
+      "step": 2934
+    },
+    {
+      "epoch": 0.2936,
+      "learning_rate": 1.0634872582684055e-05,
+      "loss": 0.5276,
+      "step": 2936
+    },
+    {
+      "epoch": 0.2938,
+      "learning_rate": 1.0641839660645795e-05,
+      "loss": 0.3586,
+      "step": 2938
+    },
+    {
+      "epoch": 0.294,
+      "learning_rate": 1.0648806425782687e-05,
+      "loss": 0.1624,
+      "step": 2940
+    },
+    {
+      "epoch": 0.2942,
+      "learning_rate": 1.0655772874699206e-05,
+      "loss": 0.6308,
+      "step": 2942
+    },
+    {
+      "epoch": 0.2944,
+      "learning_rate": 1.0662739004000012e-05,
+      "loss": 0.5563,
+      "step": 2944
+    },
+    {
+      "epoch": 0.2946,
+      "learning_rate": 1.066970481028984e-05,
+      "loss": 0.2035,
+      "step": 2946
+    },
+    {
+      "epoch": 0.2948,
+      "learning_rate": 1.0676670290173698e-05,
+      "loss": 0.3081,
+      "step": 2948
+    },
+    {
+      "epoch": 0.295,
+      "learning_rate": 1.0683635440256694e-05,
+      "loss": 0.5224,
+      "step": 2950
+    },
+    {
+      "epoch": 0.2952,
+      "learning_rate": 1.0690600257144048e-05,
+      "loss": 0.671,
+      "step": 2952
+    },
+    {
+      "epoch": 0.2954,
+      "learning_rate": 1.0697564737441242e-05,
+      "loss": 0.646,
+      "step": 2954
+    },
+    {
+      "epoch": 0.2956,
+      "learning_rate": 1.0704528877753874e-05,
+      "loss": 0.0106,
+      "step": 2956
+    },
+    {
+      "epoch": 0.2958,
+      "learning_rate": 1.0711492674687674e-05,
+      "loss": 0.5009,
+      "step": 2958
+    },
+    {
+      "epoch": 0.296,
+      "learning_rate": 1.071845612484857e-05,
+      "loss": 0.7766,
+      "step": 2960
+    },
+    {
+      "epoch": 0.2962,
+      "learning_rate": 1.0725419224842695e-05,
+      "loss": 1.0202,
+      "step": 2962
+    },
+    {
+      "epoch": 0.2964,
+      "learning_rate": 1.073238197127632e-05,
+      "loss": 0.1102,
+      "step": 2964
+    },
+    {
+      "epoch": 0.2966,
+      "learning_rate": 1.0739344360755855e-05,
+      "loss": 0.6383,
+      "step": 2966
+    },
+    {
+      "epoch": 0.2968,
+      "learning_rate": 1.0746306389887907e-05,
+      "loss": 0.4572,
+      "step": 2968
+    },
+    {
+      "epoch": 0.297,
+      "learning_rate": 1.0753268055279332e-05,
+      "loss": 0.9111,
+      "step": 2970
+    },
+    {
+      "epoch": 0.2972,
+      "learning_rate": 1.0760229353537036e-05,
+      "loss": 0.5823,
+      "step": 2972
+    },
+    {
+      "epoch": 0.2974,
+      "learning_rate": 1.0767190281268171e-05,
+      "loss": 0.6257,
+      "step": 2974
+    },
+    {
+      "epoch": 0.2976,
+      "learning_rate": 1.0774150835080119e-05,
+      "loss": 1.2132,
+      "step": 2976
+    },
+    {
+      "epoch": 0.2978,
+      "learning_rate": 1.0781111011580336e-05,
+      "loss": 1.035,
+      "step": 2978
+    },
+    {
+      "epoch": 0.298,
+      "learning_rate": 1.0788070807376536e-05,
+      "loss": 0.1538,
+      "step": 2980
+    },
+    {
+      "epoch": 0.2982,
+      "learning_rate": 1.07950302190766e-05,
+      "loss": 0.0219,
+      "step": 2982
+    },
+    {
+      "epoch": 0.2984,
+      "learning_rate": 1.0801989243288588e-05,
+      "loss": 0.0083,
+      "step": 2984
+    },
+    {
+      "epoch": 0.2986,
+      "learning_rate": 1.0808947876620766e-05,
+      "loss": 0.9021,
+      "step": 2986
+    },
+    {
+      "epoch": 0.2988,
+      "learning_rate": 1.0815906115681574e-05,
+      "loss": 0.6297,
+      "step": 2988
+    },
+    {
+      "epoch": 0.299,
+      "learning_rate": 1.0822863957079654e-05,
+      "loss": 1.4554,
+      "step": 2990
+    },
+    {
+      "epoch": 0.2992,
+      "learning_rate": 1.0829821397423839e-05,
+      "loss": 0.1735,
+      "step": 2992
+    },
+    {
+      "epoch": 0.2994,
+      "learning_rate": 1.0836778433323153e-05,
+      "loss": 0.6445,
+      "step": 2994
+    },
+    {
+      "epoch": 0.2996,
+      "learning_rate": 1.0843735061386824e-05,
+      "loss": 1.07,
+      "step": 2996
+    },
+    {
+      "epoch": 0.2998,
+      "learning_rate": 1.0850691278224277e-05,
+      "loss": 0.4537,
+      "step": 2998
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.0857647080445136e-05,
+      "loss": 0.1415,
+      "step": 3000
+    },
+    {
+      "epoch": 0.3002,
+      "learning_rate": 1.0864602464659227e-05,
+      "loss": 0.4913,
+      "step": 3002
+    },
+    {
+      "epoch": 0.3004,
+      "learning_rate": 1.0871557427476578e-05,
+      "loss": 0.7107,
+      "step": 3004
+    },
+    {
+      "epoch": 0.3006,
+      "learning_rate": 1.0878511965507428e-05,
+      "loss": 0.4472,
+      "step": 3006
+    },
+    {
+      "epoch": 0.3008,
+      "learning_rate": 1.0885466075362217e-05,
+      "loss": 0.8682,
+      "step": 3008
+    },
+    {
+      "epoch": 0.301,
+      "learning_rate": 1.0892419753651598e-05,
+      "loss": 0.4647,
+      "step": 3010
+    },
+    {
+      "epoch": 0.3012,
+      "learning_rate": 1.0899372996986432e-05,
+      "loss": 0.5412,
+      "step": 3012
+    },
+    {
+      "epoch": 0.3014,
+      "learning_rate": 1.0906325801977795e-05,
+      "loss": 0.5552,
+      "step": 3014
+    },
+    {
+      "epoch": 0.3016,
+      "learning_rate": 1.091327816523699e-05,
+      "loss": 0.1854,
+      "step": 3016
+    },
+    {
+      "epoch": 0.3018,
+      "learning_rate": 1.0920230083375465e-05,
+      "loss": 0.2317,
+      "step": 3018
+    },
+    {
+      "epoch": 0.302,
+      "learning_rate": 1.0927181553004993e-05,
+      "loss": 0.9439,
+      "step": 3020
+    },
+    {
+      "epoch": 0.3022,
+      "learning_rate": 1.0934132570737497e-05,
+      "loss": 0.5265,
+      "step": 3022
+    },
+    {
+      "epoch": 0.3024,
+      "learning_rate": 1.0941083133185136e-05,
+      "loss": 0.5584,
+      "step": 3024
+    },
+    {
+      "epoch": 0.3026,
+      "learning_rate": 1.0948033236960285e-05,
+      "loss": 0.3103,
+      "step": 3026
+    },
+    {
+      "epoch": 0.3028,
+      "learning_rate": 1.0954982878675554e-05,
+      "loss": 1.1428,
+      "step": 3028
+    },
+    {
+      "epoch": 0.303,
+      "learning_rate": 1.0961932054943785e-05,
+      "loss": 1.0685,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3032,
+      "learning_rate": 1.0968880762377982e-05,
+      "loss": 0.8028,
+      "step": 3032
+    },
+    {
+      "epoch": 0.3034,
+      "learning_rate": 1.0975828997591484e-05,
+      "loss": 0.9796,
+      "step": 3034
+    },
+    {
+      "epoch": 0.3036,
+      "learning_rate": 1.0982776757197804e-05,
+      "loss": 0.5251,
+      "step": 3036
+    },
+    {
+      "epoch": 0.3038,
+      "learning_rate": 1.098972403781064e-05,
+      "loss": 0.6307,
+      "step": 3038
+    },
+    {
+      "epoch": 0.304,
+      "learning_rate": 1.0996670836044017e-05,
+      "loss": 0.4911,
+      "step": 3040
+    },
+    {
+      "epoch": 0.3042,
+      "learning_rate": 1.1003617148512154e-05,
+      "loss": 0.9165,
+      "step": 3042
+    },
+    {
+      "epoch": 0.3044,
+      "learning_rate": 1.1010562971829468e-05,
+      "loss": 0.541,
+      "step": 3044
+    },
+    {
+      "epoch": 0.3046,
+      "learning_rate": 1.101750830261065e-05,
+      "loss": 0.4928,
+      "step": 3046
+    },
+    {
+      "epoch": 0.3048,
+      "learning_rate": 1.1024453137470678e-05,
+      "loss": 0.6768,
+      "step": 3048
+    },
+    {
+      "epoch": 0.305,
+      "learning_rate": 1.1031397473024676e-05,
+      "loss": 0.6408,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3052,
+      "learning_rate": 1.1038341305888076e-05,
+      "loss": 1.7254,
+      "step": 3052
+    },
+    {
+      "epoch": 0.3054,
+      "learning_rate": 1.104528463267652e-05,
+      "loss": 0.476,
+      "step": 3054
+    },
+    {
+      "epoch": 0.3056,
+      "learning_rate": 1.1052227450005965e-05,
+      "loss": 0.3303,
+      "step": 3056
+    },
+    {
+      "epoch": 0.3058,
+      "learning_rate": 1.1059169754492518e-05,
+      "loss": 0.4785,
+      "step": 3058
+    },
+    {
+      "epoch": 0.306,
+      "learning_rate": 1.1066111542752582e-05,
+      "loss": 0.4974,
+      "step": 3060
+    },
+    {
+      "epoch": 0.3062,
+      "learning_rate": 1.1073052811402867e-05,
+      "loss": 0.8785,
+      "step": 3062
+    },
+    {
+      "epoch": 0.3064,
+      "learning_rate": 1.1079993557060228e-05,
+      "loss": 1.0842,
+      "step": 3064
+    },
+    {
+      "epoch": 0.3066,
+      "learning_rate": 1.108693377634185e-05,
+      "loss": 0.5439,
+      "step": 3066
+    },
+    {
+      "epoch": 0.3068,
+      "learning_rate": 1.1093873465865155e-05,
+      "loss": 0.7873,
+      "step": 3068
+    },
+    {
+      "epoch": 0.307,
+      "learning_rate": 1.1100812622247821e-05,
+      "loss": 0.5531,
+      "step": 3070
+    },
+    {
+      "epoch": 0.3072,
+      "learning_rate": 1.1107751242107785e-05,
+      "loss": 0.2068,
+      "step": 3072
+    },
+    {
+      "epoch": 0.3074,
+      "learning_rate": 1.1114689322063252e-05,
+      "loss": 0.4647,
+      "step": 3074
+    },
+    {
+      "epoch": 0.3076,
+      "learning_rate": 1.1121626858732686e-05,
+      "loss": 0.5867,
+      "step": 3076
+    },
+    {
+      "epoch": 0.3078,
+      "learning_rate": 1.1128563848734815e-05,
+      "loss": 1.0616,
+      "step": 3078
+    },
+    {
+      "epoch": 0.308,
+      "learning_rate": 1.1135500288688632e-05,
+      "loss": 0.6289,
+      "step": 3080
+    },
+    {
+      "epoch": 0.3082,
+      "learning_rate": 1.1142436175213404e-05,
+      "loss": 0.5774,
+      "step": 3082
+    },
+    {
+      "epoch": 0.3084,
+      "learning_rate": 1.1149371504928662e-05,
+      "loss": 0.2121,
+      "step": 3084
+    },
+    {
+      "epoch": 0.3086,
+      "learning_rate": 1.1156306274454211e-05,
+      "loss": 0.858,
+      "step": 3086
+    },
+    {
+      "epoch": 0.3088,
+      "learning_rate": 1.116324048041013e-05,
+      "loss": 0.4409,
+      "step": 3088
+    },
+    {
+      "epoch": 0.309,
+      "learning_rate": 1.117017411941677e-05,
+      "loss": 0.9672,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3092,
+      "learning_rate": 1.1177107188094758e-05,
+      "loss": 1.1701,
+      "step": 3092
+    },
+    {
+      "epoch": 0.3094,
+      "learning_rate": 1.1184039683065002e-05,
+      "loss": 1.0452,
+      "step": 3094
+    },
+    {
+      "epoch": 0.3096,
+      "learning_rate": 1.1190971600948691e-05,
+      "loss": 0.3788,
+      "step": 3096
+    },
+    {
+      "epoch": 0.3098,
+      "learning_rate": 1.1197902938367289e-05,
+      "loss": 1.7128,
+      "step": 3098
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.1204833691942545e-05,
+      "loss": 0.8033,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3102,
+      "learning_rate": 1.1211763858296516e-05,
+      "loss": 0.1173,
+      "step": 3102
+    },
+    {
+      "epoch": 0.3104,
+      "learning_rate": 1.1218693434051467e-05,
+      "loss": 0.8495,
+      "step": 3104
+    },
+    {
+      "epoch": 0.3106,
+      "learning_rate": 1.122562241583006e-05,
+      "loss": 0.3439,
+      "step": 3106
+    },
+    {
+      "epoch": 0.3108,
+      "learning_rate": 1.1232550800255195e-05,
+      "loss": 0.197,
+      "step": 3108
+    },
+    {
+      "epoch": 0.311,
+      "learning_rate": 1.1239478583950007e-05,
+      "loss": 1.01,
+      "step": 3110
+    },
+    {
+      "epoch": 0.3112,
+      "learning_rate": 1.1246405763538033e-05,
+      "loss": 0.4684,
+      "step": 3112
+    },
+    {
+      "epoch": 0.3114,
+      "learning_rate": 1.1253332335643033e-05,
+      "loss": 1.1582,
+      "step": 3114
+    },
+    {
+      "epoch": 0.3116,
+      "learning_rate": 1.1260258296889093e-05,
+      "loss": 0.484,
+      "step": 3116
+    },
+    {
+      "epoch": 0.3118,
+      "learning_rate": 1.1267183643900534e-05,
+      "loss": 0.3266,
+      "step": 3118
+    },
+    {
+      "epoch": 0.312,
+      "learning_rate": 1.127410837330208e-05,
+      "loss": 0.7117,
+      "step": 3120
+    },
+    {
+      "epoch": 0.3122,
+      "learning_rate": 1.1281032481718701e-05,
+      "loss": 1.0194,
+      "step": 3122
+    },
+    {
+      "epoch": 0.3124,
+      "learning_rate": 1.1287955965775633e-05,
+      "loss": 0.6905,
+      "step": 3124
+    },
+    {
+      "epoch": 0.3126,
+      "learning_rate": 1.1294878822098456e-05,
+      "loss": 1.0046,
+      "step": 3126
+    },
+    {
+      "epoch": 0.3128,
+      "learning_rate": 1.1301801047313107e-05,
+      "loss": 0.4067,
+      "step": 3128
+    },
+    {
+      "epoch": 0.313,
+      "learning_rate": 1.1308722638045725e-05,
+      "loss": 0.6471,
+      "step": 3130
+    },
+    {
+      "epoch": 0.3132,
+      "learning_rate": 1.1315643590922812e-05,
+      "loss": 0.9285,
+      "step": 3132
+    },
+    {
+      "epoch": 0.3134,
+      "learning_rate": 1.1322563902571227e-05,
+      "loss": 0.1636,
+      "step": 3134
+    },
+    {
+      "epoch": 0.3136,
+      "learning_rate": 1.1329483569618047e-05,
+      "loss": 0.5744,
+      "step": 3136
+    },
+    {
+      "epoch": 0.3138,
+      "learning_rate": 1.1336402588690725e-05,
+      "loss": 1.0281,
+      "step": 3138
+    },
+    {
+      "epoch": 0.314,
+      "learning_rate": 1.1343320956416998e-05,
+      "loss": 0.3851,
+      "step": 3140
+    },
+    {
+      "epoch": 0.3142,
+      "learning_rate": 1.1350238669424993e-05,
+      "loss": 0.3526,
+      "step": 3142
+    },
+    {
+      "epoch": 0.3144,
+      "learning_rate": 1.1357155724343046e-05,
+      "loss": 0.517,
+      "step": 3144
+    },
+    {
+      "epoch": 0.3146,
+      "learning_rate": 1.1364072117799864e-05,
+      "loss": 1.0819,
+      "step": 3146
+    },
+    {
+      "epoch": 0.3148,
+      "learning_rate": 1.1370987846424542e-05,
+      "loss": 0.987,
+      "step": 3148
+    },
+    {
+      "epoch": 0.315,
+      "learning_rate": 1.137790290684638e-05,
+      "loss": 0.6563,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3152,
+      "learning_rate": 1.1384817295695082e-05,
+      "loss": 1.0395,
+      "step": 3152
+    },
+    {
+      "epoch": 0.3154,
+      "learning_rate": 1.1391731009600652e-05,
+      "loss": 0.4134,
+      "step": 3154
+    },
+    {
+      "epoch": 0.3156,
+      "learning_rate": 1.139864404519344e-05,
+      "loss": 0.3597,
+      "step": 3156
+    },
+    {
+      "epoch": 0.3158,
+      "learning_rate": 1.1405556399104108e-05,
+      "loss": 0.2046,
+      "step": 3158
+    },
+    {
+      "epoch": 0.316,
+      "learning_rate": 1.1412468067963662e-05,
+      "loss": 1.0913,
+      "step": 3160
+    },
+    {
+      "epoch": 0.3162,
+      "learning_rate": 1.141937904840344e-05,
+      "loss": 0.5704,
+      "step": 3162
+    },
+    {
+      "epoch": 0.3164,
+      "learning_rate": 1.1426289337055114e-05,
+      "loss": 0.0568,
+      "step": 3164
+    },
+    {
+      "epoch": 0.3166,
+      "learning_rate": 1.143319893055069e-05,
+      "loss": 1.1563,
+      "step": 3166
+    },
+    {
+      "epoch": 0.3168,
+      "learning_rate": 1.1440107825522517e-05,
+      "loss": 0.8895,
+      "step": 3168
+    },
+    {
+      "epoch": 0.317,
+      "learning_rate": 1.1447016018603286e-05,
+      "loss": 1.0615,
+      "step": 3170
+    },
+    {
+      "epoch": 0.3172,
+      "learning_rate": 1.1453923506426025e-05,
+      "loss": 1.2297,
+      "step": 3172
+    },
+    {
+      "epoch": 0.3174,
+      "learning_rate": 1.1460830285624112e-05,
+      "loss": 0.8373,
+      "step": 3174
+    },
+    {
+      "epoch": 0.3176,
+      "learning_rate": 1.146773635283126e-05,
+      "loss": 0.5709,
+      "step": 3176
+    },
+    {
+      "epoch": 0.3178,
+      "learning_rate": 1.1474641704681541e-05,
+      "loss": 1.0758,
+      "step": 3178
+    },
+    {
+      "epoch": 0.318,
+      "learning_rate": 1.1481546337809373e-05,
+      "loss": 0.6182,
+      "step": 3180
+    },
+    {
+      "epoch": 0.3182,
+      "learning_rate": 1.1488450248849515e-05,
+      "loss": 1.2661,
+      "step": 3182
+    },
+    {
+      "epoch": 0.3184,
+      "learning_rate": 1.1495353434437088e-05,
+      "loss": 1.1646,
+      "step": 3184
+    },
+    {
+      "epoch": 0.3186,
+      "learning_rate": 1.150225589120756e-05,
+      "loss": 0.746,
+      "step": 3186
+    },
+    {
+      "epoch": 0.3188,
+      "learning_rate": 1.1509157615796782e-05,
+      "loss": 1.1511,
+      "step": 3188
+    },
+    {
+      "epoch": 0.319,
+      "learning_rate": 1.1516058604840881e-05,
+      "loss": 0.6721,
+      "step": 3190
+    },
+    {
+      "epoch": 0.3192,
+      "learning_rate": 1.1522958854976447e-05,
+      "loss": 0.4391,
+      "step": 3192
+    },
+    {
+      "epoch": 0.3194,
+      "learning_rate": 1.1529858362840388e-05,
+      "loss": 0.8663,
+      "step": 3194
+    },
+    {
+      "epoch": 0.3196,
+      "learning_rate": 1.153675712506993e-05,
+      "loss": 0.4217,
+      "step": 3196
+    },
+    {
+      "epoch": 0.3198,
+      "learning_rate": 1.15436551383027e-05,
+      "loss": 0.6212,
+      "step": 3198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.1550552399176727e-05,
+      "loss": 0.4454,
+      "step": 3200
+    },
+    {
+      "epoch": 0.3202,
+      "learning_rate": 1.1557448904330366e-05,
+      "loss": 0.8826,
+      "step": 3202
+    },
+    {
+      "epoch": 0.3204,
+      "learning_rate": 1.1564344650402298e-05,
+      "loss": 0.7657,
+      "step": 3204
+    },
+    {
+      "epoch": 0.3206,
+      "learning_rate": 1.1571239634031666e-05,
+      "loss": 0.721,
+      "step": 3206
+    },
+    {
+      "epoch": 0.3208,
+      "learning_rate": 1.157813385185794e-05,
+      "loss": 1.0458,
+      "step": 3208
+    },
+    {
+      "epoch": 0.321,
+      "learning_rate": 1.158502730052093e-05,
+      "loss": 0.8407,
+      "step": 3210
+    },
+    {
+      "epoch": 0.3212,
+      "learning_rate": 1.1591919976660852e-05,
+      "loss": 0.5788,
+      "step": 3212
+    },
+    {
+      "epoch": 0.3214,
+      "learning_rate": 1.1598811876918352e-05,
+      "loss": 0.1289,
+      "step": 3214
+    },
+    {
+      "epoch": 0.3216,
+      "learning_rate": 1.1605702997934347e-05,
+      "loss": 0.9198,
+      "step": 3216
+    },
+    {
+      "epoch": 0.3218,
+      "learning_rate": 1.161259333635019e-05,
+      "loss": 0.3462,
+      "step": 3218
+    },
+    {
+      "epoch": 0.322,
+      "learning_rate": 1.1619482888807665e-05,
+      "loss": 0.1959,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3222,
+      "learning_rate": 1.1626371651948839e-05,
+      "loss": 0.6212,
+      "step": 3222
+    },
+    {
+      "epoch": 0.3224,
+      "learning_rate": 1.1633259622416224e-05,
+      "loss": 0.8094,
+      "step": 3224
+    },
+    {
+      "epoch": 0.3226,
+      "learning_rate": 1.1640146796852711e-05,
+      "loss": 0.1725,
+      "step": 3226
+    },
+    {
+      "epoch": 0.3228,
+      "learning_rate": 1.1647033171901573e-05,
+      "loss": 0.6775,
+      "step": 3228
+    },
+    {
+      "epoch": 0.323,
+      "learning_rate": 1.1653918744206478e-05,
+      "loss": 2.0834,
+      "step": 3230
+    },
+    {
+      "epoch": 0.3232,
+      "learning_rate": 1.1660803510411478e-05,
+      "loss": 0.1546,
+      "step": 3232
+    },
+    {
+      "epoch": 0.3234,
+      "learning_rate": 1.1667687467161021e-05,
+      "loss": 0.5176,
+      "step": 3234
+    },
+    {
+      "epoch": 0.3236,
+      "learning_rate": 1.1674570611099953e-05,
+      "loss": 0.5826,
+      "step": 3236
+    },
+    {
+      "epoch": 0.3238,
+      "learning_rate": 1.1681452938873515e-05,
+      "loss": 0.5643,
+      "step": 3238
+    },
+    {
+      "epoch": 0.324,
+      "learning_rate": 1.1688334447127337e-05,
+      "loss": 0.4974,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3242,
+      "learning_rate": 1.169521513250746e-05,
+      "loss": 0.3656,
+      "step": 3242
+    },
+    {
+      "epoch": 0.3244,
+      "learning_rate": 1.1702094991660323e-05,
+      "loss": 0.6348,
+      "step": 3244
+    },
+    {
+      "epoch": 0.3246,
+      "learning_rate": 1.1708974021232763e-05,
+      "loss": 0.0002,
+      "step": 3246
+    },
+    {
+      "epoch": 0.3248,
+      "learning_rate": 1.1715852217872025e-05,
+      "loss": 0.002,
+      "step": 3248
+    },
+    {
+      "epoch": 0.325,
+      "learning_rate": 1.1722729578225762e-05,
+      "loss": 0.4829,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3252,
+      "learning_rate": 1.172960609894203e-05,
+      "loss": 0.4845,
+      "step": 3252
+    },
+    {
+      "epoch": 0.3254,
+      "learning_rate": 1.1736481776669297e-05,
+      "loss": 0.6985,
+      "step": 3254
+    },
+    {
+      "epoch": 0.3256,
+      "learning_rate": 1.1743356608056441e-05,
+      "loss": 0.5013,
+      "step": 3256
+    },
+    {
+      "epoch": 0.3258,
+      "learning_rate": 1.1750230589752753e-05,
+      "loss": 0.8937,
+      "step": 3258
+    },
+    {
+      "epoch": 0.326,
+      "learning_rate": 1.1757103718407938e-05,
+      "loss": 1.17,
+      "step": 3260
+    },
+    {
+      "epoch": 0.3262,
+      "learning_rate": 1.1763975990672116e-05,
+      "loss": 0.1688,
+      "step": 3262
+    },
+    {
+      "epoch": 0.3264,
+      "learning_rate": 1.1770847403195826e-05,
+      "loss": 0.3461,
+      "step": 3264
+    },
+    {
+      "epoch": 0.3266,
+      "learning_rate": 1.1777717952630023e-05,
+      "loss": 0.9486,
+      "step": 3266
+    },
+    {
+      "epoch": 0.3268,
+      "learning_rate": 1.1784587635626085e-05,
+      "loss": 0.6991,
+      "step": 3268
+    },
+    {
+      "epoch": 0.327,
+      "learning_rate": 1.1791456448835815e-05,
+      "loss": 0.1506,
+      "step": 3270
+    },
+    {
+      "epoch": 0.3272,
+      "learning_rate": 1.1798324388911433e-05,
+      "loss": 0.7106,
+      "step": 3272
+    },
+    {
+      "epoch": 0.3274,
+      "learning_rate": 1.180519145250561e-05,
+      "loss": 0.826,
+      "step": 3274
+    },
+    {
+      "epoch": 0.3276,
+      "learning_rate": 1.1812057636271364e-05,
+      "loss": 1.0064,
+      "step": 3276
+    },
+    {
+      "epoch": 0.3278,
+      "learning_rate": 1.1818922936862258e-05,
+      "loss": 1.0477,
+      "step": 3278
+    },
+    {
+      "epoch": 0.328,
+      "learning_rate": 1.1825787350932227e-05,
+      "loss": 0.507,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3282,
+      "learning_rate": 1.1832650875135606e-05,
+      "loss": 0.2368,
+      "step": 3282
+    },
+    {
+      "epoch": 0.3284,
+      "learning_rate": 1.183951350612719e-05,
+      "loss": 0.7809,
+      "step": 3284
+    },
+    {
+      "epoch": 0.3286,
+      "learning_rate": 1.1846375240562274e-05,
+      "loss": 0.7872,
+      "step": 3286
+    },
+    {
+      "epoch": 0.3288,
+      "learning_rate": 1.185323607509648e-05,
+      "loss": 0.107,
+      "step": 3288
+    },
+    {
+      "epoch": 0.329,
+      "learning_rate": 1.1860096006385918e-05,
+      "loss": 0.2559,
+      "step": 3290
+    },
+    {
+      "epoch": 0.3292,
+      "learning_rate": 1.1866955031087177e-05,
+      "loss": 0.5556,
+      "step": 3292
+    },
+    {
+      "epoch": 0.3294,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 0.4961,
+      "step": 3294
+    },
+    {
+      "epoch": 0.3296,
+      "learning_rate": 1.1880670347353542e-05,
+      "loss": 0.4347,
+      "step": 3296
+    },
+    {
+      "epoch": 0.3298,
+      "learning_rate": 1.1887526632233937e-05,
+      "loss": 0.6572,
+      "step": 3298
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.1894381997156817e-05,
+      "loss": 1.326,
+      "step": 3300
+    },
+    {
+      "epoch": 0.3302,
+      "learning_rate": 1.1901236438780906e-05,
+      "loss": 0.1746,
+      "step": 3302
+    },
+    {
+      "epoch": 0.3304,
+      "learning_rate": 1.1908089953765431e-05,
+      "loss": 0.0177,
+      "step": 3304
+    },
+    {
+      "epoch": 0.3306,
+      "learning_rate": 1.191494253877013e-05,
+      "loss": 0.7133,
+      "step": 3306
+    },
+    {
+      "epoch": 0.3308,
+      "learning_rate": 1.1921794190455082e-05,
+      "loss": 0.0358,
+      "step": 3308
+    },
+    {
+      "epoch": 0.331,
+      "learning_rate": 1.192864490548089e-05,
+      "loss": 0.8192,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3312,
+      "learning_rate": 1.1935494680508604e-05,
+      "loss": 1.3504,
+      "step": 3312
+    },
+    {
+      "epoch": 0.3314,
+      "learning_rate": 1.1942343512199719e-05,
+      "loss": 0.1525,
+      "step": 3314
+    },
+    {
+      "epoch": 0.3316,
+      "learning_rate": 1.1949191397216205e-05,
+      "loss": 0.6582,
+      "step": 3316
+    },
+    {
+      "epoch": 0.3318,
+      "learning_rate": 1.195603833222048e-05,
+      "loss": 0.6181,
+      "step": 3318
+    },
+    {
+      "epoch": 0.332,
+      "learning_rate": 1.1962884313875436e-05,
+      "loss": 0.7234,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3322,
+      "learning_rate": 1.1969729338844422e-05,
+      "loss": 0.0617,
+      "step": 3322
+    },
+    {
+      "epoch": 0.3324,
+      "learning_rate": 1.197657340379126e-05,
+      "loss": 0.5637,
+      "step": 3324
+    },
+    {
+      "epoch": 0.3326,
+      "learning_rate": 1.198341650538023e-05,
+      "loss": 0.4548,
+      "step": 3326
+    },
+    {
+      "epoch": 0.3328,
+      "learning_rate": 1.1990258640276089e-05,
+      "loss": 0.7673,
+      "step": 3328
+    },
+    {
+      "epoch": 0.333,
+      "learning_rate": 1.1997099805144066e-05,
+      "loss": 0.8187,
+      "step": 3330
+    },
+    {
+      "epoch": 0.3332,
+      "learning_rate": 1.2003939996649857e-05,
+      "loss": 1.0123,
+      "step": 3332
+    },
+    {
+      "epoch": 0.3334,
+      "learning_rate": 1.2010779211459642e-05,
+      "loss": 0.7339,
+      "step": 3334
+    },
+    {
+      "epoch": 0.3336,
+      "learning_rate": 1.2017617446240063e-05,
+      "loss": 0.5736,
+      "step": 3336
+    },
+    {
+      "epoch": 0.3338,
+      "learning_rate": 1.2024454697658254e-05,
+      "loss": 0.6759,
+      "step": 3338
+    },
+    {
+      "epoch": 0.334,
+      "learning_rate": 1.2031290962381817e-05,
+      "loss": 0.3818,
+      "step": 3340
+    },
+    {
+      "epoch": 0.3342,
+      "learning_rate": 1.203812623707884e-05,
+      "loss": 0.4755,
+      "step": 3342
+    },
+    {
+      "epoch": 0.3344,
+      "learning_rate": 1.2044960518417896e-05,
+      "loss": 0.4232,
+      "step": 3344
+    },
+    {
+      "epoch": 0.3346,
+      "learning_rate": 1.2051793803068054e-05,
+      "loss": 0.5189,
+      "step": 3346
+    },
+    {
+      "epoch": 0.3348,
+      "learning_rate": 1.2058626087698806e-05,
+      "loss": 0.0822,
+      "step": 3348
+    },
+    {
+      "epoch": 0.335,
+      "learning_rate": 1.2065457368980227e-05,
+      "loss": 0.2801,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3352,
+      "learning_rate": 1.2072287643582817e-05,
+      "loss": 0.5473,
+      "step": 3352
+    },
+    {
+      "epoch": 0.3354,
+      "learning_rate": 1.20791169081776e-05,
+      "loss": 0.5232,
+      "step": 3354
+    },
+    {
+      "epoch": 0.3356,
+      "learning_rate": 1.2085945159436028e-05,
+      "loss": 0.5649,
+      "step": 3356
+    },
+    {
+      "epoch": 0.3358,
+      "learning_rate": 1.2092772394030141e-05,
+      "loss": 0.9148,
+      "step": 3358
+    },
+    {
+      "epoch": 0.336,
+      "learning_rate": 1.209959860863243e-05,
+      "loss": 1.0269,
+      "step": 3360
+    },
+    {
+      "epoch": 0.3362,
+      "learning_rate": 1.210642379991583e-05,
+      "loss": 1.3902,
+      "step": 3362
+    },
+    {
+      "epoch": 0.3364,
+      "learning_rate": 1.2113247964553876e-05,
+      "loss": 0.7041,
+      "step": 3364
+    },
+    {
+      "epoch": 0.3366,
+      "learning_rate": 1.2120071099220552e-05,
+      "loss": 0.7876,
+      "step": 3366
+    },
+    {
+      "epoch": 0.3368,
+      "learning_rate": 1.2126893200590312e-05,
+      "loss": 0.9701,
+      "step": 3368
+    },
+    {
+      "epoch": 0.337,
+      "learning_rate": 1.2133714265338148e-05,
+      "loss": 0.7745,
+      "step": 3370
+    },
+    {
+      "epoch": 0.3372,
+      "learning_rate": 1.2140534290139604e-05,
+      "loss": 1.3029,
+      "step": 3372
+    },
+    {
+      "epoch": 0.3374,
+      "learning_rate": 1.2147353271670637e-05,
+      "loss": 0.8906,
+      "step": 3374
+    },
+    {
+      "epoch": 0.3376,
+      "learning_rate": 1.2154171206607752e-05,
+      "loss": 0.6124,
+      "step": 3376
+    },
+    {
+      "epoch": 0.3378,
+      "learning_rate": 1.2160988091628006e-05,
+      "loss": 0.3282,
+      "step": 3378
+    },
+    {
+      "epoch": 0.338,
+      "learning_rate": 1.2167803923408936e-05,
+      "loss": 0.9629,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3382,
+      "learning_rate": 1.217461869862855e-05,
+      "loss": 0.6416,
+      "step": 3382
+    },
+    {
+      "epoch": 0.3384,
+      "learning_rate": 1.218143241396541e-05,
+      "loss": 0.752,
+      "step": 3384
+    },
+    {
+      "epoch": 0.3386,
+      "learning_rate": 1.2188245066098647e-05,
+      "loss": 0.6888,
+      "step": 3386
+    },
+    {
+      "epoch": 0.3388,
+      "learning_rate": 1.2195056651707804e-05,
+      "loss": 0.6999,
+      "step": 3388
+    },
+    {
+      "epoch": 0.339,
+      "learning_rate": 1.2201867167473015e-05,
+      "loss": 0.4772,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3392,
+      "learning_rate": 1.220867661007492e-05,
+      "loss": 1.2787,
+      "step": 3392
+    },
+    {
+      "epoch": 0.3394,
+      "learning_rate": 1.2215484976194673e-05,
+      "loss": 1.2183,
+      "step": 3394
+    },
+    {
+      "epoch": 0.3396,
+      "learning_rate": 1.2222292262513963e-05,
+      "loss": 0.7538,
+      "step": 3396
+    },
+    {
+      "epoch": 0.3398,
+      "learning_rate": 1.2229098465715002e-05,
+      "loss": 0.3329,
+      "step": 3398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.2235903582480528e-05,
+      "loss": 0.7288,
+      "step": 3400
+    },
+    {
+      "epoch": 0.3402,
+      "learning_rate": 1.2242707609493809e-05,
+      "loss": 1.0537,
+      "step": 3402
+    },
+    {
+      "epoch": 0.3404,
+      "learning_rate": 1.2249510543438647e-05,
+      "loss": 0.5151,
+      "step": 3404
+    },
+    {
+      "epoch": 0.3406,
+      "learning_rate": 1.2256312380999373e-05,
+      "loss": 0.5985,
+      "step": 3406
+    },
+    {
+      "epoch": 0.3408,
+      "learning_rate": 1.2263113118860857e-05,
+      "loss": 1.0554,
+      "step": 3408
+    },
+    {
+      "epoch": 0.341,
+      "learning_rate": 1.2269912753708496e-05,
+      "loss": 0.3979,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3412,
+      "learning_rate": 1.2276711282228234e-05,
+      "loss": 0.9858,
+      "step": 3412
+    },
+    {
+      "epoch": 0.3414,
+      "learning_rate": 1.2283508701106552e-05,
+      "loss": 0.5532,
+      "step": 3414
+    },
+    {
+      "epoch": 0.3416,
+      "learning_rate": 1.229030500703047e-05,
+      "loss": 1.0076,
+      "step": 3416
+    },
+    {
+      "epoch": 0.3418,
+      "learning_rate": 1.229710019668755e-05,
+      "loss": 1.3667,
+      "step": 3418
+    },
+    {
+      "epoch": 0.342,
+      "learning_rate": 1.2303894266765901e-05,
+      "loss": 0.2613,
+      "step": 3420
+    },
+    {
+      "epoch": 0.3422,
+      "learning_rate": 1.2310687213954173e-05,
+      "loss": 0.7111,
+      "step": 3422
+    },
+    {
+      "epoch": 0.3424,
+      "learning_rate": 1.2317479034941567e-05,
+      "loss": 0.4409,
+      "step": 3424
+    },
+    {
+      "epoch": 0.3426,
+      "learning_rate": 1.232426972641785e-05,
+      "loss": 0.057,
+      "step": 3426
+    },
+    {
+      "epoch": 0.3428,
+      "learning_rate": 1.2331059285073271e-05,
+      "loss": 0.6364,
+      "step": 3428
+    },
+    {
+      "epoch": 0.343,
+      "learning_rate": 1.233784770759873e-05,
+      "loss": 0.8845,
+      "step": 3430
+    },
+    {
+      "epoch": 0.3432,
+      "learning_rate": 1.2344634990685634e-05,
+      "loss": 0.8632,
+      "step": 3432
+    },
+    {
+      "epoch": 0.3434,
+      "learning_rate": 1.2351421131025891e-05,
+      "loss": 0.9902,
+      "step": 3434
+    },
+    {
+      "epoch": 0.3436,
+      "learning_rate": 1.2358206125312073e-05,
+      "loss": 0.6162,
+      "step": 3436
+    },
+    {
+      "epoch": 0.3438,
+      "learning_rate": 1.2364989970237238e-05,
+      "loss": 0.5895,
+      "step": 3438
+    },
+    {
+      "epoch": 0.344,
+      "learning_rate": 1.2371772662495038e-05,
+      "loss": 0.1641,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3442,
+      "learning_rate": 1.237855419877962e-05,
+      "loss": 1.3255,
+      "step": 3442
+    },
+    {
+      "epoch": 0.3444,
+      "learning_rate": 1.2385334575785799e-05,
+      "loss": 1.754,
+      "step": 3444
+    },
+    {
+      "epoch": 0.3446,
+      "learning_rate": 1.23921137902089e-05,
+      "loss": 0.4455,
+      "step": 3446
+    },
+    {
+      "epoch": 0.3448,
+      "learning_rate": 1.2398891838744765e-05,
+      "loss": 0.4688,
+      "step": 3448
+    },
+    {
+      "epoch": 0.345,
+      "learning_rate": 1.2405668718089906e-05,
+      "loss": 0.0629,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3452,
+      "learning_rate": 1.2412444424941354e-05,
+      "loss": 0.564,
+      "step": 3452
+    },
+    {
+      "epoch": 0.3454,
+      "learning_rate": 1.241921895599668e-05,
+      "loss": 0.3714,
+      "step": 3454
+    },
+    {
+      "epoch": 0.3456,
+      "learning_rate": 1.242599230795406e-05,
+      "loss": 0.6514,
+      "step": 3456
+    },
+    {
+      "epoch": 0.3458,
+      "learning_rate": 1.2432764477512295e-05,
+      "loss": 0.5811,
+      "step": 3458
+    },
+    {
+      "epoch": 0.346,
+      "learning_rate": 1.243953546137066e-05,
+      "loss": 1.1327,
+      "step": 3460
+    },
+    {
+      "epoch": 0.3462,
+      "learning_rate": 1.2446305256229076e-05,
+      "loss": 0.7373,
+      "step": 3462
+    },
+    {
+      "epoch": 0.3464,
+      "learning_rate": 1.2453073858788012e-05,
+      "loss": 0.4648,
+      "step": 3464
+    },
+    {
+      "epoch": 0.3466,
+      "learning_rate": 1.2459841265748582e-05,
+      "loss": 0.6596,
+      "step": 3466
+    },
+    {
+      "epoch": 0.3468,
+      "learning_rate": 1.2466607473812386e-05,
+      "loss": 0.6273,
+      "step": 3468
+    },
+    {
+      "epoch": 0.347,
+      "learning_rate": 1.2473372479681653e-05,
+      "loss": 0.5848,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3472,
+      "learning_rate": 1.2480136280059256e-05,
+      "loss": 0.0432,
+      "step": 3472
+    },
+    {
+      "epoch": 0.3474,
+      "learning_rate": 1.2486898871648547e-05,
+      "loss": 0.2211,
+      "step": 3474
+    },
+    {
+      "epoch": 0.3476,
+      "learning_rate": 1.249366025115354e-05,
+      "loss": 0.6907,
+      "step": 3476
+    },
+    {
+      "epoch": 0.3478,
+      "learning_rate": 1.2500420415278822e-05,
+      "loss": 0.456,
+      "step": 3478
+    },
+    {
+      "epoch": 0.348,
+      "learning_rate": 1.2507179360729564e-05,
+      "loss": 0.7353,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3482,
+      "learning_rate": 1.2513937084211546e-05,
+      "loss": 0.7064,
+      "step": 3482
+    },
+    {
+      "epoch": 0.3484,
+      "learning_rate": 1.2520693582431133e-05,
+      "loss": 0.3737,
+      "step": 3484
+    },
+    {
+      "epoch": 0.3486,
+      "learning_rate": 1.2527448852095292e-05,
+      "loss": 0.5242,
+      "step": 3486
+    },
+    {
+      "epoch": 0.3488,
+      "learning_rate": 1.253420288991158e-05,
+      "loss": 0.9198,
+      "step": 3488
+    },
+    {
+      "epoch": 0.349,
+      "learning_rate": 1.2540955692588167e-05,
+      "loss": 0.7628,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3492,
+      "learning_rate": 1.254770725683382e-05,
+      "loss": 0.4737,
+      "step": 3492
+    },
+    {
+      "epoch": 0.3494,
+      "learning_rate": 1.2554457579357902e-05,
+      "loss": 1.1655,
+      "step": 3494
+    },
+    {
+      "epoch": 0.3496,
+      "learning_rate": 1.2561206656870392e-05,
+      "loss": 0.4817,
+      "step": 3496
+    },
+    {
+      "epoch": 0.3498,
+      "learning_rate": 1.2567954486081873e-05,
+      "loss": 0.4389,
+      "step": 3498
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.2574701063703531e-05,
+      "loss": 1.0933,
+      "step": 3500
+    },
+    {
+      "epoch": 0.3502,
+      "learning_rate": 1.2581446386447171e-05,
+      "loss": 1.0334,
+      "step": 3502
+    },
+    {
+      "epoch": 0.3504,
+      "learning_rate": 1.2588190451025202e-05,
+      "loss": 0.7585,
+      "step": 3504
+    },
+    {
+      "epoch": 0.3506,
+      "learning_rate": 1.2594933254150647e-05,
+      "loss": 1.0609,
+      "step": 3506
+    },
+    {
+      "epoch": 0.3508,
+      "learning_rate": 1.2601674792537149e-05,
+      "loss": 0.0498,
+      "step": 3508
+    },
+    {
+      "epoch": 0.351,
+      "learning_rate": 1.2608415062898963e-05,
+      "loss": 0.9984,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3512,
+      "learning_rate": 1.2615154061950978e-05,
+      "loss": 0.1133,
+      "step": 3512
+    },
+    {
+      "epoch": 0.3514,
+      "learning_rate": 1.262189178640864e-05,
+      "loss": 0.4564,
+      "step": 3514
+    },
+    {
+      "epoch": 0.3516,
+      "learning_rate": 1.2628628232988111e-05,
+      "loss": 0.1503,
+      "step": 3516
+    },
+    {
+      "epoch": 0.3518,
+      "learning_rate": 1.2635363398406133e-05,
+      "loss": 1.1809,
+      "step": 3518
+    },
+    {
+      "epoch": 0.352,
+      "learning_rate": 1.2642097279380017e-05,
+      "loss": 0.7056,
+      "step": 3520
+    },
+    {
+      "epoch": 0.3522,
+      "learning_rate": 1.2648829872627797e-05,
+      "loss": 1.2366,
+      "step": 3522
+    },
+    {
+      "epoch": 0.3524,
+      "learning_rate": 1.2655561174868095e-05,
+      "loss": 0.0516,
+      "step": 3524
+    },
+    {
+      "epoch": 0.3526,
+      "learning_rate": 1.266229118282012e-05,
+      "loss": 0.6453,
+      "step": 3526
+    },
+    {
+      "epoch": 0.3528,
+      "learning_rate": 1.2669019893203745e-05,
+      "loss": 0.7186,
+      "step": 3528
+    },
+    {
+      "epoch": 0.353,
+      "learning_rate": 1.2675747302739516e-05,
+      "loss": 0.5776,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3532,
+      "learning_rate": 1.2682473408148574e-05,
+      "loss": 0.522,
+      "step": 3532
+    },
+    {
+      "epoch": 0.3534,
+      "learning_rate": 1.2689198206152644e-05,
+      "loss": 0.4329,
+      "step": 3534
+    },
+    {
+      "epoch": 0.3536,
+      "learning_rate": 1.2695921693474196e-05,
+      "loss": 0.5102,
+      "step": 3536
+    },
+    {
+      "epoch": 0.3538,
+      "learning_rate": 1.2702643866836281e-05,
+      "loss": 0.4806,
+      "step": 3538
+    },
+    {
+      "epoch": 0.354,
+      "learning_rate": 1.2709364722962564e-05,
+      "loss": 0.4111,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3542,
+      "learning_rate": 1.2716084258577373e-05,
+      "loss": 0.2402,
+      "step": 3542
+    },
+    {
+      "epoch": 0.3544,
+      "learning_rate": 1.2722802470405746e-05,
+      "loss": 0.4658,
+      "step": 3544
+    },
+    {
+      "epoch": 0.3546,
+      "learning_rate": 1.2729519355173254e-05,
+      "loss": 1.1571,
+      "step": 3546
+    },
+    {
+      "epoch": 0.3548,
+      "learning_rate": 1.2736234909606188e-05,
+      "loss": 0.6895,
+      "step": 3548
+    },
+    {
+      "epoch": 0.355,
+      "learning_rate": 1.2742949130431468e-05,
+      "loss": 0.8223,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3552,
+      "learning_rate": 1.274966201437666e-05,
+      "loss": 0.1338,
+      "step": 3552
+    },
+    {
+      "epoch": 0.3554,
+      "learning_rate": 1.2756373558169992e-05,
+      "loss": 0.4717,
+      "step": 3554
+    },
+    {
+      "epoch": 0.3556,
+      "learning_rate": 1.276308375854032e-05,
+      "loss": 1.306,
+      "step": 3556
+    },
+    {
+      "epoch": 0.3558,
+      "learning_rate": 1.2769792612217224e-05,
+      "loss": 0.8632,
+      "step": 3558
+    },
+    {
+      "epoch": 0.356,
+      "learning_rate": 1.2776500115930838e-05,
+      "loss": 0.5108,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3562,
+      "learning_rate": 1.2783206266412028e-05,
+      "loss": 0.8448,
+      "step": 3562
+    },
+    {
+      "epoch": 0.3564,
+      "learning_rate": 1.278991106039229e-05,
+      "loss": 0.7342,
+      "step": 3564
+    },
+    {
+      "epoch": 0.3566,
+      "learning_rate": 1.2796614494603795e-05,
+      "loss": 0.5437,
+      "step": 3566
+    },
+    {
+      "epoch": 0.3568,
+      "learning_rate": 1.280331656577937e-05,
+      "loss": 0.0912,
+      "step": 3568
+    },
+    {
+      "epoch": 0.357,
+      "learning_rate": 1.2810017270652508e-05,
+      "loss": 0.7027,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3572,
+      "learning_rate": 1.2816716605957363e-05,
+      "loss": 0.738,
+      "step": 3572
+    },
+    {
+      "epoch": 0.3574,
+      "learning_rate": 1.282341456842876e-05,
+      "loss": 0.2262,
+      "step": 3574
+    },
+    {
+      "epoch": 0.3576,
+      "learning_rate": 1.2830111154802198e-05,
+      "loss": 0.7844,
+      "step": 3576
+    },
+    {
+      "epoch": 0.3578,
+      "learning_rate": 1.283680636181384e-05,
+      "loss": 1.0604,
+      "step": 3578
+    },
+    {
+      "epoch": 0.358,
+      "learning_rate": 1.2843500186200522e-05,
+      "loss": 0.1503,
+      "step": 3580
+    },
+    {
+      "epoch": 0.3582,
+      "learning_rate": 1.2850192624699756e-05,
+      "loss": 0.2712,
+      "step": 3582
+    },
+    {
+      "epoch": 0.3584,
+      "learning_rate": 1.2856883674049747e-05,
+      "loss": 0.31,
+      "step": 3584
+    },
+    {
+      "epoch": 0.3586,
+      "learning_rate": 1.2863573330989308e-05,
+      "loss": 0.4479,
+      "step": 3586
+    },
+    {
+      "epoch": 0.3588,
+      "learning_rate": 1.2870261592258028e-05,
+      "loss": 0.2167,
+      "step": 3588
+    },
+    {
+      "epoch": 0.359,
+      "learning_rate": 1.2876948454596122e-05,
+      "loss": 0.7998,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3592,
+      "learning_rate": 1.2883633914744485e-05,
+      "loss": 0.661,
+      "step": 3592
+    },
+    {
+      "epoch": 0.3594,
+      "learning_rate": 1.2890317969444708e-05,
+      "loss": 0.5851,
+      "step": 3594
+    },
+    {
+      "epoch": 0.3596,
+      "learning_rate": 1.2897000615439066e-05,
+      "loss": 0.8869,
+      "step": 3596
+    },
+    {
+      "epoch": 0.3598,
+      "learning_rate": 1.2903681849470535e-05,
+      "loss": 0.6899,
+      "step": 3598
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.291036166828271e-05,
+      "loss": 0.5083,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3602,
+      "learning_rate": 1.291704006861998e-05,
+      "loss": 0.7021,
+      "step": 3602
+    },
+    {
+      "epoch": 0.3604,
+      "learning_rate": 1.2923717047227373e-05,
+      "loss": 0.1198,
+      "step": 3604
+    },
+    {
+      "epoch": 0.3606,
+      "learning_rate": 1.2930392600850565e-05,
+      "loss": 0.1381,
+      "step": 3606
+    },
+    {
+      "epoch": 0.3608,
+      "learning_rate": 1.2937066726236016e-05,
+      "loss": 0.7398,
+      "step": 3608
+    },
+    {
+      "epoch": 0.361,
+      "learning_rate": 1.2943739420130843e-05,
+      "loss": 0.4982,
+      "step": 3610
+    },
+    {
+      "epoch": 0.3612,
+      "learning_rate": 1.2950410679282818e-05,
+      "loss": 0.9701,
+      "step": 3612
+    },
+    {
+      "epoch": 0.3614,
+      "learning_rate": 1.2957080500440455e-05,
+      "loss": 0.6179,
+      "step": 3614
+    },
+    {
+      "epoch": 0.3616,
+      "learning_rate": 1.2963748880352995e-05,
+      "loss": 0.5483,
+      "step": 3616
+    },
+    {
+      "epoch": 0.3618,
+      "learning_rate": 1.2970415815770353e-05,
+      "loss": 0.5708,
+      "step": 3618
+    },
+    {
+      "epoch": 0.362,
+      "learning_rate": 1.297708130344311e-05,
+      "loss": 0.3799,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3622,
+      "learning_rate": 1.2983745340122589e-05,
+      "loss": 0.7213,
+      "step": 3622
+    },
+    {
+      "epoch": 0.3624,
+      "learning_rate": 1.299040792256087e-05,
+      "loss": 0.5235,
+      "step": 3624
+    },
+    {
+      "epoch": 0.3626,
+      "learning_rate": 1.299706904751064e-05,
+      "loss": 0.8309,
+      "step": 3626
+    },
+    {
+      "epoch": 0.3628,
+      "learning_rate": 1.3003728711725347e-05,
+      "loss": 0.2273,
+      "step": 3628
+    },
+    {
+      "epoch": 0.363,
+      "learning_rate": 1.3010386911959205e-05,
+      "loss": 0.5663,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3632,
+      "learning_rate": 1.3017043644967036e-05,
+      "loss": 0.3303,
+      "step": 3632
+    },
+    {
+      "epoch": 0.3634,
+      "learning_rate": 1.3023698907504447e-05,
+      "loss": 0.8499,
+      "step": 3634
+    },
+    {
+      "epoch": 0.3636,
+      "learning_rate": 1.303035269632774e-05,
+      "loss": 1.3761,
+      "step": 3636
+    },
+    {
+      "epoch": 0.3638,
+      "learning_rate": 1.3037005008193944e-05,
+      "loss": 1.0667,
+      "step": 3638
+    },
+    {
+      "epoch": 0.364,
+      "learning_rate": 1.3043655839860801e-05,
+      "loss": 0.6698,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3642,
+      "learning_rate": 1.3050305188086757e-05,
+      "loss": 0.4949,
+      "step": 3642
+    },
+    {
+      "epoch": 0.3644,
+      "learning_rate": 1.3056953049631056e-05,
+      "loss": 0.4918,
+      "step": 3644
+    },
+    {
+      "epoch": 0.3646,
+      "learning_rate": 1.3063599421253556e-05,
+      "loss": 0.6559,
+      "step": 3646
+    },
+    {
+      "epoch": 0.3648,
+      "learning_rate": 1.3070244299714918e-05,
+      "loss": 0.5215,
+      "step": 3648
+    },
+    {
+      "epoch": 0.365,
+      "learning_rate": 1.3076887681776504e-05,
+      "loss": 0.808,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3652,
+      "learning_rate": 1.3083529564200412e-05,
+      "loss": 0.5596,
+      "step": 3652
+    },
+    {
+      "epoch": 0.3654,
+      "learning_rate": 1.309016994374947e-05,
+      "loss": 0.4771,
+      "step": 3654
+    },
+    {
+      "epoch": 0.3656,
+      "learning_rate": 1.3096808817187238e-05,
+      "loss": 0.8142,
+      "step": 3656
+    },
+    {
+      "epoch": 0.3658,
+      "learning_rate": 1.310344618127801e-05,
+      "loss": 0.563,
+      "step": 3658
+    },
+    {
+      "epoch": 0.366,
+      "learning_rate": 1.3110082032786813e-05,
+      "loss": 0.1576,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3662,
+      "learning_rate": 1.3116716368479415e-05,
+      "loss": 0.867,
+      "step": 3662
+    },
+    {
+      "epoch": 0.3664,
+      "learning_rate": 1.3123349185122321e-05,
+      "loss": 0.4655,
+      "step": 3664
+    },
+    {
+      "epoch": 0.3666,
+      "learning_rate": 1.3129980479482776e-05,
+      "loss": 0.1479,
+      "step": 3666
+    },
+    {
+      "epoch": 0.3668,
+      "learning_rate": 1.3136610248328769e-05,
+      "loss": 0.8621,
+      "step": 3668
+    },
+    {
+      "epoch": 0.367,
+      "learning_rate": 1.3143238488429049e-05,
+      "loss": 1.2677,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3672,
+      "learning_rate": 1.314986519655304e-05,
+      "loss": 0.7012,
+      "step": 3672
+    },
+    {
+      "epoch": 0.3674,
+      "learning_rate": 1.3156490369471018e-05,
+      "loss": 0.2761,
+      "step": 3674
+    },
+    {
+      "epoch": 0.3676,
+      "learning_rate": 1.3163114003953932e-05,
+      "loss": 0.8818,
+      "step": 3676
+    },
+    {
+      "epoch": 0.3678,
+      "learning_rate": 1.316973609677351e-05,
+      "loss": 0.3122,
+      "step": 3678
+    },
+    {
+      "epoch": 0.368,
+      "learning_rate": 1.3176356644702217e-05,
+      "loss": 1.079,
+      "step": 3680
+    },
+    {
+      "epoch": 0.3682,
+      "learning_rate": 1.3182975644513286e-05,
+      "loss": 0.5597,
+      "step": 3682
+    },
+    {
+      "epoch": 0.3684,
+      "learning_rate": 1.3189593092980708e-05,
+      "loss": 0.4921,
+      "step": 3684
+    },
+    {
+      "epoch": 0.3686,
+      "learning_rate": 1.319620898687917e-05,
+      "loss": 0.9193,
+      "step": 3686
+    },
+    {
+      "epoch": 0.3688,
+      "learning_rate": 1.3202823322984218e-05,
+      "loss": 0.9631,
+      "step": 3688
+    },
+    {
+      "epoch": 0.369,
+      "learning_rate": 1.3209436098072102e-05,
+      "loss": 0.9692,
+      "step": 3690
+    },
+    {
+      "epoch": 0.3692,
+      "learning_rate": 1.3216047308919804e-05,
+      "loss": 0.6023,
+      "step": 3692
+    },
+    {
+      "epoch": 0.3694,
+      "learning_rate": 1.32226569523051e-05,
+      "loss": 0.5885,
+      "step": 3694
+    },
+    {
+      "epoch": 0.3696,
+      "learning_rate": 1.3229265025006579e-05,
+      "loss": 0.9593,
+      "step": 3696
+    },
+    {
+      "epoch": 0.3698,
+      "learning_rate": 1.3235871523803501e-05,
+      "loss": 0.1786,
+      "step": 3698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.3242476445475932e-05,
+      "loss": 0.203,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3702,
+      "learning_rate": 1.324907978680475e-05,
+      "loss": 0.5274,
+      "step": 3702
+    },
+    {
+      "epoch": 0.3704,
+      "learning_rate": 1.325568154457157e-05,
+      "loss": 0.4156,
+      "step": 3704
+    },
+    {
+      "epoch": 0.3706,
+      "learning_rate": 1.3262281715558738e-05,
+      "loss": 0.61,
+      "step": 3706
+    },
+    {
+      "epoch": 0.3708,
+      "learning_rate": 1.326888029654941e-05,
+      "loss": 0.2144,
+      "step": 3708
+    },
+    {
+      "epoch": 0.371,
+      "learning_rate": 1.3275477284327572e-05,
+      "loss": 0.5426,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3712,
+      "learning_rate": 1.328207267567788e-05,
+      "loss": 1.0489,
+      "step": 3712
+    },
+    {
+      "epoch": 0.3714,
+      "learning_rate": 1.3288666467385815e-05,
+      "loss": 0.6415,
+      "step": 3714
+    },
+    {
+      "epoch": 0.3716,
+      "learning_rate": 1.3295258656237703e-05,
+      "loss": 0.99,
+      "step": 3716
+    },
+    {
+      "epoch": 0.3718,
+      "learning_rate": 1.3301849239020537e-05,
+      "loss": 0.721,
+      "step": 3718
+    },
+    {
+      "epoch": 0.372,
+      "learning_rate": 1.3308438212522164e-05,
+      "loss": 0.8002,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3722,
+      "learning_rate": 1.3315025573531193e-05,
+      "loss": 0.7925,
+      "step": 3722
+    },
+    {
+      "epoch": 0.3724,
+      "learning_rate": 1.3321611318837033e-05,
+      "loss": 0.5674,
+      "step": 3724
+    },
+    {
+      "epoch": 0.3726,
+      "learning_rate": 1.3328195445229865e-05,
+      "loss": 0.887,
+      "step": 3726
+    },
+    {
+      "epoch": 0.3728,
+      "learning_rate": 1.333477794950067e-05,
+      "loss": 0.8267,
+      "step": 3728
+    },
+    {
+      "epoch": 0.373,
+      "learning_rate": 1.3341358828441214e-05,
+      "loss": 0.7039,
+      "step": 3730
+    },
+    {
+      "epoch": 0.3732,
+      "learning_rate": 1.3347938078844053e-05,
+      "loss": 0.9569,
+      "step": 3732
+    },
+    {
+      "epoch": 0.3734,
+      "learning_rate": 1.3354515697502548e-05,
+      "loss": 0.6893,
+      "step": 3734
+    },
+    {
+      "epoch": 0.3736,
+      "learning_rate": 1.336109168121084e-05,
+      "loss": 0.5652,
+      "step": 3736
+    },
+    {
+      "epoch": 0.3738,
+      "learning_rate": 1.3367666026763879e-05,
+      "loss": 0.6812,
+      "step": 3738
+    },
+    {
+      "epoch": 0.374,
+      "learning_rate": 1.3374238730957408e-05,
+      "loss": 0.5861,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3742,
+      "learning_rate": 1.338080979058797e-05,
+      "loss": 0.3547,
+      "step": 3742
+    },
+    {
+      "epoch": 0.3744,
+      "learning_rate": 1.3387379202452908e-05,
+      "loss": 0.4347,
+      "step": 3744
+    },
+    {
+      "epoch": 0.3746,
+      "learning_rate": 1.3393946963350378e-05,
+      "loss": 1.072,
+      "step": 3746
+    },
+    {
+      "epoch": 0.3748,
+      "learning_rate": 1.3400513070079324e-05,
+      "loss": 0.2405,
+      "step": 3748
+    },
+    {
+      "epoch": 0.375,
+      "learning_rate": 1.340707751943951e-05,
+      "loss": 1.1901,
+      "step": 3750
+    },
+    {
+      "epoch": 0.3752,
+      "learning_rate": 1.3413640308231504e-05,
+      "loss": 0.1048,
+      "step": 3752
+    },
+    {
+      "epoch": 0.3754,
+      "learning_rate": 1.3420201433256682e-05,
+      "loss": 1.3169,
+      "step": 3754
+    },
+    {
+      "epoch": 0.3756,
+      "learning_rate": 1.3426760891317243e-05,
+      "loss": 0.2276,
+      "step": 3756
+    },
+    {
+      "epoch": 0.3758,
+      "learning_rate": 1.3433318679216145e-05,
+      "loss": 0.8541,
+      "step": 3758
+    },
+    {
+      "epoch": 0.376,
+      "learning_rate": 1.3439874793757249e-05,
+      "loss": 1.2314,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3762,
+      "learning_rate": 1.3446429231745162e-05,
+      "loss": 0.7877,
+      "step": 3762
+    },
+    {
+      "epoch": 0.3764,
+      "learning_rate": 1.3452981989985339e-05,
+      "loss": 0.19,
+      "step": 3764
+    },
+    {
+      "epoch": 0.3766,
+      "learning_rate": 1.3459533065284039e-05,
+      "loss": 0.4233,
+      "step": 3766
+    },
+    {
+      "epoch": 0.3768,
+      "learning_rate": 1.346608245444835e-05,
+      "loss": 0.819,
+      "step": 3768
+    },
+    {
+      "epoch": 0.377,
+      "learning_rate": 1.3472630154286197e-05,
+      "loss": 1.2813,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3772,
+      "learning_rate": 1.3479176161606257e-05,
+      "loss": 0.5543,
+      "step": 3772
+    },
+    {
+      "epoch": 0.3774,
+      "learning_rate": 1.348572047321814e-05,
+      "loss": 0.5149,
+      "step": 3774
+    },
+    {
+      "epoch": 0.3776,
+      "learning_rate": 1.3492263085932226e-05,
+      "loss": 0.8118,
+      "step": 3776
+    },
+    {
+      "epoch": 0.3778,
+      "learning_rate": 1.3498803996559692e-05,
+      "loss": 0.8384,
+      "step": 3778
+    },
+    {
+      "epoch": 0.378,
+      "learning_rate": 1.3505343201912577e-05,
+      "loss": 0.4651,
+      "step": 3780
+    },
+    {
+      "epoch": 0.3782,
+      "learning_rate": 1.3511880698803803e-05,
+      "loss": 0.4809,
+      "step": 3782
+    },
+    {
+      "epoch": 0.3784,
+      "learning_rate": 1.3518416484047021e-05,
+      "loss": 0.0795,
+      "step": 3784
+    },
+    {
+      "epoch": 0.3786,
+      "learning_rate": 1.3524950554456773e-05,
+      "loss": 0.5084,
+      "step": 3786
+    },
+    {
+      "epoch": 0.3788,
+      "learning_rate": 1.3531482906848476e-05,
+      "loss": 0.661,
+      "step": 3788
+    },
+    {
+      "epoch": 0.379,
+      "learning_rate": 1.3538013538038296e-05,
+      "loss": 0.4328,
+      "step": 3790
+    },
+    {
+      "epoch": 0.3792,
+      "learning_rate": 1.3544542444843298e-05,
+      "loss": 0.8622,
+      "step": 3792
+    },
+    {
+      "epoch": 0.3794,
+      "learning_rate": 1.3551069624081356e-05,
+      "loss": 0.2311,
+      "step": 3794
+    },
+    {
+      "epoch": 0.3796,
+      "learning_rate": 1.355759507257125e-05,
+      "loss": 0.1075,
+      "step": 3796
+    },
+    {
+      "epoch": 0.3798,
+      "learning_rate": 1.3564118787132507e-05,
+      "loss": 0.5593,
+      "step": 3798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.357064076458555e-05,
+      "loss": 0.4445,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3802,
+      "learning_rate": 1.3577161001751692e-05,
+      "loss": 0.6785,
+      "step": 3802
+    },
+    {
+      "epoch": 0.3804,
+      "learning_rate": 1.3583679495453e-05,
+      "loss": 0.8244,
+      "step": 3804
+    },
+    {
+      "epoch": 0.3806,
+      "learning_rate": 1.3590196242512461e-05,
+      "loss": 0.523,
+      "step": 3806
+    },
+    {
+      "epoch": 0.3808,
+      "learning_rate": 1.3596711239753885e-05,
+      "loss": 0.6698,
+      "step": 3808
+    },
+    {
+      "epoch": 0.381,
+      "learning_rate": 1.3603224484001944e-05,
+      "loss": 0.4852,
+      "step": 3810
+    },
+    {
+      "epoch": 0.3812,
+      "learning_rate": 1.3609735972082163e-05,
+      "loss": 1.8795,
+      "step": 3812
+    },
+    {
+      "epoch": 0.3814,
+      "learning_rate": 1.361624570082092e-05,
+      "loss": 0.4041,
+      "step": 3814
+    },
+    {
+      "epoch": 0.3816,
+      "learning_rate": 1.3622753667045453e-05,
+      "loss": 0.666,
+      "step": 3816
+    },
+    {
+      "epoch": 0.3818,
+      "learning_rate": 1.362925986758386e-05,
+      "loss": 0.8822,
+      "step": 3818
+    },
+    {
+      "epoch": 0.382,
+      "learning_rate": 1.3635764299265094e-05,
+      "loss": 0.1391,
+      "step": 3820
+    },
+    {
+      "epoch": 0.3822,
+      "learning_rate": 1.364226695891898e-05,
+      "loss": 0.874,
+      "step": 3822
+    },
+    {
+      "epoch": 0.3824,
+      "learning_rate": 1.3648767843376191e-05,
+      "loss": 0.8123,
+      "step": 3824
+    },
+    {
+      "epoch": 0.3826,
+      "learning_rate": 1.3655266949468287e-05,
+      "loss": 0.1427,
+      "step": 3826
+    },
+    {
+      "epoch": 0.3828,
+      "learning_rate": 1.3661764274027674e-05,
+      "loss": 0.1002,
+      "step": 3828
+    },
+    {
+      "epoch": 0.383,
+      "learning_rate": 1.3668259813887637e-05,
+      "loss": 0.3836,
+      "step": 3830
+    },
+    {
+      "epoch": 0.3832,
+      "learning_rate": 1.3674753565882329e-05,
+      "loss": 1.3379,
+      "step": 3832
+    },
+    {
+      "epoch": 0.3834,
+      "learning_rate": 1.3681245526846773e-05,
+      "loss": 0.5053,
+      "step": 3834
+    },
+    {
+      "epoch": 0.3836,
+      "learning_rate": 1.3687735693616868e-05,
+      "loss": 0.9122,
+      "step": 3836
+    },
+    {
+      "epoch": 0.3838,
+      "learning_rate": 1.3694224063029386e-05,
+      "loss": 0.8416,
+      "step": 3838
+    },
+    {
+      "epoch": 0.384,
+      "learning_rate": 1.3700710631921976e-05,
+      "loss": 0.8203,
+      "step": 3840
+    },
+    {
+      "epoch": 0.3842,
+      "learning_rate": 1.3707195397133176e-05,
+      "loss": 0.0394,
+      "step": 3842
+    },
+    {
+      "epoch": 0.3844,
+      "learning_rate": 1.371367835550234e-05,
+      "loss": 0.24,
+      "step": 3844
+    },
+    {
+      "epoch": 0.3846,
+      "learning_rate": 1.3720159503869806e-05,
+      "loss": 1.8128,
+      "step": 3846
+    },
+    {
+      "epoch": 0.3848,
+      "learning_rate": 1.3726638839076739e-05,
+      "loss": 0.7194,
+      "step": 3848
+    },
+    {
+      "epoch": 0.385,
+      "learning_rate": 1.3733116357965156e-05,
+      "loss": 0.3922,
+      "step": 3850
+    },
+    {
+      "epoch": 0.3852,
+      "learning_rate": 1.3739592057377994e-05,
+      "loss": 0.7691,
+      "step": 3852
+    },
+    {
+      "epoch": 0.3854,
+      "learning_rate": 1.374606593415911e-05,
+      "loss": 0.5839,
+      "step": 3854
+    },
+    {
+      "epoch": 0.3856,
+      "learning_rate": 1.3752537985153213e-05,
+      "loss": 0.6217,
+      "step": 3856
+    },
+    {
+      "epoch": 0.3858,
+      "learning_rate": 1.3759008207205855e-05,
+      "loss": 0.4681,
+      "step": 3858
+    },
+    {
+      "epoch": 0.386,
+      "learning_rate": 1.3765476597163581e-05,
+      "loss": 0.7355,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3862,
+      "learning_rate": 1.377194315187377e-05,
+      "loss": 0.5564,
+      "step": 3862
+    },
+    {
+      "epoch": 0.3864,
+      "learning_rate": 1.3778407868184675e-05,
+      "loss": 1.0848,
+      "step": 3864
+    },
+    {
+      "epoch": 0.3866,
+      "learning_rate": 1.3784870742945468e-05,
+      "loss": 0.1948,
+      "step": 3866
+    },
+    {
+      "epoch": 0.3868,
+      "learning_rate": 1.3791331773006273e-05,
+      "loss": 0.798,
+      "step": 3868
+    },
+    {
+      "epoch": 0.387,
+      "learning_rate": 1.3797790955218014e-05,
+      "loss": 0.2664,
+      "step": 3870
+    },
+    {
+      "epoch": 0.3872,
+      "learning_rate": 1.3804248286432562e-05,
+      "loss": 0.5977,
+      "step": 3872
+    },
+    {
+      "epoch": 0.3874,
+      "learning_rate": 1.3810703763502744e-05,
+      "loss": 0.943,
+      "step": 3874
+    },
+    {
+      "epoch": 0.3876,
+      "learning_rate": 1.3817157383282184e-05,
+      "loss": 0.3214,
+      "step": 3876
+    },
+    {
+      "epoch": 0.3878,
+      "learning_rate": 1.3823609142625492e-05,
+      "loss": 0.2788,
+      "step": 3878
+    },
+    {
+      "epoch": 0.388,
+      "learning_rate": 1.3830059038388136e-05,
+      "loss": 0.5747,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3882,
+      "learning_rate": 1.3836507067426563e-05,
+      "loss": 0.7373,
+      "step": 3882
+    },
+    {
+      "epoch": 0.3884,
+      "learning_rate": 1.3842953226598036e-05,
+      "loss": 0.2118,
+      "step": 3884
+    },
+    {
+      "epoch": 0.3886,
+      "learning_rate": 1.3849397512760793e-05,
+      "loss": 0.4715,
+      "step": 3886
+    },
+    {
+      "epoch": 0.3888,
+      "learning_rate": 1.3855839922773963e-05,
+      "loss": 0.0755,
+      "step": 3888
+    },
+    {
+      "epoch": 0.389,
+      "learning_rate": 1.38622804534976e-05,
+      "loss": 0.7263,
+      "step": 3890
+    },
+    {
+      "epoch": 0.3892,
+      "learning_rate": 1.386871910179266e-05,
+      "loss": 0.6195,
+      "step": 3892
+    },
+    {
+      "epoch": 0.3894,
+      "learning_rate": 1.3875155864521027e-05,
+      "loss": 1.0616,
+      "step": 3894
+    },
+    {
+      "epoch": 0.3896,
+      "learning_rate": 1.3881590738545503e-05,
+      "loss": 1.356,
+      "step": 3896
+    },
+    {
+      "epoch": 0.3898,
+      "learning_rate": 1.3888023720729806e-05,
+      "loss": 0.2528,
+      "step": 3898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894454807938582e-05,
+      "loss": 0.7197,
+      "step": 3900
+    },
+    {
+      "epoch": 0.3902,
+      "learning_rate": 1.3900883997037393e-05,
+      "loss": 0.588,
+      "step": 3902
+    },
+    {
+      "epoch": 0.3904,
+      "learning_rate": 1.3907311284892733e-05,
+      "loss": 0.4138,
+      "step": 3904
+    },
+    {
+      "epoch": 0.3906,
+      "learning_rate": 1.391373666837202e-05,
+      "loss": 0.4004,
+      "step": 3906
+    },
+    {
+      "epoch": 0.3908,
+      "learning_rate": 1.39201601443436e-05,
+      "loss": 0.2185,
+      "step": 3908
+    },
+    {
+      "epoch": 0.391,
+      "learning_rate": 1.3926581709676746e-05,
+      "loss": 0.4092,
+      "step": 3910
+    },
+    {
+      "epoch": 0.3912,
+      "learning_rate": 1.3933001361241667e-05,
+      "loss": 0.993,
+      "step": 3912
+    },
+    {
+      "epoch": 0.3914,
+      "learning_rate": 1.3939419095909506e-05,
+      "loss": 1.0319,
+      "step": 3914
+    },
+    {
+      "epoch": 0.3916,
+      "learning_rate": 1.3945834910552332e-05,
+      "loss": 0.0528,
+      "step": 3916
+    },
+    {
+      "epoch": 0.3918,
+      "learning_rate": 1.3952248802043158e-05,
+      "loss": 0.3083,
+      "step": 3918
+    },
+    {
+      "epoch": 0.392,
+      "learning_rate": 1.395866076725593e-05,
+      "loss": 0.8809,
+      "step": 3920
+    },
+    {
+      "epoch": 0.3922,
+      "learning_rate": 1.396507080306555e-05,
+      "loss": 1.301,
+      "step": 3922
+    },
+    {
+      "epoch": 0.3924,
+      "learning_rate": 1.3971478906347798e-05,
+      "loss": 1.4473,
+      "step": 3924
+    },
+    {
+      "epoch": 0.3926,
+      "learning_rate": 1.397788507397949e-05,
+      "loss": 0.1568,
+      "step": 3926
+    },
+    {
+      "epoch": 0.3928,
+      "learning_rate": 1.3984289302838334e-05,
+      "loss": 0.6458,
+      "step": 3928
+    },
+    {
+      "epoch": 0.393,
+      "learning_rate": 1.3990691589802943e-05,
+      "loss": 0.4917,
+      "step": 3930
+    },
+    {
+      "epoch": 0.3932,
+      "learning_rate": 1.3997091931752966e-05,
+      "loss": 0.0864,
+      "step": 3932
+    },
+    {
+      "epoch": 0.3934,
+      "learning_rate": 1.4003490325568956e-05,
+      "loss": 0.4464,
+      "step": 3934
+    },
+    {
+      "epoch": 0.3936,
+      "learning_rate": 1.4009886768132378e-05,
+      "loss": 0.8698,
+      "step": 3936
+    },
+    {
+      "epoch": 0.3938,
+      "learning_rate": 1.4016281256325688e-05,
+      "loss": 0.4884,
+      "step": 3938
+    },
+    {
+      "epoch": 0.394,
+      "learning_rate": 1.402267378703232e-05,
+      "loss": 0.7461,
+      "step": 3940
+    },
+    {
+      "epoch": 0.3942,
+      "learning_rate": 1.4029064357136632e-05,
+      "loss": 0.1211,
+      "step": 3942
+    },
+    {
+      "epoch": 0.3944,
+      "learning_rate": 1.4035452963523888e-05,
+      "loss": 0.6199,
+      "step": 3944
+    },
+    {
+      "epoch": 0.3946,
+      "learning_rate": 1.4041839603080411e-05,
+      "loss": 1.153,
+      "step": 3946
+    },
+    {
+      "epoch": 0.3948,
+      "learning_rate": 1.4048224272693427e-05,
+      "loss": 0.9284,
+      "step": 3948
+    },
+    {
+      "epoch": 0.395,
+      "learning_rate": 1.4054606969251096e-05,
+      "loss": 0.079,
+      "step": 3950
+    },
+    {
+      "epoch": 0.3952,
+      "learning_rate": 1.4060987689642566e-05,
+      "loss": 0.4135,
+      "step": 3952
+    },
+    {
+      "epoch": 0.3954,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 0.9188,
+      "step": 3954
+    },
+    {
+      "epoch": 0.3956,
+      "learning_rate": 1.4073743189488438e-05,
+      "loss": 0.6067,
+      "step": 3956
+    },
+    {
+      "epoch": 0.3958,
+      "learning_rate": 1.4080117962725929e-05,
+      "loss": 1.0444,
+      "step": 3958
+    },
+    {
+      "epoch": 0.396,
+      "learning_rate": 1.4086490747363492e-05,
+      "loss": 0.6673,
+      "step": 3960
+    },
+    {
+      "epoch": 0.3962,
+      "learning_rate": 1.4092861540295107e-05,
+      "loss": 0.3211,
+      "step": 3962
+    },
+    {
+      "epoch": 0.3964,
+      "learning_rate": 1.4099230338415728e-05,
+      "loss": 0.5796,
+      "step": 3964
+    },
+    {
+      "epoch": 0.3966,
+      "learning_rate": 1.4105597138621281e-05,
+      "loss": 0.856,
+      "step": 3966
+    },
+    {
+      "epoch": 0.3968,
+      "learning_rate": 1.4111961937808663e-05,
+      "loss": 1.0951,
+      "step": 3968
+    },
+    {
+      "epoch": 0.397,
+      "learning_rate": 1.411832473287575e-05,
+      "loss": 0.8239,
+      "step": 3970
+    },
+    {
+      "epoch": 0.3972,
+      "learning_rate": 1.4124685520721391e-05,
+      "loss": 0.1183,
+      "step": 3972
+    },
+    {
+      "epoch": 0.3974,
+      "learning_rate": 1.4131044298245416e-05,
+      "loss": 0.5294,
+      "step": 3974
+    },
+    {
+      "epoch": 0.3976,
+      "learning_rate": 1.4137401062348634e-05,
+      "loss": 0.4781,
+      "step": 3976
+    },
+    {
+      "epoch": 0.3978,
+      "learning_rate": 1.414375580993284e-05,
+      "loss": 2.335,
+      "step": 3978
+    },
+    {
+      "epoch": 0.398,
+      "learning_rate": 1.41501085379008e-05,
+      "loss": 0.6665,
+      "step": 3980
+    },
+    {
+      "epoch": 0.3982,
+      "learning_rate": 1.4156459243156275e-05,
+      "loss": 0.6381,
+      "step": 3982
+    },
+    {
+      "epoch": 0.3984,
+      "learning_rate": 1.4162807922604009e-05,
+      "loss": 0.4633,
+      "step": 3984
+    },
+    {
+      "epoch": 0.3986,
+      "learning_rate": 1.416915457314973e-05,
+      "loss": 0.5546,
+      "step": 3986
+    },
+    {
+      "epoch": 0.3988,
+      "learning_rate": 1.4175499191700162e-05,
+      "loss": 0.7114,
+      "step": 3988
+    },
+    {
+      "epoch": 0.399,
+      "learning_rate": 1.418184177516301e-05,
+      "loss": 0.7128,
+      "step": 3990
+    },
+    {
+      "epoch": 0.3992,
+      "learning_rate": 1.4188182320446982e-05,
+      "loss": 0.31,
+      "step": 3992
+    },
+    {
+      "epoch": 0.3994,
+      "learning_rate": 1.4194520824461782e-05,
+      "loss": 1.1051,
+      "step": 3994
+    },
+    {
+      "epoch": 0.3996,
+      "learning_rate": 1.4200857284118056e-05,
+      "loss": 1.1069,
+      "step": 3996
+    },
+    {
+      "epoch": 0.3998,
+      "learning_rate": 1.420719169632754e-05,
+      "loss": 0.6453,
+      "step": 3998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.4213524058002901e-05,
+      "loss": 0.0484,
+      "step": 4000
+    },
+    {
+      "epoch": 0.4002,
+      "learning_rate": 1.4219854366057821e-05,
+      "loss": 0.4207,
+      "step": 4002
+    },
+    {
+      "epoch": 0.4004,
+      "learning_rate": 1.4226182617406988e-05,
+      "loss": 0.3493,
+      "step": 4004
+    },
+    {
+      "epoch": 0.4006,
+      "learning_rate": 1.4232508808966085e-05,
+      "loss": 0.5596,
+      "step": 4006
+    },
+    {
+      "epoch": 0.4008,
+      "learning_rate": 1.4238832937651823e-05,
+      "loss": 1.2118,
+      "step": 4008
+    },
+    {
+      "epoch": 0.401,
+      "learning_rate": 1.424515500038185e-05,
+      "loss": 0.2718,
+      "step": 4010
+    },
+    {
+      "epoch": 0.4012,
+      "learning_rate": 1.4251474994074917e-05,
+      "loss": 0.5675,
+      "step": 4012
+    },
+    {
+      "epoch": 0.4014,
+      "learning_rate": 1.4257792915650735e-05,
+      "loss": 0.5914,
+      "step": 4014
+    },
+    {
+      "epoch": 0.4016,
+      "learning_rate": 1.4264108762029975e-05,
+      "loss": 0.5574,
+      "step": 4016
+    },
+    {
+      "epoch": 0.4018,
+      "learning_rate": 1.4270422530134425e-05,
+      "loss": 0.6303,
+      "step": 4018
+    },
+    {
+      "epoch": 0.402,
+      "learning_rate": 1.4276734216886825e-05,
+      "loss": 0.9845,
+      "step": 4020
+    },
+    {
+      "epoch": 0.4022,
+      "learning_rate": 1.4283043819210906e-05,
+      "loss": 0.2503,
+      "step": 4022
+    },
+    {
+      "epoch": 0.4024,
+      "learning_rate": 1.4289351334031448e-05,
+      "loss": 0.811,
+      "step": 4024
+    },
+    {
+      "epoch": 0.4026,
+      "learning_rate": 1.4295656758274288e-05,
+      "loss": 0.6662,
+      "step": 4026
+    },
+    {
+      "epoch": 0.4028,
+      "learning_rate": 1.430196008886619e-05,
+      "loss": 0.2228,
+      "step": 4028
+    },
+    {
+      "epoch": 0.403,
+      "learning_rate": 1.430826132273499e-05,
+      "loss": 0.5143,
+      "step": 4030
+    },
+    {
+      "epoch": 0.4032,
+      "learning_rate": 1.4314560456809577e-05,
+      "loss": 0.6437,
+      "step": 4032
+    },
+    {
+      "epoch": 0.4034,
+      "learning_rate": 1.4320857488019826e-05,
+      "loss": 0.2149,
+      "step": 4034
+    },
+    {
+      "epoch": 0.4036,
+      "learning_rate": 1.432715241329661e-05,
+      "loss": 0.5796,
+      "step": 4036
+    },
+    {
+      "epoch": 0.4038,
+      "learning_rate": 1.4333445229571857e-05,
+      "loss": 0.3911,
+      "step": 4038
+    },
+    {
+      "epoch": 0.404,
+      "learning_rate": 1.4339735933778576e-05,
+      "loss": 0.3845,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4042,
+      "learning_rate": 1.4346024522850704e-05,
+      "loss": 0.4109,
+      "step": 4042
+    },
+    {
+      "epoch": 0.4044,
+      "learning_rate": 1.4352310993723276e-05,
+      "loss": 0.6074,
+      "step": 4044
+    },
+    {
+      "epoch": 0.4046,
+      "learning_rate": 1.4358595343332342e-05,
+      "loss": 0.3561,
+      "step": 4046
+    },
+    {
+      "epoch": 0.4048,
+      "learning_rate": 1.4364877568614985e-05,
+      "loss": 0.8855,
+      "step": 4048
+    },
+    {
+      "epoch": 0.405,
+      "learning_rate": 1.437115766650933e-05,
+      "loss": 0.6485,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4052,
+      "learning_rate": 1.4377435633954526e-05,
+      "loss": 0.3745,
+      "step": 4052
+    },
+    {
+      "epoch": 0.4054,
+      "learning_rate": 1.4383711467890772e-05,
+      "loss": 0.9903,
+      "step": 4054
+    },
+    {
+      "epoch": 0.4056,
+      "learning_rate": 1.4389985165259304e-05,
+      "loss": 0.7568,
+      "step": 4056
+    },
+    {
+      "epoch": 0.4058,
+      "learning_rate": 1.4396256723002398e-05,
+      "loss": 0.2312,
+      "step": 4058
+    },
+    {
+      "epoch": 0.406,
+      "learning_rate": 1.4402526138063368e-05,
+      "loss": 0.0149,
+      "step": 4060
+    },
+    {
+      "epoch": 0.4062,
+      "learning_rate": 1.4408793407386584e-05,
+      "loss": 0.501,
+      "step": 4062
+    },
+    {
+      "epoch": 0.4064,
+      "learning_rate": 1.4415058527917449e-05,
+      "loss": 1.1656,
+      "step": 4064
+    },
+    {
+      "epoch": 0.4066,
+      "learning_rate": 1.4421321496602423e-05,
+      "loss": 0.4997,
+      "step": 4066
+    },
+    {
+      "epoch": 0.4068,
+      "learning_rate": 1.4427582310389013e-05,
+      "loss": 0.9297,
+      "step": 4068
+    },
+    {
+      "epoch": 0.407,
+      "learning_rate": 1.4433840966225767e-05,
+      "loss": 1.1731,
+      "step": 4070
+    },
+    {
+      "epoch": 0.4072,
+      "learning_rate": 1.4440097461062301e-05,
+      "loss": 0.2173,
+      "step": 4072
+    },
+    {
+      "epoch": 0.4074,
+      "learning_rate": 1.444635179184927e-05,
+      "loss": 0.2012,
+      "step": 4074
+    },
+    {
+      "epoch": 0.4076,
+      "learning_rate": 1.445260395553839e-05,
+      "loss": 0.9541,
+      "step": 4076
+    },
+    {
+      "epoch": 0.4078,
+      "learning_rate": 1.4458853949082434e-05,
+      "loss": 0.678,
+      "step": 4078
+    },
+    {
+      "epoch": 0.408,
+      "learning_rate": 1.4465101769435245e-05,
+      "loss": 1.1287,
+      "step": 4080
+    },
+    {
+      "epoch": 0.4082,
+      "learning_rate": 1.4471347413551665e-05,
+      "loss": 0.6401,
+      "step": 4082
+    },
+    {
+      "epoch": 0.4084,
+      "learning_rate": 1.447759087838769e-05,
+      "loss": 0.6256,
+      "step": 4084
+    },
+    {
+      "epoch": 0.4086,
+      "learning_rate": 1.4483832160900332e-05,
+      "loss": 0.5521,
+      "step": 4086
+    },
+    {
+      "epoch": 0.4088,
+      "learning_rate": 1.4490071258047616e-05,
+      "loss": 0.6462,
+      "step": 4088
+    },
+    {
+      "epoch": 0.409,
+      "learning_rate": 1.4496308166788731e-05,
+      "loss": 1.0206,
+      "step": 4090
+    },
+    {
+      "epoch": 0.4092,
+      "learning_rate": 1.4502542884083868e-05,
+      "loss": 0.6083,
+      "step": 4092
+    },
+    {
+      "epoch": 0.4094,
+      "learning_rate": 1.4508775406894315e-05,
+      "loss": 0.682,
+      "step": 4094
+    },
+    {
+      "epoch": 0.4096,
+      "learning_rate": 1.451500573218237e-05,
+      "loss": 1.0215,
+      "step": 4096
+    },
+    {
+      "epoch": 0.4098,
+      "learning_rate": 1.4521233856911499e-05,
+      "loss": 1.237,
+      "step": 4098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.4527459778046185e-05,
+      "loss": 1.0253,
+      "step": 4100
+    },
+    {
+      "epoch": 0.4102,
+      "learning_rate": 1.4533683492551942e-05,
+      "loss": 0.7195,
+      "step": 4102
+    },
+    {
+      "epoch": 0.4104,
+      "learning_rate": 1.4539904997395458e-05,
+      "loss": 0.7775,
+      "step": 4104
+    },
+    {
+      "epoch": 0.4106,
+      "learning_rate": 1.4546124289544446e-05,
+      "loss": 0.9084,
+      "step": 4106
+    },
+    {
+      "epoch": 0.4108,
+      "learning_rate": 1.4552341365967664e-05,
+      "loss": 0.8296,
+      "step": 4108
+    },
+    {
+      "epoch": 0.411,
+      "learning_rate": 1.4558556223634988e-05,
+      "loss": 1.1081,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4112,
+      "learning_rate": 1.4564768859517417e-05,
+      "loss": 1.2403,
+      "step": 4112
+    },
+    {
+      "epoch": 0.4114,
+      "learning_rate": 1.4570979270586944e-05,
+      "loss": 0.0475,
+      "step": 4114
+    },
+    {
+      "epoch": 0.4116,
+      "learning_rate": 1.4577187453816703e-05,
+      "loss": 0.5526,
+      "step": 4116
+    },
+    {
+      "epoch": 0.4118,
+      "learning_rate": 1.4583393406180886e-05,
+      "loss": 0.0165,
+      "step": 4118
+    },
+    {
+      "epoch": 0.412,
+      "learning_rate": 1.4589597124654834e-05,
+      "loss": 0.0702,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4122,
+      "learning_rate": 1.4595798606214882e-05,
+      "loss": 0.8134,
+      "step": 4122
+    },
+    {
+      "epoch": 0.4124,
+      "learning_rate": 1.4601997847838503e-05,
+      "loss": 1.0517,
+      "step": 4124
+    },
+    {
+      "epoch": 0.4126,
+      "learning_rate": 1.460819484650431e-05,
+      "loss": 0.4552,
+      "step": 4126
+    },
+    {
+      "epoch": 0.4128,
+      "learning_rate": 1.4614389599191917e-05,
+      "loss": 0.76,
+      "step": 4128
+    },
+    {
+      "epoch": 0.413,
+      "learning_rate": 1.4620582102882086e-05,
+      "loss": 0.0394,
+      "step": 4130
+    },
+    {
+      "epoch": 0.4132,
+      "learning_rate": 1.462677235455667e-05,
+      "loss": 0.8698,
+      "step": 4132
+    },
+    {
+      "epoch": 0.4134,
+      "learning_rate": 1.4632960351198618e-05,
+      "loss": 0.5726,
+      "step": 4134
+    },
+    {
+      "epoch": 0.4136,
+      "learning_rate": 1.463914608979197e-05,
+      "loss": 0.5841,
+      "step": 4136
+    },
+    {
+      "epoch": 0.4138,
+      "learning_rate": 1.4645329567321875e-05,
+      "loss": 0.7149,
+      "step": 4138
+    },
+    {
+      "epoch": 0.414,
+      "learning_rate": 1.465151078077458e-05,
+      "loss": 0.4402,
+      "step": 4140
+    },
+    {
+      "epoch": 0.4142,
+      "learning_rate": 1.4657689727137441e-05,
+      "loss": 0.3774,
+      "step": 4142
+    },
+    {
+      "epoch": 0.4144,
+      "learning_rate": 1.466386640339891e-05,
+      "loss": 1.0791,
+      "step": 4144
+    },
+    {
+      "epoch": 0.4146,
+      "learning_rate": 1.4670040806548551e-05,
+      "loss": 0.8347,
+      "step": 4146
+    },
+    {
+      "epoch": 0.4148,
+      "learning_rate": 1.4676212933577034e-05,
+      "loss": 0.9572,
+      "step": 4148
+    },
+    {
+      "epoch": 0.415,
+      "learning_rate": 1.468238278147614e-05,
+      "loss": 0.6601,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4152,
+      "learning_rate": 1.4688550347238763e-05,
+      "loss": 0.6807,
+      "step": 4152
+    },
+    {
+      "epoch": 0.4154,
+      "learning_rate": 1.4694715627858904e-05,
+      "loss": 0.5631,
+      "step": 4154
+    },
+    {
+      "epoch": 0.4156,
+      "learning_rate": 1.4700878620331677e-05,
+      "loss": 0.804,
+      "step": 4156
+    },
+    {
+      "epoch": 0.4158,
+      "learning_rate": 1.470703932165332e-05,
+      "loss": 0.2926,
+      "step": 4158
+    },
+    {
+      "epoch": 0.416,
+      "learning_rate": 1.4713197728821178e-05,
+      "loss": 0.7925,
+      "step": 4160
+    },
+    {
+      "epoch": 0.4162,
+      "learning_rate": 1.471935383883372e-05,
+      "loss": 1.1056,
+      "step": 4162
+    },
+    {
+      "epoch": 0.4164,
+      "learning_rate": 1.4725507648690534e-05,
+      "loss": 0.3803,
+      "step": 4164
+    },
+    {
+      "epoch": 0.4166,
+      "learning_rate": 1.4731659155392339e-05,
+      "loss": 0.5206,
+      "step": 4166
+    },
+    {
+      "epoch": 0.4168,
+      "learning_rate": 1.4737808355940924e-05,
+      "loss": 0.6328,
+      "step": 4168
+    },
+    {
+      "epoch": 0.417,
+      "learning_rate": 1.4743955247339286e-05,
+      "loss": 0.6936,
+      "step": 4170
+    },
+    {
+      "epoch": 0.4172,
+      "learning_rate": 1.4750099826591506e-05,
+      "loss": 0.7331,
+      "step": 4172
+    },
+    {
+      "epoch": 0.4174,
+      "learning_rate": 1.4756242090702744e-05,
+      "loss": 0.7535,
+      "step": 4174
+    },
+    {
+      "epoch": 0.4176,
+      "learning_rate": 1.4762382036679381e-05,
+      "loss": 0.3165,
+      "step": 4176
+    },
+    {
+      "epoch": 0.4178,
+      "learning_rate": 1.476851966152887e-05,
+      "loss": 0.1571,
+      "step": 4178
+    },
+    {
+      "epoch": 0.418,
+      "learning_rate": 1.477465496225982e-05,
+      "loss": 1.4124,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4182,
+      "learning_rate": 1.4780787935881913e-05,
+      "loss": 0.9578,
+      "step": 4182
+    },
+    {
+      "epoch": 0.4184,
+      "learning_rate": 1.4786918579406057e-05,
+      "loss": 0.2272,
+      "step": 4184
+    },
+    {
+      "epoch": 0.4186,
+      "learning_rate": 1.4793046889844255e-05,
+      "loss": 0.6074,
+      "step": 4186
+    },
+    {
+      "epoch": 0.4188,
+      "learning_rate": 1.479917286420961e-05,
+      "loss": 0.5505,
+      "step": 4188
+    },
+    {
+      "epoch": 0.419,
+      "learning_rate": 1.4805296499516397e-05,
+      "loss": 0.5304,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4192,
+      "learning_rate": 1.4811417792780075e-05,
+      "loss": 1.2202,
+      "step": 4192
+    },
+    {
+      "epoch": 0.4194,
+      "learning_rate": 1.4817536741017155e-05,
+      "loss": 0.551,
+      "step": 4194
+    },
+    {
+      "epoch": 0.4196,
+      "learning_rate": 1.4823653341245339e-05,
+      "loss": 1.0077,
+      "step": 4196
+    },
+    {
+      "epoch": 0.4198,
+      "learning_rate": 1.482976759048351e-05,
+      "loss": 0.5521,
+      "step": 4198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.4835879485751617e-05,
+      "loss": 0.0472,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4202,
+      "learning_rate": 1.4841989024070809e-05,
+      "loss": 1.22,
+      "step": 4202
+    },
+    {
+      "epoch": 0.4204,
+      "learning_rate": 1.4848096202463356e-05,
+      "loss": 0.3681,
+      "step": 4204
+    },
+    {
+      "epoch": 0.4206,
+      "learning_rate": 1.485420101795274e-05,
+      "loss": 0.7803,
+      "step": 4206
+    },
+    {
+      "epoch": 0.4208,
+      "learning_rate": 1.4860303467563504e-05,
+      "loss": 0.6836,
+      "step": 4208
+    },
+    {
+      "epoch": 0.421,
+      "learning_rate": 1.4866403548321385e-05,
+      "loss": 0.4814,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4212,
+      "learning_rate": 1.4872501257253325e-05,
+      "loss": 1.1212,
+      "step": 4212
+    },
+    {
+      "epoch": 0.4214,
+      "learning_rate": 1.4878596591387327e-05,
+      "loss": 0.6145,
+      "step": 4214
+    },
+    {
+      "epoch": 0.4216,
+      "learning_rate": 1.4884689547752617e-05,
+      "loss": 0.0173,
+      "step": 4216
+    },
+    {
+      "epoch": 0.4218,
+      "learning_rate": 1.4890780123379563e-05,
+      "loss": 1.0132,
+      "step": 4218
+    },
+    {
+      "epoch": 0.422,
+      "learning_rate": 1.4896868315299689e-05,
+      "loss": 0.781,
+      "step": 4220
+    },
+    {
+      "epoch": 0.4222,
+      "learning_rate": 1.4902954120545686e-05,
+      "loss": 0.6917,
+      "step": 4222
+    },
+    {
+      "epoch": 0.4224,
+      "learning_rate": 1.4909037536151405e-05,
+      "loss": 0.5102,
+      "step": 4224
+    },
+    {
+      "epoch": 0.4226,
+      "learning_rate": 1.491511855915187e-05,
+      "loss": 0.8484,
+      "step": 4226
+    },
+    {
+      "epoch": 0.4228,
+      "learning_rate": 1.4921197186583251e-05,
+      "loss": 0.882,
+      "step": 4228
+    },
+    {
+      "epoch": 0.423,
+      "learning_rate": 1.4927273415482913e-05,
+      "loss": 1.1065,
+      "step": 4230
+    },
+    {
+      "epoch": 0.4232,
+      "learning_rate": 1.4933347242889366e-05,
+      "loss": 0.069,
+      "step": 4232
+    },
+    {
+      "epoch": 0.4234,
+      "learning_rate": 1.4939418665842307e-05,
+      "loss": 0.6195,
+      "step": 4234
+    },
+    {
+      "epoch": 0.4236,
+      "learning_rate": 1.494548768138259e-05,
+      "loss": 1.0915,
+      "step": 4236
+    },
+    {
+      "epoch": 0.4238,
+      "learning_rate": 1.4951554286552261e-05,
+      "loss": 0.6437,
+      "step": 4238
+    },
+    {
+      "epoch": 0.424,
+      "learning_rate": 1.4957618478394522e-05,
+      "loss": 0.1732,
+      "step": 4240
+    },
+    {
+      "epoch": 0.4242,
+      "learning_rate": 1.4963680253953763e-05,
+      "loss": 0.4962,
+      "step": 4242
+    },
+    {
+      "epoch": 0.4244,
+      "learning_rate": 1.4969739610275546e-05,
+      "loss": 0.1759,
+      "step": 4244
+    },
+    {
+      "epoch": 0.4246,
+      "learning_rate": 1.4975796544406617e-05,
+      "loss": 0.4505,
+      "step": 4246
+    },
+    {
+      "epoch": 0.4248,
+      "learning_rate": 1.4981851053394903e-05,
+      "loss": 0.4678,
+      "step": 4248
+    },
+    {
+      "epoch": 0.425,
+      "learning_rate": 1.49879031342895e-05,
+      "loss": 0.5611,
+      "step": 4250
+    },
+    {
+      "epoch": 0.4252,
+      "learning_rate": 1.4993952784140722e-05,
+      "loss": 0.446,
+      "step": 4252
+    },
+    {
+      "epoch": 0.4254,
+      "learning_rate": 1.4999999999999992e-05,
+      "loss": 0.8424,
+      "step": 4254
+    },
+    {
+      "epoch": 0.4256,
+      "learning_rate": 1.5006044778920016e-05,
+      "loss": 1.2241,
+      "step": 4256
+    },
+    {
+      "epoch": 0.4258,
+      "learning_rate": 1.501208711795465e-05,
+      "loss": 0.5397,
+      "step": 4258
+    },
+    {
+      "epoch": 0.426,
+      "learning_rate": 1.5018127014158874e-05,
+      "loss": 0.1946,
+      "step": 4260
+    },
+    {
+      "epoch": 0.4262,
+      "learning_rate": 1.502416446458897e-05,
+      "loss": 0.9243,
+      "step": 4262
+    },
+    {
+      "epoch": 0.4264,
+      "learning_rate": 1.5030199466302357e-05,
+      "loss": 0.6444,
+      "step": 4264
+    },
+    {
+      "epoch": 0.4266,
+      "learning_rate": 1.5036232016357613e-05,
+      "loss": 1.1924,
+      "step": 4266
+    },
+    {
+      "epoch": 0.4268,
+      "learning_rate": 1.5042262111814552e-05,
+      "loss": 0.7569,
+      "step": 4268
+    },
+    {
+      "epoch": 0.427,
+      "learning_rate": 1.5048289749734206e-05,
+      "loss": 0.8744,
+      "step": 4270
+    },
+    {
+      "epoch": 0.4272,
+      "learning_rate": 1.5054314927178779e-05,
+      "loss": 0.338,
+      "step": 4272
+    },
+    {
+      "epoch": 0.4274,
+      "learning_rate": 1.5060337641211642e-05,
+      "loss": 0.4811,
+      "step": 4274
+    },
+    {
+      "epoch": 0.4276,
+      "learning_rate": 1.5066357888897396e-05,
+      "loss": 1.0436,
+      "step": 4276
+    },
+    {
+      "epoch": 0.4278,
+      "learning_rate": 1.5072375667301895e-05,
+      "loss": 0.8197,
+      "step": 4278
+    },
+    {
+      "epoch": 0.428,
+      "learning_rate": 1.5078390973492094e-05,
+      "loss": 0.8051,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4282,
+      "learning_rate": 1.5084403804536214e-05,
+      "loss": 0.0915,
+      "step": 4282
+    },
+    {
+      "epoch": 0.4284,
+      "learning_rate": 1.5090414157503715e-05,
+      "loss": 0.9132,
+      "step": 4284
+    },
+    {
+      "epoch": 0.4286,
+      "learning_rate": 1.5096422029465178e-05,
+      "loss": 0.4084,
+      "step": 4286
+    },
+    {
+      "epoch": 0.4288,
+      "learning_rate": 1.510242741749246e-05,
+      "loss": 0.81,
+      "step": 4288
+    },
+    {
+      "epoch": 0.429,
+      "learning_rate": 1.5108430318658597e-05,
+      "loss": 0.2885,
+      "step": 4290
+    },
+    {
+      "epoch": 0.4292,
+      "learning_rate": 1.511443073003786e-05,
+      "loss": 0.7916,
+      "step": 4292
+    },
+    {
+      "epoch": 0.4294,
+      "learning_rate": 1.5120428648705714e-05,
+      "loss": 0.8923,
+      "step": 4294
+    },
+    {
+      "epoch": 0.4296,
+      "learning_rate": 1.5126424071738836e-05,
+      "loss": 0.3761,
+      "step": 4296
+    },
+    {
+      "epoch": 0.4298,
+      "learning_rate": 1.513241699621517e-05,
+      "loss": 0.7993,
+      "step": 4298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.5138407419213792e-05,
+      "loss": 1.1389,
+      "step": 4300
+    },
+    {
+      "epoch": 0.4302,
+      "learning_rate": 1.5144395337815064e-05,
+      "loss": 0.3969,
+      "step": 4302
+    },
+    {
+      "epoch": 0.4304,
+      "learning_rate": 1.515038074910054e-05,
+      "loss": 0.2218,
+      "step": 4304
+    },
+    {
+      "epoch": 0.4306,
+      "learning_rate": 1.5156363650153008e-05,
+      "loss": 0.6942,
+      "step": 4306
+    },
+    {
+      "epoch": 0.4308,
+      "learning_rate": 1.5162344038056471e-05,
+      "loss": 0.4414,
+      "step": 4308
+    },
+    {
+      "epoch": 0.431,
+      "learning_rate": 1.5168321909896166e-05,
+      "loss": 0.3336,
+      "step": 4310
+    },
+    {
+      "epoch": 0.4312,
+      "learning_rate": 1.5174297262758546e-05,
+      "loss": 0.2963,
+      "step": 4312
+    },
+    {
+      "epoch": 0.4314,
+      "learning_rate": 1.51802700937313e-05,
+      "loss": 1.1014,
+      "step": 4314
+    },
+    {
+      "epoch": 0.4316,
+      "learning_rate": 1.5186240399903338e-05,
+      "loss": 0.8493,
+      "step": 4316
+    },
+    {
+      "epoch": 0.4318,
+      "learning_rate": 1.5192208178364808e-05,
+      "loss": 0.9942,
+      "step": 4318
+    },
+    {
+      "epoch": 0.432,
+      "learning_rate": 1.519817342620709e-05,
+      "loss": 0.338,
+      "step": 4320
+    },
+    {
+      "epoch": 0.4322,
+      "learning_rate": 1.5204136140522792e-05,
+      "loss": 0.1357,
+      "step": 4322
+    },
+    {
+      "epoch": 0.4324,
+      "learning_rate": 1.5210096318405774e-05,
+      "loss": 0.3019,
+      "step": 4324
+    },
+    {
+      "epoch": 0.4326,
+      "learning_rate": 1.521605395695107e-05,
+      "loss": 0.6336,
+      "step": 4326
+    },
+    {
+      "epoch": 0.4328,
+      "learning_rate": 1.5222009053255054e-05,
+      "loss": 0.7856,
+      "step": 4328
+    },
+    {
+      "epoch": 0.433,
+      "learning_rate": 1.522796160441526e-05,
+      "loss": 0.6399,
+      "step": 4330
+    },
+    {
+      "epoch": 0.4332,
+      "learning_rate": 1.523391160753049e-05,
+      "loss": 0.9059,
+      "step": 4332
+    },
+    {
+      "epoch": 0.4334,
+      "learning_rate": 1.5239859059700784e-05,
+      "loss": 0.5126,
+      "step": 4334
+    },
+    {
+      "epoch": 0.4336,
+      "learning_rate": 1.5245803958027427e-05,
+      "loss": 1.1906,
+      "step": 4336
+    },
+    {
+      "epoch": 0.4338,
+      "learning_rate": 1.5251746299612964e-05,
+      "loss": 0.153,
+      "step": 4338
+    },
+    {
+      "epoch": 0.434,
+      "learning_rate": 1.5257686081561122e-05,
+      "loss": 0.4066,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4342,
+      "learning_rate": 1.526362330097697e-05,
+      "loss": 0.146,
+      "step": 4342
+    },
+    {
+      "epoch": 0.4344,
+      "learning_rate": 1.526955795496678e-05,
+      "loss": 0.5695,
+      "step": 4344
+    },
+    {
+      "epoch": 0.4346,
+      "learning_rate": 1.5275490040638044e-05,
+      "loss": 1.0401,
+      "step": 4346
+    },
+    {
+      "epoch": 0.4348,
+      "learning_rate": 1.5281419555099533e-05,
+      "loss": 0.81,
+      "step": 4348
+    },
+    {
+      "epoch": 0.435,
+      "learning_rate": 1.5287346495461322e-05,
+      "loss": 1.1334,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4352,
+      "learning_rate": 1.5293270858834643e-05,
+      "loss": 0.449,
+      "step": 4352
+    },
+    {
+      "epoch": 0.4354,
+      "learning_rate": 1.529919264233204e-05,
+      "loss": 0.1131,
+      "step": 4354
+    },
+    {
+      "epoch": 0.4356,
+      "learning_rate": 1.530511184306733e-05,
+      "loss": 0.8451,
+      "step": 4356
+    },
+    {
+      "epoch": 0.4358,
+      "learning_rate": 1.531102845815557e-05,
+      "loss": 1.1482,
+      "step": 4358
+    },
+    {
+      "epoch": 0.436,
+      "learning_rate": 1.5316942484713043e-05,
+      "loss": 0.4643,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4362,
+      "learning_rate": 1.5322853919857327e-05,
+      "loss": 0.513,
+      "step": 4362
+    },
+    {
+      "epoch": 0.4364,
+      "learning_rate": 1.5328762760707304e-05,
+      "loss": 0.7121,
+      "step": 4364
+    },
+    {
+      "epoch": 0.4366,
+      "learning_rate": 1.5334669004383025e-05,
+      "loss": 0.3694,
+      "step": 4366
+    },
+    {
+      "epoch": 0.4368,
+      "learning_rate": 1.534057264800587e-05,
+      "loss": 0.2365,
+      "step": 4368
+    },
+    {
+      "epoch": 0.437,
+      "learning_rate": 1.5346473688698514e-05,
+      "loss": 0.8222,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4372,
+      "learning_rate": 1.5352372123584816e-05,
+      "loss": 0.4671,
+      "step": 4372
+    },
+    {
+      "epoch": 0.4374,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 0.8775,
+      "step": 4374
+    },
+    {
+      "epoch": 0.4376,
+      "learning_rate": 1.5364161164440413e-05,
+      "loss": 0.4123,
+      "step": 4376
+    },
+    {
+      "epoch": 0.4378,
+      "learning_rate": 1.537005176466387e-05,
+      "loss": 0.889,
+      "step": 4378
+    },
+    {
+      "epoch": 0.438,
+      "learning_rate": 1.537593974758933e-05,
+      "loss": 0.6095,
+      "step": 4380
+    },
+    {
+      "epoch": 0.4382,
+      "learning_rate": 1.5381825110347072e-05,
+      "loss": 1.2619,
+      "step": 4382
+    },
+    {
+      "epoch": 0.4384,
+      "learning_rate": 1.538770785006863e-05,
+      "loss": 0.5079,
+      "step": 4384
+    },
+    {
+      "epoch": 0.4386,
+      "learning_rate": 1.539358796388683e-05,
+      "loss": 0.3705,
+      "step": 4386
+    },
+    {
+      "epoch": 0.4388,
+      "learning_rate": 1.5399465448935785e-05,
+      "loss": 0.3359,
+      "step": 4388
+    },
+    {
+      "epoch": 0.439,
+      "learning_rate": 1.540534030235087e-05,
+      "loss": 0.3946,
+      "step": 4390
+    },
+    {
+      "epoch": 0.4392,
+      "learning_rate": 1.5411212521268753e-05,
+      "loss": 0.3162,
+      "step": 4392
+    },
+    {
+      "epoch": 0.4394,
+      "learning_rate": 1.5417082102827397e-05,
+      "loss": 1.328,
+      "step": 4394
+    },
+    {
+      "epoch": 0.4396,
+      "learning_rate": 1.5422949044166028e-05,
+      "loss": 0.8007,
+      "step": 4396
+    },
+    {
+      "epoch": 0.4398,
+      "learning_rate": 1.542881334242517e-05,
+      "loss": 0.3016,
+      "step": 4398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.5434674994746642e-05,
+      "loss": 0.666,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4402,
+      "learning_rate": 1.5440533998273542e-05,
+      "loss": 1.0226,
+      "step": 4402
+    },
+    {
+      "epoch": 0.4404,
+      "learning_rate": 1.5446390350150265e-05,
+      "loss": 0.9761,
+      "step": 4404
+    },
+    {
+      "epoch": 0.4406,
+      "learning_rate": 1.5452244047522493e-05,
+      "loss": 0.436,
+      "step": 4406
+    },
+    {
+      "epoch": 0.4408,
+      "learning_rate": 1.545809508753721e-05,
+      "loss": 1.0851,
+      "step": 4408
+    },
+    {
+      "epoch": 0.441,
+      "learning_rate": 1.54639434673427e-05,
+      "loss": 0.4528,
+      "step": 4410
+    },
+    {
+      "epoch": 0.4412,
+      "learning_rate": 1.546978918408849e-05,
+      "loss": 0.4942,
+      "step": 4412
+    },
+    {
+      "epoch": 0.4414,
+      "learning_rate": 1.5475632234925495e-05,
+      "loss": 0.4929,
+      "step": 4414
+    },
+    {
+      "epoch": 0.4416,
+      "learning_rate": 1.5481472617005867e-05,
+      "loss": 0.6371,
+      "step": 4416
+    },
+    {
+      "epoch": 0.4418,
+      "learning_rate": 1.548731032748309e-05,
+      "loss": 0.4461,
+      "step": 4418
+    },
+    {
+      "epoch": 0.442,
+      "learning_rate": 1.5493145363511893e-05,
+      "loss": 1.5423,
+      "step": 4420
+    },
+    {
+      "epoch": 0.4422,
+      "learning_rate": 1.5498977722248388e-05,
+      "loss": 0.4189,
+      "step": 4422
+    },
+    {
+      "epoch": 0.4424,
+      "learning_rate": 1.550480740084996e-05,
+      "loss": 0.0927,
+      "step": 4424
+    },
+    {
+      "epoch": 0.4426,
+      "learning_rate": 1.551063439647525e-05,
+      "loss": 0.8923,
+      "step": 4426
+    },
+    {
+      "epoch": 0.4428,
+      "learning_rate": 1.5516458706284293e-05,
+      "loss": 0.7277,
+      "step": 4428
+    },
+    {
+      "epoch": 0.443,
+      "learning_rate": 1.552228032743839e-05,
+      "loss": 0.4788,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4432,
+      "learning_rate": 1.5528099257100133e-05,
+      "loss": 0.078,
+      "step": 4432
+    },
+    {
+      "epoch": 0.4434,
+      "learning_rate": 1.553391549243343e-05,
+      "loss": 0.7094,
+      "step": 4434
+    },
+    {
+      "epoch": 0.4436,
+      "learning_rate": 1.5539729030603577e-05,
+      "loss": 1.0044,
+      "step": 4436
+    },
+    {
+      "epoch": 0.4438,
+      "learning_rate": 1.5545539868777075e-05,
+      "loss": 0.1408,
+      "step": 4438
+    },
+    {
+      "epoch": 0.444,
+      "learning_rate": 1.5551348004121797e-05,
+      "loss": 0.7497,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4442,
+      "learning_rate": 1.5557153433806954e-05,
+      "loss": 0.5897,
+      "step": 4442
+    },
+    {
+      "epoch": 0.4444,
+      "learning_rate": 1.556295615500305e-05,
+      "loss": 1.2312,
+      "step": 4444
+    },
+    {
+      "epoch": 0.4446,
+      "learning_rate": 1.556875616488188e-05,
+      "loss": 0.7953,
+      "step": 4446
+    },
+    {
+      "epoch": 0.4448,
+      "learning_rate": 1.5574553460616594e-05,
+      "loss": 0.6401,
+      "step": 4448
+    },
+    {
+      "epoch": 0.445,
+      "learning_rate": 1.55803480393817e-05,
+      "loss": 0.4421,
+      "step": 4450
+    },
+    {
+      "epoch": 0.4452,
+      "learning_rate": 1.558613989835295e-05,
+      "loss": 0.1039,
+      "step": 4452
+    },
+    {
+      "epoch": 0.4454,
+      "learning_rate": 1.5591929034707468e-05,
+      "loss": 0.7975,
+      "step": 4454
+    },
+    {
+      "epoch": 0.4456,
+      "learning_rate": 1.559771544562371e-05,
+      "loss": 0.0902,
+      "step": 4456
+    },
+    {
+      "epoch": 0.4458,
+      "learning_rate": 1.5603499128281447e-05,
+      "loss": 0.5265,
+      "step": 4458
+    },
+    {
+      "epoch": 0.446,
+      "learning_rate": 1.5609280079861778e-05,
+      "loss": 0.5703,
+      "step": 4460
+    },
+    {
+      "epoch": 0.4462,
+      "learning_rate": 1.5615058297547144e-05,
+      "loss": 0.4338,
+      "step": 4462
+    },
+    {
+      "epoch": 0.4464,
+      "learning_rate": 1.5620833778521306e-05,
+      "loss": 0.7222,
+      "step": 4464
+    },
+    {
+      "epoch": 0.4466,
+      "learning_rate": 1.5626606519969366e-05,
+      "loss": 0.1597,
+      "step": 4466
+    },
+    {
+      "epoch": 0.4468,
+      "learning_rate": 1.5632376519077762e-05,
+      "loss": 1.3639,
+      "step": 4468
+    },
+    {
+      "epoch": 0.447,
+      "learning_rate": 1.5638143773034268e-05,
+      "loss": 1.0375,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4472,
+      "learning_rate": 1.5643908279027987e-05,
+      "loss": 0.4113,
+      "step": 4472
+    },
+    {
+      "epoch": 0.4474,
+      "learning_rate": 1.5649670034249376e-05,
+      "loss": 0.0407,
+      "step": 4474
+    },
+    {
+      "epoch": 0.4476,
+      "learning_rate": 1.5655429035890224e-05,
+      "loss": 0.1759,
+      "step": 4476
+    },
+    {
+      "epoch": 0.4478,
+      "learning_rate": 1.5661185281143663e-05,
+      "loss": 0.5912,
+      "step": 4478
+    },
+    {
+      "epoch": 0.448,
+      "learning_rate": 1.5666938767204166e-05,
+      "loss": 0.5081,
+      "step": 4480
+    },
+    {
+      "epoch": 0.4482,
+      "learning_rate": 1.5672689491267562e-05,
+      "loss": 0.5758,
+      "step": 4482
+    },
+    {
+      "epoch": 0.4484,
+      "learning_rate": 1.5678437450531007e-05,
+      "loss": 0.6109,
+      "step": 4484
+    },
+    {
+      "epoch": 0.4486,
+      "learning_rate": 1.5684182642193024e-05,
+      "loss": 0.8331,
+      "step": 4486
+    },
+    {
+      "epoch": 0.4488,
+      "learning_rate": 1.5689925063453473e-05,
+      "loss": 0.6659,
+      "step": 4488
+    },
+    {
+      "epoch": 0.449,
+      "learning_rate": 1.5695664711513582e-05,
+      "loss": 0.081,
+      "step": 4490
+    },
+    {
+      "epoch": 0.4492,
+      "learning_rate": 1.5701401583575876e-05,
+      "loss": 0.7613,
+      "step": 4492
+    },
+    {
+      "epoch": 0.4494,
+      "learning_rate": 1.5707135676844312e-05,
+      "loss": 0.8807,
+      "step": 4494
+    },
+    {
+      "epoch": 0.4496,
+      "learning_rate": 1.5712866988524164e-05,
+      "loss": 0.9784,
+      "step": 4496
+    },
+    {
+      "epoch": 0.4498,
+      "learning_rate": 1.5718595515822016e-05,
+      "loss": 1.1183,
+      "step": 4498
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.57243212559459e-05,
+      "loss": 0.5289,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4502,
+      "learning_rate": 1.5730044206105146e-05,
+      "loss": 1.0596,
+      "step": 4502
+    },
+    {
+      "epoch": 0.4504,
+      "learning_rate": 1.5735764363510465e-05,
+      "loss": 0.7385,
+      "step": 4504
+    },
+    {
+      "epoch": 0.4506,
+      "learning_rate": 1.574148172537389e-05,
+      "loss": 1.2509,
+      "step": 4506
+    },
+    {
+      "epoch": 0.4508,
+      "learning_rate": 1.574719628890888e-05,
+      "loss": 1.1249,
+      "step": 4508
+    },
+    {
+      "epoch": 0.451,
+      "learning_rate": 1.5752908051330232e-05,
+      "loss": 0.1437,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4512,
+      "learning_rate": 1.5758617009854054e-05,
+      "loss": 0.8037,
+      "step": 4512
+    },
+    {
+      "epoch": 0.4514,
+      "learning_rate": 1.5764323161697923e-05,
+      "loss": 0.9234,
+      "step": 4514
+    },
+    {
+      "epoch": 0.4516,
+      "learning_rate": 1.577002650408072e-05,
+      "loss": 0.6356,
+      "step": 4516
+    },
+    {
+      "epoch": 0.4518,
+      "learning_rate": 1.577572703422268e-05,
+      "loss": 0.4358,
+      "step": 4518
+    },
+    {
+      "epoch": 0.452,
+      "learning_rate": 1.5781424749345434e-05,
+      "loss": 1.137,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4522,
+      "learning_rate": 1.5787119646672025e-05,
+      "loss": 1.1422,
+      "step": 4522
+    },
+    {
+      "epoch": 0.4524,
+      "learning_rate": 1.579281172342679e-05,
+      "loss": 1.0306,
+      "step": 4524
+    },
+    {
+      "epoch": 0.4526,
+      "learning_rate": 1.579850097683548e-05,
+      "loss": 0.8333,
+      "step": 4526
+    },
+    {
+      "epoch": 0.4528,
+      "learning_rate": 1.5804187404125265e-05,
+      "loss": 1.1905,
+      "step": 4528
+    },
+    {
+      "epoch": 0.453,
+      "learning_rate": 1.58098710025246e-05,
+      "loss": 0.5599,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4532,
+      "learning_rate": 1.5815551769263387e-05,
+      "loss": 1.2228,
+      "step": 4532
+    },
+    {
+      "epoch": 0.4534,
+      "learning_rate": 1.582122970157288e-05,
+      "loss": 1.0019,
+      "step": 4534
+    },
+    {
+      "epoch": 0.4536,
+      "learning_rate": 1.582690479668576e-05,
+      "loss": 0.3956,
+      "step": 4536
+    },
+    {
+      "epoch": 0.4538,
+      "learning_rate": 1.5832577051836016e-05,
+      "loss": 2.09,
+      "step": 4538
+    },
+    {
+      "epoch": 0.454,
+      "learning_rate": 1.583824646425907e-05,
+      "loss": 0.5143,
+      "step": 4540
+    },
+    {
+      "epoch": 0.4542,
+      "learning_rate": 1.5843913031191722e-05,
+      "loss": 1.0277,
+      "step": 4542
+    },
+    {
+      "epoch": 0.4544,
+      "learning_rate": 1.5849576749872152e-05,
+      "loss": 0.5897,
+      "step": 4544
+    },
+    {
+      "epoch": 0.4546,
+      "learning_rate": 1.585523761753994e-05,
+      "loss": 1.0795,
+      "step": 4546
+    },
+    {
+      "epoch": 0.4548,
+      "learning_rate": 1.586089563143604e-05,
+      "loss": 1.1917,
+      "step": 4548
+    },
+    {
+      "epoch": 0.455,
+      "learning_rate": 1.586655078880281e-05,
+      "loss": 0.627,
+      "step": 4550
+    },
+    {
+      "epoch": 0.4552,
+      "learning_rate": 1.5872203086883993e-05,
+      "loss": 1.0132,
+      "step": 4552
+    },
+    {
+      "epoch": 0.4554,
+      "learning_rate": 1.587785252292473e-05,
+      "loss": 0.5922,
+      "step": 4554
+    },
+    {
+      "epoch": 0.4556,
+      "learning_rate": 1.588349909417155e-05,
+      "loss": 0.5236,
+      "step": 4556
+    },
+    {
+      "epoch": 0.4558,
+      "learning_rate": 1.5889142797872383e-05,
+      "loss": 0.6929,
+      "step": 4558
+    },
+    {
+      "epoch": 0.456,
+      "learning_rate": 1.589478363127656e-05,
+      "loss": 0.657,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4562,
+      "learning_rate": 1.5900421591634806e-05,
+      "loss": 0.4763,
+      "step": 4562
+    },
+    {
+      "epoch": 0.4564,
+      "learning_rate": 1.590605667619925e-05,
+      "loss": 0.8024,
+      "step": 4564
+    },
+    {
+      "epoch": 0.4566,
+      "learning_rate": 1.5911688882223415e-05,
+      "loss": 0.5092,
+      "step": 4566
+    },
+    {
+      "epoch": 0.4568,
+      "learning_rate": 1.5917318206962235e-05,
+      "loss": 1.007,
+      "step": 4568
+    },
+    {
+      "epoch": 0.457,
+      "learning_rate": 1.5922944647672044e-05,
+      "loss": 0.5411,
+      "step": 4570
+    },
+    {
+      "epoch": 0.4572,
+      "learning_rate": 1.5928568201610587e-05,
+      "loss": 0.4549,
+      "step": 4572
+    },
+    {
+      "epoch": 0.4574,
+      "learning_rate": 1.5934188866037007e-05,
+      "loss": 0.0193,
+      "step": 4574
+    },
+    {
+      "epoch": 0.4576,
+      "learning_rate": 1.593980663821188e-05,
+      "loss": 0.8634,
+      "step": 4576
+    },
+    {
+      "epoch": 0.4578,
+      "learning_rate": 1.5945421515397125e-05,
+      "loss": 0.1204,
+      "step": 4578
+    },
+    {
+      "epoch": 0.458,
+      "learning_rate": 1.5951033494856164e-05,
+      "loss": 1.1454,
+      "step": 4580
+    },
+    {
+      "epoch": 0.4582,
+      "learning_rate": 1.5956642573853787e-05,
+      "loss": 0.5011,
+      "step": 4582
+    },
+    {
+      "epoch": 0.4584,
+      "learning_rate": 1.596224874965615e-05,
+      "loss": 0.2916,
+      "step": 4584
+    },
+    {
+      "epoch": 0.4586,
+      "learning_rate": 1.5967852019530918e-05,
+      "loss": 0.1484,
+      "step": 4586
+    },
+    {
+      "epoch": 0.4588,
+      "learning_rate": 1.5973452380747128e-05,
+      "loss": 0.438,
+      "step": 4588
+    },
+    {
+      "epoch": 0.459,
+      "learning_rate": 1.5979049830575193e-05,
+      "loss": 0.3745,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4592,
+      "learning_rate": 1.5984644366286994e-05,
+      "loss": 0.8553,
+      "step": 4592
+    },
+    {
+      "epoch": 0.4594,
+      "learning_rate": 1.599023598515585e-05,
+      "loss": 0.5074,
+      "step": 4594
+    },
+    {
+      "epoch": 0.4596,
+      "learning_rate": 1.5995824684456468e-05,
+      "loss": 0.2482,
+      "step": 4596
+    },
+    {
+      "epoch": 0.4598,
+      "learning_rate": 1.6001410461464945e-05,
+      "loss": 0.778,
+      "step": 4598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6006993313458883e-05,
+      "loss": 1.0657,
+      "step": 4600
+    },
+    {
+      "epoch": 0.4602,
+      "learning_rate": 1.601257323771727e-05,
+      "loss": 0.4022,
+      "step": 4602
+    },
+    {
+      "epoch": 0.4604,
+      "learning_rate": 1.6018150231520486e-05,
+      "loss": 0.4355,
+      "step": 4604
+    },
+    {
+      "epoch": 0.4606,
+      "learning_rate": 1.6023724292150377e-05,
+      "loss": 1.094,
+      "step": 4606
+    },
+    {
+      "epoch": 0.4608,
+      "learning_rate": 1.602929541689025e-05,
+      "loss": 0.3769,
+      "step": 4608
+    },
+    {
+      "epoch": 0.461,
+      "learning_rate": 1.6034863603024768e-05,
+      "loss": 1.2227,
+      "step": 4610
+    },
+    {
+      "epoch": 0.4612,
+      "learning_rate": 1.6040428847840078e-05,
+      "loss": 0.0003,
+      "step": 4612
+    },
+    {
+      "epoch": 0.4614,
+      "learning_rate": 1.604599114862375e-05,
+      "loss": 0.6189,
+      "step": 4614
+    },
+    {
+      "epoch": 0.4616,
+      "learning_rate": 1.605155050266478e-05,
+      "loss": 0.9045,
+      "step": 4616
+    },
+    {
+      "epoch": 0.4618,
+      "learning_rate": 1.6057106907253614e-05,
+      "loss": 0.4142,
+      "step": 4618
+    },
+    {
+      "epoch": 0.462,
+      "learning_rate": 1.606266035968211e-05,
+      "loss": 1.1323,
+      "step": 4620
+    },
+    {
+      "epoch": 0.4622,
+      "learning_rate": 1.606821085724362e-05,
+      "loss": 0.6853,
+      "step": 4622
+    },
+    {
+      "epoch": 0.4624,
+      "learning_rate": 1.607375839723287e-05,
+      "loss": 0.142,
+      "step": 4624
+    },
+    {
+      "epoch": 0.4626,
+      "learning_rate": 1.6079302976946052e-05,
+      "loss": 0.8922,
+      "step": 4626
+    },
+    {
+      "epoch": 0.4628,
+      "learning_rate": 1.608484459368082e-05,
+      "loss": 0.5856,
+      "step": 4628
+    },
+    {
+      "epoch": 0.463,
+      "learning_rate": 1.6090383244736253e-05,
+      "loss": 0.6174,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4632,
+      "learning_rate": 1.609591892741288e-05,
+      "loss": 0.498,
+      "step": 4632
+    },
+    {
+      "epoch": 0.4634,
+      "learning_rate": 1.6101451639012675e-05,
+      "loss": 0.2184,
+      "step": 4634
+    },
+    {
+      "epoch": 0.4636,
+      "learning_rate": 1.610698137683906e-05,
+      "loss": 0.0692,
+      "step": 4636
+    },
+    {
+      "epoch": 0.4638,
+      "learning_rate": 1.6112508138196912e-05,
+      "loss": 0.2929,
+      "step": 4638
+    },
+    {
+      "epoch": 0.464,
+      "learning_rate": 1.6118031920392555e-05,
+      "loss": 0.7306,
+      "step": 4640
+    },
+    {
+      "epoch": 0.4642,
+      "learning_rate": 1.6123552720733763e-05,
+      "loss": 0.6077,
+      "step": 4642
+    },
+    {
+      "epoch": 0.4644,
+      "learning_rate": 1.6129070536529763e-05,
+      "loss": 1.1394,
+      "step": 4644
+    },
+    {
+      "epoch": 0.4646,
+      "learning_rate": 1.613458536509124e-05,
+      "loss": 0.9547,
+      "step": 4646
+    },
+    {
+      "epoch": 0.4648,
+      "learning_rate": 1.6140097203730347e-05,
+      "loss": 0.9407,
+      "step": 4648
+    },
+    {
+      "epoch": 0.465,
+      "learning_rate": 1.614560604976064e-05,
+      "loss": 1.3551,
+      "step": 4650
+    },
+    {
+      "epoch": 0.4652,
+      "learning_rate": 1.6151111900497215e-05,
+      "loss": 0.4113,
+      "step": 4652
+    },
+    {
+      "epoch": 0.4654,
+      "learning_rate": 1.615661475325658e-05,
+      "loss": 0.7502,
+      "step": 4654
+    },
+    {
+      "epoch": 0.4656,
+      "learning_rate": 1.61621146053567e-05,
+      "loss": 1.6514,
+      "step": 4656
+    },
+    {
+      "epoch": 0.4658,
+      "learning_rate": 1.616761145411702e-05,
+      "loss": 0.2619,
+      "step": 4658
+    },
+    {
+      "epoch": 0.466,
+      "learning_rate": 1.6173105296858444e-05,
+      "loss": 0.2818,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4662,
+      "learning_rate": 1.6178596130903352e-05,
+      "loss": 1.0056,
+      "step": 4662
+    },
+    {
+      "epoch": 0.4664,
+      "learning_rate": 1.6184083953575536e-05,
+      "loss": 0.1599,
+      "step": 4664
+    },
+    {
+      "epoch": 0.4666,
+      "learning_rate": 1.618956876220034e-05,
+      "loss": 0.5448,
+      "step": 4666
+    },
+    {
+      "epoch": 0.4668,
+      "learning_rate": 1.6195050554104536e-05,
+      "loss": 0.4948,
+      "step": 4668
+    },
+    {
+      "epoch": 0.467,
+      "learning_rate": 1.620052932661632e-05,
+      "loss": 0.6748,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4672,
+      "learning_rate": 1.6206005077065447e-05,
+      "loss": 0.6631,
+      "step": 4672
+    },
+    {
+      "epoch": 0.4674,
+      "learning_rate": 1.621147780278311e-05,
+      "loss": 1.2028,
+      "step": 4674
+    },
+    {
+      "epoch": 0.4676,
+      "learning_rate": 1.6216947501101933e-05,
+      "loss": 0.836,
+      "step": 4676
+    },
+    {
+      "epoch": 0.4678,
+      "learning_rate": 1.6222414169356056e-05,
+      "loss": 1.0624,
+      "step": 4678
+    },
+    {
+      "epoch": 0.468,
+      "learning_rate": 1.6227877804881116e-05,
+      "loss": 1.0547,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4682,
+      "learning_rate": 1.6233338405014204e-05,
+      "loss": 0.1753,
+      "step": 4682
+    },
+    {
+      "epoch": 0.4684,
+      "learning_rate": 1.6238795967093865e-05,
+      "loss": 1.0729,
+      "step": 4684
+    },
+    {
+      "epoch": 0.4686,
+      "learning_rate": 1.6244250488460146e-05,
+      "loss": 1.183,
+      "step": 4686
+    },
+    {
+      "epoch": 0.4688,
+      "learning_rate": 1.6249701966454626e-05,
+      "loss": 0.6149,
+      "step": 4688
+    },
+    {
+      "epoch": 0.469,
+      "learning_rate": 1.6255150398420273e-05,
+      "loss": 0.9201,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4692,
+      "learning_rate": 1.6260595781701592e-05,
+      "loss": 0.6249,
+      "step": 4692
+    },
+    {
+      "epoch": 0.4694,
+      "learning_rate": 1.6266038113644605e-05,
+      "loss": 0.9035,
+      "step": 4694
+    },
+    {
+      "epoch": 0.4696,
+      "learning_rate": 1.6271477391596754e-05,
+      "loss": 0.6188,
+      "step": 4696
+    },
+    {
+      "epoch": 0.4698,
+      "learning_rate": 1.6276913612907005e-05,
+      "loss": 0.927,
+      "step": 4698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6282346774925816e-05,
+      "loss": 0.7763,
+      "step": 4700
+    },
+    {
+      "epoch": 0.4702,
+      "learning_rate": 1.6287776875005127e-05,
+      "loss": 0.4574,
+      "step": 4702
+    },
+    {
+      "epoch": 0.4704,
+      "learning_rate": 1.6293203910498375e-05,
+      "loss": 0.6704,
+      "step": 4704
+    },
+    {
+      "epoch": 0.4706,
+      "learning_rate": 1.6298627878760488e-05,
+      "loss": 1.0952,
+      "step": 4706
+    },
+    {
+      "epoch": 0.4708,
+      "learning_rate": 1.6304048777147887e-05,
+      "loss": 0.326,
+      "step": 4708
+    },
+    {
+      "epoch": 0.471,
+      "learning_rate": 1.6309466603018497e-05,
+      "loss": 0.8348,
+      "step": 4710
+    },
+    {
+      "epoch": 0.4712,
+      "learning_rate": 1.631488135373173e-05,
+      "loss": 0.5042,
+      "step": 4712
+    },
+    {
+      "epoch": 0.4714,
+      "learning_rate": 1.6320293026648508e-05,
+      "loss": 0.7304,
+      "step": 4714
+    },
+    {
+      "epoch": 0.4716,
+      "learning_rate": 1.6325701619131243e-05,
+      "loss": 0.3688,
+      "step": 4716
+    },
+    {
+      "epoch": 0.4718,
+      "learning_rate": 1.6331107128543856e-05,
+      "loss": 0.9338,
+      "step": 4718
+    },
+    {
+      "epoch": 0.472,
+      "learning_rate": 1.6336509552251763e-05,
+      "loss": 0.7092,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4722,
+      "learning_rate": 1.634190888762189e-05,
+      "loss": 1.0564,
+      "step": 4722
+    },
+    {
+      "epoch": 0.4724,
+      "learning_rate": 1.6347305132022674e-05,
+      "loss": 0.3267,
+      "step": 4724
+    },
+    {
+      "epoch": 0.4726,
+      "learning_rate": 1.635269828282404e-05,
+      "loss": 0.8322,
+      "step": 4726
+    },
+    {
+      "epoch": 0.4728,
+      "learning_rate": 1.6358088337397437e-05,
+      "loss": 0.4251,
+      "step": 4728
+    },
+    {
+      "epoch": 0.473,
+      "learning_rate": 1.6363475293115818e-05,
+      "loss": 0.3022,
+      "step": 4730
+    },
+    {
+      "epoch": 0.4732,
+      "learning_rate": 1.6368859147353642e-05,
+      "loss": 0.4053,
+      "step": 4732
+    },
+    {
+      "epoch": 0.4734,
+      "learning_rate": 1.6374239897486905e-05,
+      "loss": 0.1159,
+      "step": 4734
+    },
+    {
+      "epoch": 0.4736,
+      "learning_rate": 1.637961754089305e-05,
+      "loss": 0.3582,
+      "step": 4736
+    },
+    {
+      "epoch": 0.4738,
+      "learning_rate": 1.6384992074951118e-05,
+      "loss": 1.3968,
+      "step": 4738
+    },
+    {
+      "epoch": 0.474,
+      "learning_rate": 1.6390363497041614e-05,
+      "loss": 0.6968,
+      "step": 4740
+    },
+    {
+      "epoch": 0.4742,
+      "learning_rate": 1.6395731804546575e-05,
+      "loss": 0.0905,
+      "step": 4742
+    },
+    {
+      "epoch": 0.4744,
+      "learning_rate": 1.6401096994849548e-05,
+      "loss": 0.9006,
+      "step": 4744
+    },
+    {
+      "epoch": 0.4746,
+      "learning_rate": 1.640645906533561e-05,
+      "loss": 0.5432,
+      "step": 4746
+    },
+    {
+      "epoch": 0.4748,
+      "learning_rate": 1.641181801339136e-05,
+      "loss": 0.8531,
+      "step": 4748
+    },
+    {
+      "epoch": 0.475,
+      "learning_rate": 1.6417173836404878e-05,
+      "loss": 0.7349,
+      "step": 4750
+    },
+    {
+      "epoch": 0.4752,
+      "learning_rate": 1.6422526531765836e-05,
+      "loss": 1.2384,
+      "step": 4752
+    },
+    {
+      "epoch": 0.4754,
+      "learning_rate": 1.6427876096865397e-05,
+      "loss": 0.9486,
+      "step": 4754
+    },
+    {
+      "epoch": 0.4756,
+      "learning_rate": 1.643322252909621e-05,
+      "loss": 0.8719,
+      "step": 4756
+    },
+    {
+      "epoch": 0.4758,
+      "learning_rate": 1.643856582585253e-05,
+      "loss": 1.4088,
+      "step": 4758
+    },
+    {
+      "epoch": 0.476,
+      "learning_rate": 1.6443905984530092e-05,
+      "loss": 0.5806,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4762,
+      "learning_rate": 1.6449243002526146e-05,
+      "loss": 1.1422,
+      "step": 4762
+    },
+    {
+      "epoch": 0.4764,
+      "learning_rate": 1.6454576877239495e-05,
+      "loss": 0.4522,
+      "step": 4764
+    },
+    {
+      "epoch": 0.4766,
+      "learning_rate": 1.6459907606070513e-05,
+      "loss": 0.8294,
+      "step": 4766
+    },
+    {
+      "epoch": 0.4768,
+      "learning_rate": 1.6465235186421024e-05,
+      "loss": 0.6026,
+      "step": 4768
+    },
+    {
+      "epoch": 0.477,
+      "learning_rate": 1.6470559615694445e-05,
+      "loss": 0.8923,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4772,
+      "learning_rate": 1.6475880891295702e-05,
+      "loss": 0.7269,
+      "step": 4772
+    },
+    {
+      "epoch": 0.4774,
+      "learning_rate": 1.6481199010631312e-05,
+      "loss": 0.68,
+      "step": 4774
+    },
+    {
+      "epoch": 0.4776,
+      "learning_rate": 1.648651397110924e-05,
+      "loss": 0.4557,
+      "step": 4776
+    },
+    {
+      "epoch": 0.4778,
+      "learning_rate": 1.649182577013905e-05,
+      "loss": 0.6797,
+      "step": 4778
+    },
+    {
+      "epoch": 0.478,
+      "learning_rate": 1.649713440513187e-05,
+      "loss": 1.4137,
+      "step": 4780
+    },
+    {
+      "epoch": 0.4782,
+      "learning_rate": 1.650243987350029e-05,
+      "loss": 0.9078,
+      "step": 4782
+    },
+    {
+      "epoch": 0.4784,
+      "learning_rate": 1.650774217265851e-05,
+      "loss": 0.7481,
+      "step": 4784
+    },
+    {
+      "epoch": 0.4786,
+      "learning_rate": 1.6513041300022253e-05,
+      "loss": 0.9319,
+      "step": 4786
+    },
+    {
+      "epoch": 0.4788,
+      "learning_rate": 1.6518337253008787e-05,
+      "loss": 2.7572,
+      "step": 4788
+    },
+    {
+      "epoch": 0.479,
+      "learning_rate": 1.652363002903693e-05,
+      "loss": 0.4725,
+      "step": 4790
+    },
+    {
+      "epoch": 0.4792,
+      "learning_rate": 1.6528919625527047e-05,
+      "loss": 0.581,
+      "step": 4792
+    },
+    {
+      "epoch": 0.4794,
+      "learning_rate": 1.6534206039901054e-05,
+      "loss": 0.4585,
+      "step": 4794
+    },
+    {
+      "epoch": 0.4796,
+      "learning_rate": 1.6539489269582414e-05,
+      "loss": 0.2324,
+      "step": 4796
+    },
+    {
+      "epoch": 0.4798,
+      "learning_rate": 1.6544769311996146e-05,
+      "loss": 0.9985,
+      "step": 4798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.6550046164568823e-05,
+      "loss": 0.6177,
+      "step": 4800
+    },
+    {
+      "epoch": 0.4802,
+      "learning_rate": 1.655531982472857e-05,
+      "loss": 1.2434,
+      "step": 4802
+    },
+    {
+      "epoch": 0.4804,
+      "learning_rate": 1.656059028990507e-05,
+      "loss": 0.5329,
+      "step": 4804
+    },
+    {
+      "epoch": 0.4806,
+      "learning_rate": 1.656585755752956e-05,
+      "loss": 1.2959,
+      "step": 4806
+    },
+    {
+      "epoch": 0.4808,
+      "learning_rate": 1.6571121625034843e-05,
+      "loss": 0.6861,
+      "step": 4808
+    },
+    {
+      "epoch": 0.481,
+      "learning_rate": 1.657638248985527e-05,
+      "loss": 0.6016,
+      "step": 4810
+    },
+    {
+      "epoch": 0.4812,
+      "learning_rate": 1.6581640149426763e-05,
+      "loss": 1.1006,
+      "step": 4812
+    },
+    {
+      "epoch": 0.4814,
+      "learning_rate": 1.65868946011868e-05,
+      "loss": 0.4904,
+      "step": 4814
+    },
+    {
+      "epoch": 0.4816,
+      "learning_rate": 1.659214584257443e-05,
+      "loss": 0.863,
+      "step": 4816
+    },
+    {
+      "epoch": 0.4818,
+      "learning_rate": 1.6597393871030257e-05,
+      "loss": 0.6517,
+      "step": 4818
+    },
+    {
+      "epoch": 0.482,
+      "learning_rate": 1.660263868399647e-05,
+      "loss": 0.7301,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4822,
+      "learning_rate": 1.660788027891677e-05,
+      "loss": 0.9025,
+      "step": 4822
+    },
+    {
+      "epoch": 0.4824,
+      "learning_rate": 1.6613118653236513e-05,
+      "loss": 0.8878,
+      "step": 4824
+    },
+    {
+      "epoch": 0.4826,
+      "learning_rate": 1.6618353804402573e-05,
+      "loss": 0.8559,
+      "step": 4826
+    },
+    {
+      "epoch": 0.4828,
+      "learning_rate": 1.662358572986336e-05,
+      "loss": 0.9701,
+      "step": 4828
+    },
+    {
+      "epoch": 0.483,
+      "learning_rate": 1.6628814427068944e-05,
+      "loss": 0.8296,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4832,
+      "learning_rate": 1.6634039893470902e-05,
+      "loss": 0.7792,
+      "step": 4832
+    },
+    {
+      "epoch": 0.4834,
+      "learning_rate": 1.663926212652242e-05,
+      "loss": 0.1946,
+      "step": 4834
+    },
+    {
+      "epoch": 0.4836,
+      "learning_rate": 1.6644481123678206e-05,
+      "loss": 0.68,
+      "step": 4836
+    },
+    {
+      "epoch": 0.4838,
+      "learning_rate": 1.6649696882394625e-05,
+      "loss": 0.4659,
+      "step": 4838
+    },
+    {
+      "epoch": 0.484,
+      "learning_rate": 1.665490940012958e-05,
+      "loss": 0.6142,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4842,
+      "learning_rate": 1.666011867434252e-05,
+      "loss": 1.4057,
+      "step": 4842
+    },
+    {
+      "epoch": 0.4844,
+      "learning_rate": 1.6665324702494517e-05,
+      "loss": 0.9208,
+      "step": 4844
+    },
+    {
+      "epoch": 0.4846,
+      "learning_rate": 1.667052748204825e-05,
+      "loss": 0.5762,
+      "step": 4846
+    },
+    {
+      "epoch": 0.4848,
+      "learning_rate": 1.667572701046791e-05,
+      "loss": 0.561,
+      "step": 4848
+    },
+    {
+      "epoch": 0.485,
+      "learning_rate": 1.6680923285219308e-05,
+      "loss": 0.6001,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4852,
+      "learning_rate": 1.6686116303769884e-05,
+      "loss": 0.6044,
+      "step": 4852
+    },
+    {
+      "epoch": 0.4854,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.7218,
+      "step": 4854
+    },
+    {
+      "epoch": 0.4856,
+      "learning_rate": 1.6696492562145996e-05,
+      "loss": 0.736,
+      "step": 4856
+    },
+    {
+      "epoch": 0.4858,
+      "learning_rate": 1.6701675796914273e-05,
+      "loss": 0.4332,
+      "step": 4858
+    },
+    {
+      "epoch": 0.486,
+      "learning_rate": 1.6706855765367202e-05,
+      "loss": 0.3239,
+      "step": 4860
+    },
+    {
+      "epoch": 0.4862,
+      "learning_rate": 1.6712032464980094e-05,
+      "loss": 0.8499,
+      "step": 4862
+    },
+    {
+      "epoch": 0.4864,
+      "learning_rate": 1.6717205893229887e-05,
+      "loss": 0.7193,
+      "step": 4864
+    },
+    {
+      "epoch": 0.4866,
+      "learning_rate": 1.672237604759516e-05,
+      "loss": 0.4909,
+      "step": 4866
+    },
+    {
+      "epoch": 0.4868,
+      "learning_rate": 1.6727542925556e-05,
+      "loss": 0.8322,
+      "step": 4868
+    },
+    {
+      "epoch": 0.487,
+      "learning_rate": 1.6732706524594138e-05,
+      "loss": 0.11,
+      "step": 4870
+    },
+    {
+      "epoch": 0.4872,
+      "learning_rate": 1.6737866842192908e-05,
+      "loss": 0.537,
+      "step": 4872
+    },
+    {
+      "epoch": 0.4874,
+      "learning_rate": 1.6743023875837233e-05,
+      "loss": 1.4523,
+      "step": 4874
+    },
+    {
+      "epoch": 0.4876,
+      "learning_rate": 1.6748177623013635e-05,
+      "loss": 0.7209,
+      "step": 4876
+    },
+    {
+      "epoch": 0.4878,
+      "learning_rate": 1.6753328081210244e-05,
+      "loss": 0.2256,
+      "step": 4878
+    },
+    {
+      "epoch": 0.488,
+      "learning_rate": 1.6758475247916783e-05,
+      "loss": 0.5102,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4882,
+      "learning_rate": 1.6763619120624592e-05,
+      "loss": 0.6257,
+      "step": 4882
+    },
+    {
+      "epoch": 0.4884,
+      "learning_rate": 1.6768759696826605e-05,
+      "loss": 1.196,
+      "step": 4884
+    },
+    {
+      "epoch": 0.4886,
+      "learning_rate": 1.6773896974017373e-05,
+      "loss": 0.7225,
+      "step": 4886
+    },
+    {
+      "epoch": 0.4888,
+      "learning_rate": 1.677903094969304e-05,
+      "loss": 0.9494,
+      "step": 4888
+    },
+    {
+      "epoch": 0.489,
+      "learning_rate": 1.6784161621351377e-05,
+      "loss": 0.0902,
+      "step": 4890
+    },
+    {
+      "epoch": 0.4892,
+      "learning_rate": 1.6789288986491758e-05,
+      "loss": 0.5346,
+      "step": 4892
+    },
+    {
+      "epoch": 0.4894,
+      "learning_rate": 1.679441304261516e-05,
+      "loss": 1.2416,
+      "step": 4894
+    },
+    {
+      "epoch": 0.4896,
+      "learning_rate": 1.6799533787224186e-05,
+      "loss": 0.6112,
+      "step": 4896
+    },
+    {
+      "epoch": 0.4898,
+      "learning_rate": 1.6804651217823048e-05,
+      "loss": 1.0033,
+      "step": 4898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.680976533191757e-05,
+      "loss": 0.635,
+      "step": 4900
+    },
+    {
+      "epoch": 0.4902,
+      "learning_rate": 1.681487612701519e-05,
+      "loss": 1.2309,
+      "step": 4902
+    },
+    {
+      "epoch": 0.4904,
+      "learning_rate": 1.681998360062498e-05,
+      "loss": 0.6057,
+      "step": 4904
+    },
+    {
+      "epoch": 0.4906,
+      "learning_rate": 1.6825087750257624e-05,
+      "loss": 1.074,
+      "step": 4906
+    },
+    {
+      "epoch": 0.4908,
+      "learning_rate": 1.6830188573425382e-05,
+      "loss": 0.921,
+      "step": 4908
+    },
+    {
+      "epoch": 0.491,
+      "learning_rate": 1.683528606764222e-05,
+      "loss": 1.1538,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4912,
+      "learning_rate": 1.6840380230423676e-05,
+      "loss": 0.9872,
+      "step": 4912
+    },
+    {
+      "epoch": 0.4914,
+      "learning_rate": 1.6845471059286893e-05,
+      "loss": 0.5783,
+      "step": 4914
+    },
+    {
+      "epoch": 0.4916,
+      "learning_rate": 1.685055855175066e-05,
+      "loss": 0.4505,
+      "step": 4916
+    },
+    {
+      "epoch": 0.4918,
+      "learning_rate": 1.6855642705335428e-05,
+      "loss": 1.1305,
+      "step": 4918
+    },
+    {
+      "epoch": 0.492,
+      "learning_rate": 1.6860723517563235e-05,
+      "loss": 0.2436,
+      "step": 4920
+    },
+    {
+      "epoch": 0.4922,
+      "learning_rate": 1.6865800985957718e-05,
+      "loss": 0.9211,
+      "step": 4922
+    },
+    {
+      "epoch": 0.4924,
+      "learning_rate": 1.687087510804422e-05,
+      "loss": 0.8298,
+      "step": 4924
+    },
+    {
+      "epoch": 0.4926,
+      "learning_rate": 1.687594588134968e-05,
+      "loss": 0.5613,
+      "step": 4926
+    },
+    {
+      "epoch": 0.4928,
+      "learning_rate": 1.688101330340263e-05,
+      "loss": 0.126,
+      "step": 4928
+    },
+    {
+      "epoch": 0.493,
+      "learning_rate": 1.6886077371733275e-05,
+      "loss": 0.88,
+      "step": 4930
+    },
+    {
+      "epoch": 0.4932,
+      "learning_rate": 1.6891138083873486e-05,
+      "loss": 0.8991,
+      "step": 4932
+    },
+    {
+      "epoch": 0.4934,
+      "learning_rate": 1.68961954373567e-05,
+      "loss": 0.4666,
+      "step": 4934
+    },
+    {
+      "epoch": 0.4936,
+      "learning_rate": 1.690124942971802e-05,
+      "loss": 0.6236,
+      "step": 4936
+    },
+    {
+      "epoch": 0.4938,
+      "learning_rate": 1.690630005849423e-05,
+      "loss": 0.9559,
+      "step": 4938
+    },
+    {
+      "epoch": 0.494,
+      "learning_rate": 1.691134732122368e-05,
+      "loss": 0.7848,
+      "step": 4940
+    },
+    {
+      "epoch": 0.4942,
+      "learning_rate": 1.6916391215446403e-05,
+      "loss": 1.162,
+      "step": 4942
+    },
+    {
+      "epoch": 0.4944,
+      "learning_rate": 1.6921431738704056e-05,
+      "loss": 1.2839,
+      "step": 4944
+    },
+    {
+      "epoch": 0.4946,
+      "learning_rate": 1.6926468888539988e-05,
+      "loss": 1.1284,
+      "step": 4946
+    },
+    {
+      "epoch": 0.4948,
+      "learning_rate": 1.6931502662499116e-05,
+      "loss": 0.3261,
+      "step": 4948
+    },
+    {
+      "epoch": 0.495,
+      "learning_rate": 1.693653305812805e-05,
+      "loss": 0.5532,
+      "step": 4950
+    },
+    {
+      "epoch": 0.4952,
+      "learning_rate": 1.6941560072975037e-05,
+      "loss": 0.7245,
+      "step": 4952
+    },
+    {
+      "epoch": 0.4954,
+      "learning_rate": 1.6946583704589973e-05,
+      "loss": 0.4874,
+      "step": 4954
+    },
+    {
+      "epoch": 0.4956,
+      "learning_rate": 1.69516039505244e-05,
+      "loss": 1.1326,
+      "step": 4956
+    },
+    {
+      "epoch": 0.4958,
+      "learning_rate": 1.6956620808331505e-05,
+      "loss": 1.0361,
+      "step": 4958
+    },
+    {
+      "epoch": 0.496,
+      "learning_rate": 1.6961634275566144e-05,
+      "loss": 1.4247,
+      "step": 4960
+    },
+    {
+      "epoch": 0.4962,
+      "learning_rate": 1.6966644349784805e-05,
+      "loss": 0.7915,
+      "step": 4962
+    },
+    {
+      "epoch": 0.4964,
+      "learning_rate": 1.6971651028545645e-05,
+      "loss": 0.2669,
+      "step": 4964
+    },
+    {
+      "epoch": 0.4966,
+      "learning_rate": 1.697665430940846e-05,
+      "loss": 1.6265,
+      "step": 4966
+    },
+    {
+      "epoch": 0.4968,
+      "learning_rate": 1.6981654189934726e-05,
+      "loss": 1.332,
+      "step": 4968
+    },
+    {
+      "epoch": 0.497,
+      "learning_rate": 1.698665066768755e-05,
+      "loss": 0.6792,
+      "step": 4970
+    },
+    {
+      "epoch": 0.4972,
+      "learning_rate": 1.699164374023171e-05,
+      "loss": 0.8436,
+      "step": 4972
+    },
+    {
+      "epoch": 0.4974,
+      "learning_rate": 1.699663340513365e-05,
+      "loss": 1.1987,
+      "step": 4974
+    },
+    {
+      "epoch": 0.4976,
+      "learning_rate": 1.7001619659961464e-05,
+      "loss": 0.618,
+      "step": 4976
+    },
+    {
+      "epoch": 0.4978,
+      "learning_rate": 1.7006602502284913e-05,
+      "loss": 0.7405,
+      "step": 4978
+    },
+    {
+      "epoch": 0.498,
+      "learning_rate": 1.701158192967542e-05,
+      "loss": 0.7183,
+      "step": 4980
+    },
+    {
+      "epoch": 0.4982,
+      "learning_rate": 1.7016557939706068e-05,
+      "loss": 0.9725,
+      "step": 4982
+    },
+    {
+      "epoch": 0.4984,
+      "learning_rate": 1.7021530529951617e-05,
+      "loss": 0.3656,
+      "step": 4984
+    },
+    {
+      "epoch": 0.4986,
+      "learning_rate": 1.70264996979885e-05,
+      "loss": 0.4061,
+      "step": 4986
+    },
+    {
+      "epoch": 0.4988,
+      "learning_rate": 1.703146544139476e-05,
+      "loss": 0.0297,
+      "step": 4988
+    },
+    {
+      "epoch": 0.499,
+      "learning_rate": 1.7036427757750198e-05,
+      "loss": 0.5103,
+      "step": 4990
+    },
+    {
+      "epoch": 0.4992,
+      "learning_rate": 1.7041386644636238e-05,
+      "loss": 1.1174,
+      "step": 4992
+    },
+    {
+      "epoch": 0.4994,
+      "learning_rate": 1.7046342099635938e-05,
+      "loss": 1.1631,
+      "step": 4994
+    },
+    {
+      "epoch": 0.4996,
+      "learning_rate": 1.7051294120334116e-05,
+      "loss": 0.9564,
+      "step": 4996
+    },
+    {
+      "epoch": 0.4998,
+      "learning_rate": 1.7056242704317212e-05,
+      "loss": 0.9553,
+      "step": 4998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.706118784917332e-05,
+      "loss": 0.4561,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5002,
+      "learning_rate": 1.706612955249224e-05,
+      "loss": 1.1596,
+      "step": 5002
+    },
+    {
+      "epoch": 0.5004,
+      "learning_rate": 1.7071067811865467e-05,
+      "loss": 0.617,
+      "step": 5004
+    },
+    {
+      "epoch": 0.5006,
+      "learning_rate": 1.7076002624886156e-05,
+      "loss": 0.3822,
+      "step": 5006
+    },
+    {
+      "epoch": 0.5008,
+      "learning_rate": 1.70809339891491e-05,
+      "loss": 1.1268,
+      "step": 5008
+    },
+    {
+      "epoch": 0.501,
+      "learning_rate": 1.708586190225085e-05,
+      "loss": 0.0987,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5012,
+      "learning_rate": 1.7090786361789606e-05,
+      "loss": 0.5814,
+      "step": 5012
+    },
+    {
+      "epoch": 0.5014,
+      "learning_rate": 1.709570736536521e-05,
+      "loss": 0.7264,
+      "step": 5014
+    },
+    {
+      "epoch": 0.5016,
+      "learning_rate": 1.7100624910579235e-05,
+      "loss": 0.028,
+      "step": 5016
+    },
+    {
+      "epoch": 0.5018,
+      "learning_rate": 1.710553899503496e-05,
+      "loss": 0.8199,
+      "step": 5018
+    },
+    {
+      "epoch": 0.502,
+      "learning_rate": 1.711044961633729e-05,
+      "loss": 0.5077,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5022,
+      "learning_rate": 1.7115356772092844e-05,
+      "loss": 0.584,
+      "step": 5022
+    },
+    {
+      "epoch": 0.5024,
+      "learning_rate": 1.712026045990997e-05,
+      "loss": 0.5381,
+      "step": 5024
+    },
+    {
+      "epoch": 0.5026,
+      "learning_rate": 1.7125160677398625e-05,
+      "loss": 0.7683,
+      "step": 5026
+    },
+    {
+      "epoch": 0.5028,
+      "learning_rate": 1.713005742217053e-05,
+      "loss": 0.4843,
+      "step": 5028
+    },
+    {
+      "epoch": 0.503,
+      "learning_rate": 1.7134950691839063e-05,
+      "loss": 0.7033,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5032,
+      "learning_rate": 1.7139840484019306e-05,
+      "loss": 0.8213,
+      "step": 5032
+    },
+    {
+      "epoch": 0.5034,
+      "learning_rate": 1.7144726796328034e-05,
+      "loss": 0.5283,
+      "step": 5034
+    },
+    {
+      "epoch": 0.5036,
+      "learning_rate": 1.7149609626383714e-05,
+      "loss": 0.5189,
+      "step": 5036
+    },
+    {
+      "epoch": 0.5038,
+      "learning_rate": 1.7154488971806518e-05,
+      "loss": 0.944,
+      "step": 5038
+    },
+    {
+      "epoch": 0.504,
+      "learning_rate": 1.7159364830218312e-05,
+      "loss": 0.1303,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5042,
+      "learning_rate": 1.716423719924266e-05,
+      "loss": 0.5659,
+      "step": 5042
+    },
+    {
+      "epoch": 0.5044,
+      "learning_rate": 1.7169106076504823e-05,
+      "loss": 0.9141,
+      "step": 5044
+    },
+    {
+      "epoch": 0.5046,
+      "learning_rate": 1.7173971459631783e-05,
+      "loss": 0.6398,
+      "step": 5046
+    },
+    {
+      "epoch": 0.5048,
+      "learning_rate": 1.71788333462522e-05,
+      "loss": 0.8223,
+      "step": 5048
+    },
+    {
+      "epoch": 0.505,
+      "learning_rate": 1.718369173399646e-05,
+      "loss": 1.024,
+      "step": 5050
+    },
+    {
+      "epoch": 0.5052,
+      "learning_rate": 1.718854662049663e-05,
+      "loss": 0.0976,
+      "step": 5052
+    },
+    {
+      "epoch": 0.5054,
+      "learning_rate": 1.7193398003386507e-05,
+      "loss": 0.8196,
+      "step": 5054
+    },
+    {
+      "epoch": 0.5056,
+      "learning_rate": 1.7198245880301586e-05,
+      "loss": 0.4801,
+      "step": 5056
+    },
+    {
+      "epoch": 0.5058,
+      "learning_rate": 1.7203090248879063e-05,
+      "loss": 1.1504,
+      "step": 5058
+    },
+    {
+      "epoch": 0.506,
+      "learning_rate": 1.7207931106757864e-05,
+      "loss": 0.9992,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5062,
+      "learning_rate": 1.7212768451578602e-05,
+      "loss": 1.0247,
+      "step": 5062
+    },
+    {
+      "epoch": 0.5064,
+      "learning_rate": 1.721760228098363e-05,
+      "loss": 0.3998,
+      "step": 5064
+    },
+    {
+      "epoch": 0.5066,
+      "learning_rate": 1.7222432592616963e-05,
+      "loss": 1.0277,
+      "step": 5066
+    },
+    {
+      "epoch": 0.5068,
+      "learning_rate": 1.72272593841244e-05,
+      "loss": 0.6995,
+      "step": 5068
+    },
+    {
+      "epoch": 0.507,
+      "learning_rate": 1.7232082653153416e-05,
+      "loss": 0.7283,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5072,
+      "learning_rate": 1.723690239735321e-05,
+      "loss": 0.9888,
+      "step": 5072
+    },
+    {
+      "epoch": 0.5074,
+      "learning_rate": 1.724171861437467e-05,
+      "loss": 0.9474,
+      "step": 5074
+    },
+    {
+      "epoch": 0.5076,
+      "learning_rate": 1.724653130187046e-05,
+      "loss": 0.0943,
+      "step": 5076
+    },
+    {
+      "epoch": 0.5078,
+      "learning_rate": 1.7251340457494937e-05,
+      "loss": 1.0327,
+      "step": 5078
+    },
+    {
+      "epoch": 0.508,
+      "learning_rate": 1.7256146078904143e-05,
+      "loss": 0.7248,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5082,
+      "learning_rate": 1.726094816375591e-05,
+      "loss": 1.2109,
+      "step": 5082
+    },
+    {
+      "epoch": 0.5084,
+      "learning_rate": 1.7265746709709766e-05,
+      "loss": 0.5444,
+      "step": 5084
+    },
+    {
+      "epoch": 0.5086,
+      "learning_rate": 1.7270541714426923e-05,
+      "loss": 0.79,
+      "step": 5086
+    },
+    {
+      "epoch": 0.5088,
+      "learning_rate": 1.727533317557036e-05,
+      "loss": 0.3095,
+      "step": 5088
+    },
+    {
+      "epoch": 0.509,
+      "learning_rate": 1.7280121090804817e-05,
+      "loss": 0.8383,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5092,
+      "learning_rate": 1.7284905457796678e-05,
+      "loss": 0.5676,
+      "step": 5092
+    },
+    {
+      "epoch": 0.5094,
+      "learning_rate": 1.7289686274214106e-05,
+      "loss": 2.279,
+      "step": 5094
+    },
+    {
+      "epoch": 0.5096,
+      "learning_rate": 1.7294463537727012e-05,
+      "loss": 0.6753,
+      "step": 5096
+    },
+    {
+      "epoch": 0.5098,
+      "learning_rate": 1.7299237246007018e-05,
+      "loss": 0.5406,
+      "step": 5098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.730400739672745e-05,
+      "loss": 0.7745,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5102,
+      "learning_rate": 1.7308773987563393e-05,
+      "loss": 0.3873,
+      "step": 5102
+    },
+    {
+      "epoch": 0.5104,
+      "learning_rate": 1.7313537016191706e-05,
+      "loss": 1.1422,
+      "step": 5104
+    },
+    {
+      "epoch": 0.5106,
+      "learning_rate": 1.7318296480290912e-05,
+      "loss": 1.0113,
+      "step": 5106
+    },
+    {
+      "epoch": 0.5108,
+      "learning_rate": 1.732305237754132e-05,
+      "loss": 1.1497,
+      "step": 5108
+    },
+    {
+      "epoch": 0.511,
+      "learning_rate": 1.732780470562496e-05,
+      "loss": 0.7511,
+      "step": 5110
+    },
+    {
+      "epoch": 0.5112,
+      "learning_rate": 1.73325534622256e-05,
+      "loss": 0.6439,
+      "step": 5112
+    },
+    {
+      "epoch": 0.5114,
+      "learning_rate": 1.7337298645028764e-05,
+      "loss": 0.6478,
+      "step": 5114
+    },
+    {
+      "epoch": 0.5116,
+      "learning_rate": 1.7342040251721702e-05,
+      "loss": 0.9459,
+      "step": 5116
+    },
+    {
+      "epoch": 0.5118,
+      "learning_rate": 1.7346778279993413e-05,
+      "loss": 0.8949,
+      "step": 5118
+    },
+    {
+      "epoch": 0.512,
+      "learning_rate": 1.735151272753464e-05,
+      "loss": 0.5336,
+      "step": 5120
+    },
+    {
+      "epoch": 0.5122,
+      "learning_rate": 1.7356243592037872e-05,
+      "loss": 0.567,
+      "step": 5122
+    },
+    {
+      "epoch": 0.5124,
+      "learning_rate": 1.7360970871197344e-05,
+      "loss": 0.0395,
+      "step": 5124
+    },
+    {
+      "epoch": 0.5126,
+      "learning_rate": 1.736569456270903e-05,
+      "loss": 0.7874,
+      "step": 5126
+    },
+    {
+      "epoch": 0.5128,
+      "learning_rate": 1.7370414664270668e-05,
+      "loss": 0.851,
+      "step": 5128
+    },
+    {
+      "epoch": 0.513,
+      "learning_rate": 1.7375131173581737e-05,
+      "loss": 0.4565,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5132,
+      "learning_rate": 1.7379844088343466e-05,
+      "loss": 0.7925,
+      "step": 5132
+    },
+    {
+      "epoch": 0.5134,
+      "learning_rate": 1.7384553406258836e-05,
+      "loss": 1.0737,
+      "step": 5134
+    },
+    {
+      "epoch": 0.5136,
+      "learning_rate": 1.7389259125032584e-05,
+      "loss": 1.6573,
+      "step": 5136
+    },
+    {
+      "epoch": 0.5138,
+      "learning_rate": 1.73939612423712e-05,
+      "loss": 0.8234,
+      "step": 5138
+    },
+    {
+      "epoch": 0.514,
+      "learning_rate": 1.739865975598293e-05,
+      "loss": 0.8217,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5142,
+      "learning_rate": 1.740335466357778e-05,
+      "loss": 0.4941,
+      "step": 5142
+    },
+    {
+      "epoch": 0.5144,
+      "learning_rate": 1.7408045962867508e-05,
+      "loss": 0.5349,
+      "step": 5144
+    },
+    {
+      "epoch": 0.5146,
+      "learning_rate": 1.7412733651565607e-05,
+      "loss": 0.1346,
+      "step": 5146
+    },
+    {
+      "epoch": 0.5148,
+      "learning_rate": 1.7417417727387385e-05,
+      "loss": 0.3895,
+      "step": 5148
+    },
+    {
+      "epoch": 0.515,
+      "learning_rate": 1.7422098188049888e-05,
+      "loss": 0.6821,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5152,
+      "learning_rate": 1.742677503127187e-05,
+      "loss": 0.7471,
+      "step": 5152
+    },
+    {
+      "epoch": 0.5154,
+      "learning_rate": 1.7431448254773936e-05,
+      "loss": 0.6644,
+      "step": 5154
+    },
+    {
+      "epoch": 0.5156,
+      "learning_rate": 1.7436117856278404e-05,
+      "loss": 0.5308,
+      "step": 5156
+    },
+    {
+      "epoch": 0.5158,
+      "learning_rate": 1.7440783833509373e-05,
+      "loss": 1.4865,
+      "step": 5158
+    },
+    {
+      "epoch": 0.516,
+      "learning_rate": 1.744544618419267e-05,
+      "loss": 1.2349,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5162,
+      "learning_rate": 1.7450104906055956e-05,
+      "loss": 1.1049,
+      "step": 5162
+    },
+    {
+      "epoch": 0.5164,
+      "learning_rate": 1.745475999682863e-05,
+      "loss": 1.2529,
+      "step": 5164
+    },
+    {
+      "epoch": 0.5166,
+      "learning_rate": 1.7459411454241816e-05,
+      "loss": 0.6686,
+      "step": 5166
+    },
+    {
+      "epoch": 0.5168,
+      "learning_rate": 1.746405927602849e-05,
+      "loss": 0.1131,
+      "step": 5168
+    },
+    {
+      "epoch": 0.517,
+      "learning_rate": 1.746870345992336e-05,
+      "loss": 1.0399,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5172,
+      "learning_rate": 1.747334400366288e-05,
+      "loss": 0.4478,
+      "step": 5172
+    },
+    {
+      "epoch": 0.5174,
+      "learning_rate": 1.747798090498531e-05,
+      "loss": 0.9382,
+      "step": 5174
+    },
+    {
+      "epoch": 0.5176,
+      "learning_rate": 1.7482614161630714e-05,
+      "loss": 0.4797,
+      "step": 5176
+    },
+    {
+      "epoch": 0.5178,
+      "learning_rate": 1.7487243771340865e-05,
+      "loss": 1.395,
+      "step": 5178
+    },
+    {
+      "epoch": 0.518,
+      "learning_rate": 1.7491869731859356e-05,
+      "loss": 0.6039,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5182,
+      "learning_rate": 1.749649204093154e-05,
+      "loss": 0.4313,
+      "step": 5182
+    },
+    {
+      "epoch": 0.5184,
+      "learning_rate": 1.7501110696304598e-05,
+      "loss": 0.254,
+      "step": 5184
+    },
+    {
+      "epoch": 0.5186,
+      "learning_rate": 1.750572569572741e-05,
+      "loss": 0.5815,
+      "step": 5186
+    },
+    {
+      "epoch": 0.5188,
+      "learning_rate": 1.7510337036950693e-05,
+      "loss": 0.8845,
+      "step": 5188
+    },
+    {
+      "epoch": 0.519,
+      "learning_rate": 1.7514944717726962e-05,
+      "loss": 0.3638,
+      "step": 5190
+    },
+    {
+      "epoch": 0.5192,
+      "learning_rate": 1.7519548735810456e-05,
+      "loss": 0.6544,
+      "step": 5192
+    },
+    {
+      "epoch": 0.5194,
+      "learning_rate": 1.7524149088957244e-05,
+      "loss": 0.6046,
+      "step": 5194
+    },
+    {
+      "epoch": 0.5196,
+      "learning_rate": 1.752874577492517e-05,
+      "loss": 0.5483,
+      "step": 5196
+    },
+    {
+      "epoch": 0.5198,
+      "learning_rate": 1.753333879147387e-05,
+      "loss": 1.1658,
+      "step": 5198
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7537928136364756e-05,
+      "loss": 0.3291,
+      "step": 5200
+    },
+    {
+      "epoch": 0.5202,
+      "learning_rate": 1.7542513807361037e-05,
+      "loss": 1.0052,
+      "step": 5202
+    },
+    {
+      "epoch": 0.5204,
+      "learning_rate": 1.754709580222772e-05,
+      "loss": 2.1536,
+      "step": 5204
+    },
+    {
+      "epoch": 0.5206,
+      "learning_rate": 1.755167411873159e-05,
+      "loss": 0.819,
+      "step": 5206
+    },
+    {
+      "epoch": 0.5208,
+      "learning_rate": 1.7556248754641234e-05,
+      "loss": 0.1319,
+      "step": 5208
+    },
+    {
+      "epoch": 0.521,
+      "learning_rate": 1.7560819707727027e-05,
+      "loss": 1.3194,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5212,
+      "learning_rate": 1.7565386975761147e-05,
+      "loss": 0.7322,
+      "step": 5212
+    },
+    {
+      "epoch": 0.5214,
+      "learning_rate": 1.7569950556517563e-05,
+      "loss": 0.51,
+      "step": 5214
+    },
+    {
+      "epoch": 0.5216,
+      "learning_rate": 1.7574510447772034e-05,
+      "loss": 0.3213,
+      "step": 5216
+    },
+    {
+      "epoch": 0.5218,
+      "learning_rate": 1.757906664730213e-05,
+      "loss": 0.526,
+      "step": 5218
+    },
+    {
+      "epoch": 0.522,
+      "learning_rate": 1.758361915288722e-05,
+      "loss": 0.5978,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5222,
+      "learning_rate": 1.758816796230845e-05,
+      "loss": 0.9505,
+      "step": 5222
+    },
+    {
+      "epoch": 0.5224,
+      "learning_rate": 1.7592713073348805e-05,
+      "loss": 0.5402,
+      "step": 5224
+    },
+    {
+      "epoch": 0.5226,
+      "learning_rate": 1.759725448379304e-05,
+      "loss": 0.5921,
+      "step": 5226
+    },
+    {
+      "epoch": 0.5228,
+      "learning_rate": 1.7601792191427738e-05,
+      "loss": 0.4167,
+      "step": 5228
+    },
+    {
+      "epoch": 0.523,
+      "learning_rate": 1.7606326194041278e-05,
+      "loss": 0.6477,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5232,
+      "learning_rate": 1.7610856489423813e-05,
+      "loss": 0.1095,
+      "step": 5232
+    },
+    {
+      "epoch": 0.5234,
+      "learning_rate": 1.7615383075367363e-05,
+      "loss": 0.419,
+      "step": 5234
+    },
+    {
+      "epoch": 0.5236,
+      "learning_rate": 1.7619905949665725e-05,
+      "loss": 0.7019,
+      "step": 5236
+    },
+    {
+      "epoch": 0.5238,
+      "learning_rate": 1.762442511011447e-05,
+      "loss": 0.3126,
+      "step": 5238
+    },
+    {
+      "epoch": 0.524,
+      "learning_rate": 1.7628940554511057e-05,
+      "loss": 0.254,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5242,
+      "learning_rate": 1.763345228065469e-05,
+      "loss": 0.4397,
+      "step": 5242
+    },
+    {
+      "epoch": 0.5244,
+      "learning_rate": 1.7637960286346427e-05,
+      "loss": 1.3191,
+      "step": 5244
+    },
+    {
+      "epoch": 0.5246,
+      "learning_rate": 1.7642464569389083e-05,
+      "loss": 1.2893,
+      "step": 5246
+    },
+    {
+      "epoch": 0.5248,
+      "learning_rate": 1.7646965127587363e-05,
+      "loss": 0.5132,
+      "step": 5248
+    },
+    {
+      "epoch": 0.525,
+      "learning_rate": 1.7651461958747745e-05,
+      "loss": 0.6225,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5252,
+      "learning_rate": 1.765595506067851e-05,
+      "loss": 0.1316,
+      "step": 5252
+    },
+    {
+      "epoch": 0.5254,
+      "learning_rate": 1.766044443118977e-05,
+      "loss": 0.6758,
+      "step": 5254
+    },
+    {
+      "epoch": 0.5256,
+      "learning_rate": 1.7664930068093503e-05,
+      "loss": 0.713,
+      "step": 5256
+    },
+    {
+      "epoch": 0.5258,
+      "learning_rate": 1.766941196920342e-05,
+      "loss": 0.5146,
+      "step": 5258
+    },
+    {
+      "epoch": 0.526,
+      "learning_rate": 1.76738901323351e-05,
+      "loss": 0.9672,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5262,
+      "learning_rate": 1.767836455530598e-05,
+      "loss": 0.8052,
+      "step": 5262
+    },
+    {
+      "epoch": 0.5264,
+      "learning_rate": 1.7682835235935236e-05,
+      "loss": 0.9739,
+      "step": 5264
+    },
+    {
+      "epoch": 0.5266,
+      "learning_rate": 1.7687302172043933e-05,
+      "loss": 0.5922,
+      "step": 5266
+    },
+    {
+      "epoch": 0.5268,
+      "learning_rate": 1.769176536145494e-05,
+      "loss": 0.5581,
+      "step": 5268
+    },
+    {
+      "epoch": 0.527,
+      "learning_rate": 1.7696224801992947e-05,
+      "loss": 0.2483,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5272,
+      "learning_rate": 1.770068049148448e-05,
+      "loss": 0.604,
+      "step": 5272
+    },
+    {
+      "epoch": 0.5274,
+      "learning_rate": 1.7705132427757885e-05,
+      "loss": 0.5052,
+      "step": 5274
+    },
+    {
+      "epoch": 0.5276,
+      "learning_rate": 1.770958060864336e-05,
+      "loss": 1.0087,
+      "step": 5276
+    },
+    {
+      "epoch": 0.5278,
+      "learning_rate": 1.77140250319729e-05,
+      "loss": 0.6325,
+      "step": 5278
+    },
+    {
+      "epoch": 0.528,
+      "learning_rate": 1.771846569558035e-05,
+      "loss": 0.4536,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5282,
+      "learning_rate": 1.7722902597301385e-05,
+      "loss": 0.1595,
+      "step": 5282
+    },
+    {
+      "epoch": 0.5284,
+      "learning_rate": 1.772733573497351e-05,
+      "loss": 0.6238,
+      "step": 5284
+    },
+    {
+      "epoch": 0.5286,
+      "learning_rate": 1.7731765106436073e-05,
+      "loss": 1.1158,
+      "step": 5286
+    },
+    {
+      "epoch": 0.5288,
+      "learning_rate": 1.7736190709530247e-05,
+      "loss": 0.6865,
+      "step": 5288
+    },
+    {
+      "epoch": 0.529,
+      "learning_rate": 1.774061254209905e-05,
+      "loss": 0.7401,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5292,
+      "learning_rate": 1.7745030601987338e-05,
+      "loss": 0.4349,
+      "step": 5292
+    },
+    {
+      "epoch": 0.5294,
+      "learning_rate": 1.7749444887041793e-05,
+      "loss": 0.4787,
+      "step": 5294
+    },
+    {
+      "epoch": 0.5296,
+      "learning_rate": 1.7753855395110956e-05,
+      "loss": 0.8561,
+      "step": 5296
+    },
+    {
+      "epoch": 0.5298,
+      "learning_rate": 1.7758262124045192e-05,
+      "loss": 0.4178,
+      "step": 5298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7762665071696717e-05,
+      "loss": 0.7065,
+      "step": 5300
+    },
+    {
+      "epoch": 0.5302,
+      "learning_rate": 1.776706423591959e-05,
+      "loss": 0.6355,
+      "step": 5302
+    },
+    {
+      "epoch": 0.5304,
+      "learning_rate": 1.7771459614569707e-05,
+      "loss": 0.5603,
+      "step": 5304
+    },
+    {
+      "epoch": 0.5306,
+      "learning_rate": 1.7775851205504816e-05,
+      "loss": 0.7178,
+      "step": 5306
+    },
+    {
+      "epoch": 0.5308,
+      "learning_rate": 1.7780239006584512e-05,
+      "loss": 0.5637,
+      "step": 5308
+    },
+    {
+      "epoch": 0.531,
+      "learning_rate": 1.778462301567023e-05,
+      "loss": 0.6351,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5312,
+      "learning_rate": 1.778900323062526e-05,
+      "loss": 0.1656,
+      "step": 5312
+    },
+    {
+      "epoch": 0.5314,
+      "learning_rate": 1.7793379649314736e-05,
+      "loss": 0.9517,
+      "step": 5314
+    },
+    {
+      "epoch": 0.5316,
+      "learning_rate": 1.779775226960566e-05,
+      "loss": 0.6343,
+      "step": 5316
+    },
+    {
+      "epoch": 0.5318,
+      "learning_rate": 1.7802121089366832e-05,
+      "loss": 1.0649,
+      "step": 5318
+    },
+    {
+      "epoch": 0.532,
+      "learning_rate": 1.7806486106468973e-05,
+      "loss": 0.5041,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5322,
+      "learning_rate": 1.7810847318784635e-05,
+      "loss": 0.2015,
+      "step": 5322
+    },
+    {
+      "epoch": 0.5324,
+      "learning_rate": 1.7815204724188182e-05,
+      "loss": 0.9046,
+      "step": 5324
+    },
+    {
+      "epoch": 0.5326,
+      "learning_rate": 1.7819558320555895e-05,
+      "loss": 0.7726,
+      "step": 5326
+    },
+    {
+      "epoch": 0.5328,
+      "learning_rate": 1.7823908105765886e-05,
+      "loss": 0.6799,
+      "step": 5328
+    },
+    {
+      "epoch": 0.533,
+      "learning_rate": 1.7828254077698103e-05,
+      "loss": 0.8014,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5332,
+      "learning_rate": 1.7832596234234366e-05,
+      "loss": 0.6224,
+      "step": 5332
+    },
+    {
+      "epoch": 0.5334,
+      "learning_rate": 1.7836934573258392e-05,
+      "loss": 0.5607,
+      "step": 5334
+    },
+    {
+      "epoch": 0.5336,
+      "learning_rate": 1.784126909265572e-05,
+      "loss": 0.9873,
+      "step": 5336
+    },
+    {
+      "epoch": 0.5338,
+      "learning_rate": 1.7845599790313735e-05,
+      "loss": 0.984,
+      "step": 5338
+    },
+    {
+      "epoch": 0.534,
+      "learning_rate": 1.7849926664121716e-05,
+      "loss": 0.6584,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5342,
+      "learning_rate": 1.785424971197082e-05,
+      "loss": 0.4082,
+      "step": 5342
+    },
+    {
+      "epoch": 0.5344,
+      "learning_rate": 1.785856893175402e-05,
+      "loss": 0.7575,
+      "step": 5344
+    },
+    {
+      "epoch": 0.5346,
+      "learning_rate": 1.786288432136618e-05,
+      "loss": 0.8452,
+      "step": 5346
+    },
+    {
+      "epoch": 0.5348,
+      "learning_rate": 1.7867195878704062e-05,
+      "loss": 0.5444,
+      "step": 5348
+    },
+    {
+      "epoch": 0.535,
+      "learning_rate": 1.7871503601666233e-05,
+      "loss": 0.8221,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5352,
+      "learning_rate": 1.7875807488153173e-05,
+      "loss": 0.5556,
+      "step": 5352
+    },
+    {
+      "epoch": 0.5354,
+      "learning_rate": 1.788010753606722e-05,
+      "loss": 0.2679,
+      "step": 5354
+    },
+    {
+      "epoch": 0.5356,
+      "learning_rate": 1.7884403743312583e-05,
+      "loss": 0.6054,
+      "step": 5356
+    },
+    {
+      "epoch": 0.5358,
+      "learning_rate": 1.7888696107795343e-05,
+      "loss": 0.0819,
+      "step": 5358
+    },
+    {
+      "epoch": 0.536,
+      "learning_rate": 1.789298462742344e-05,
+      "loss": 0.9014,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5362,
+      "learning_rate": 1.7897269300106735e-05,
+      "loss": 1.5304,
+      "step": 5362
+    },
+    {
+      "epoch": 0.5364,
+      "learning_rate": 1.7901550123756902e-05,
+      "loss": 0.5208,
+      "step": 5364
+    },
+    {
+      "epoch": 0.5366,
+      "learning_rate": 1.790582709628753e-05,
+      "loss": 0.5905,
+      "step": 5366
+    },
+    {
+      "epoch": 0.5368,
+      "learning_rate": 1.791010021561407e-05,
+      "loss": 0.0692,
+      "step": 5368
+    },
+    {
+      "epoch": 0.537,
+      "learning_rate": 1.7914369479653854e-05,
+      "loss": 0.6554,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5372,
+      "learning_rate": 1.7918634886326102e-05,
+      "loss": 0.102,
+      "step": 5372
+    },
+    {
+      "epoch": 0.5374,
+      "learning_rate": 1.7922896433551903e-05,
+      "loss": 0.4042,
+      "step": 5374
+    },
+    {
+      "epoch": 0.5376,
+      "learning_rate": 1.7927154119254234e-05,
+      "loss": 1.2878,
+      "step": 5376
+    },
+    {
+      "epoch": 0.5378,
+      "learning_rate": 1.7931407941357945e-05,
+      "loss": 0.584,
+      "step": 5378
+    },
+    {
+      "epoch": 0.538,
+      "learning_rate": 1.7935657897789774e-05,
+      "loss": 1.2187,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5382,
+      "learning_rate": 1.793990398647835e-05,
+      "loss": 1.0817,
+      "step": 5382
+    },
+    {
+      "epoch": 0.5384,
+      "learning_rate": 1.794414620535418e-05,
+      "loss": 0.326,
+      "step": 5384
+    },
+    {
+      "epoch": 0.5386,
+      "learning_rate": 1.7948384552349655e-05,
+      "loss": 1.1729,
+      "step": 5386
+    },
+    {
+      "epoch": 0.5388,
+      "learning_rate": 1.7952619025399066e-05,
+      "loss": 0.662,
+      "step": 5388
+    },
+    {
+      "epoch": 0.539,
+      "learning_rate": 1.795684962243855e-05,
+      "loss": 0.2034,
+      "step": 5390
+    },
+    {
+      "epoch": 0.5392,
+      "learning_rate": 1.7961076341406202e-05,
+      "loss": 0.6619,
+      "step": 5392
+    },
+    {
+      "epoch": 0.5394,
+      "learning_rate": 1.796529918024196e-05,
+      "loss": 0.6765,
+      "step": 5394
+    },
+    {
+      "epoch": 0.5396,
+      "learning_rate": 1.7969518136887657e-05,
+      "loss": 0.6777,
+      "step": 5396
+    },
+    {
+      "epoch": 0.5398,
+      "learning_rate": 1.7973733209287032e-05,
+      "loss": 0.5912,
+      "step": 5398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7977944395385706e-05,
+      "loss": 0.5988,
+      "step": 5400
+    },
+    {
+      "epoch": 0.5402,
+      "learning_rate": 1.798215169313121e-05,
+      "loss": 0.7102,
+      "step": 5402
+    },
+    {
+      "epoch": 0.5404,
+      "learning_rate": 1.7986355100472922e-05,
+      "loss": 0.8057,
+      "step": 5404
+    },
+    {
+      "epoch": 0.5406,
+      "learning_rate": 1.7990554615362193e-05,
+      "loss": 0.5254,
+      "step": 5406
+    },
+    {
+      "epoch": 0.5408,
+      "learning_rate": 1.799475023575222e-05,
+      "loss": 0.8623,
+      "step": 5408
+    },
+    {
+      "epoch": 0.541,
+      "learning_rate": 1.79989419595981e-05,
+      "loss": 1.1223,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5412,
+      "learning_rate": 1.8003129784856826e-05,
+      "loss": 0.9058,
+      "step": 5412
+    },
+    {
+      "epoch": 0.5414,
+      "learning_rate": 1.800731370948734e-05,
+      "loss": 0.2894,
+      "step": 5414
+    },
+    {
+      "epoch": 0.5416,
+      "learning_rate": 1.8011493731450412e-05,
+      "loss": 1.1587,
+      "step": 5416
+    },
+    {
+      "epoch": 0.5418,
+      "learning_rate": 1.8015669848708757e-05,
+      "loss": 1.3021,
+      "step": 5418
+    },
+    {
+      "epoch": 0.542,
+      "learning_rate": 1.8019842059227005e-05,
+      "loss": 1.0548,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5422,
+      "learning_rate": 1.802401036097167e-05,
+      "loss": 0.7707,
+      "step": 5422
+    },
+    {
+      "epoch": 0.5424,
+      "learning_rate": 1.8028174751911147e-05,
+      "loss": 0.5724,
+      "step": 5424
+    },
+    {
+      "epoch": 0.5426,
+      "learning_rate": 1.803233523001577e-05,
+      "loss": 1.243,
+      "step": 5426
+    },
+    {
+      "epoch": 0.5428,
+      "learning_rate": 1.80364917932578e-05,
+      "loss": 0.9044,
+      "step": 5428
+    },
+    {
+      "epoch": 0.543,
+      "learning_rate": 1.804064443961135e-05,
+      "loss": 0.7482,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5432,
+      "learning_rate": 1.8044793167052466e-05,
+      "loss": 1.2576,
+      "step": 5432
+    },
+    {
+      "epoch": 0.5434,
+      "learning_rate": 1.804893797355914e-05,
+      "loss": 1.1196,
+      "step": 5434
+    },
+    {
+      "epoch": 0.5436,
+      "learning_rate": 1.8053078857111218e-05,
+      "loss": 0.5714,
+      "step": 5436
+    },
+    {
+      "epoch": 0.5438,
+      "learning_rate": 1.8057215815690494e-05,
+      "loss": 1.1987,
+      "step": 5438
+    },
+    {
+      "epoch": 0.544,
+      "learning_rate": 1.8061348847280658e-05,
+      "loss": 0.8601,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5442,
+      "learning_rate": 1.8065477949867327e-05,
+      "loss": 0.99,
+      "step": 5442
+    },
+    {
+      "epoch": 0.5444,
+      "learning_rate": 1.8069603121438018e-05,
+      "loss": 0.4188,
+      "step": 5444
+    },
+    {
+      "epoch": 0.5446,
+      "learning_rate": 1.8073724359982184e-05,
+      "loss": 0.8651,
+      "step": 5446
+    },
+    {
+      "epoch": 0.5448,
+      "learning_rate": 1.8077841663491174e-05,
+      "loss": 0.5128,
+      "step": 5448
+    },
+    {
+      "epoch": 0.545,
+      "learning_rate": 1.808195502995827e-05,
+      "loss": 0.4111,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5452,
+      "learning_rate": 1.8086064457378663e-05,
+      "loss": 0.8093,
+      "step": 5452
+    },
+    {
+      "epoch": 0.5454,
+      "learning_rate": 1.809016994374947e-05,
+      "loss": 1.0997,
+      "step": 5454
+    },
+    {
+      "epoch": 0.5456,
+      "learning_rate": 1.809427148706973e-05,
+      "loss": 0.4011,
+      "step": 5456
+    },
+    {
+      "epoch": 0.5458,
+      "learning_rate": 1.8098369085340397e-05,
+      "loss": 0.5343,
+      "step": 5458
+    },
+    {
+      "epoch": 0.546,
+      "learning_rate": 1.810246273656435e-05,
+      "loss": 0.5791,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5462,
+      "learning_rate": 1.81065524387464e-05,
+      "loss": 0.6573,
+      "step": 5462
+    },
+    {
+      "epoch": 0.5464,
+      "learning_rate": 1.8110638189893263e-05,
+      "loss": 1.029,
+      "step": 5464
+    },
+    {
+      "epoch": 0.5466,
+      "learning_rate": 1.8114719988013606e-05,
+      "loss": 0.5632,
+      "step": 5466
+    },
+    {
+      "epoch": 0.5468,
+      "learning_rate": 1.8118797831118003e-05,
+      "loss": 0.5125,
+      "step": 5468
+    },
+    {
+      "epoch": 0.547,
+      "learning_rate": 1.8122871717218968e-05,
+      "loss": 0.7954,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5472,
+      "learning_rate": 1.8126941644330936e-05,
+      "loss": 0.4595,
+      "step": 5472
+    },
+    {
+      "epoch": 0.5474,
+      "learning_rate": 1.813100761047028e-05,
+      "loss": 0.2687,
+      "step": 5474
+    },
+    {
+      "epoch": 0.5476,
+      "learning_rate": 1.8135069613655276e-05,
+      "loss": 1.1733,
+      "step": 5476
+    },
+    {
+      "epoch": 0.5478,
+      "learning_rate": 1.8139127651906176e-05,
+      "loss": 1.0713,
+      "step": 5478
+    },
+    {
+      "epoch": 0.548,
+      "learning_rate": 1.8143181723245137e-05,
+      "loss": 0.2184,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5482,
+      "learning_rate": 1.8147231825696258e-05,
+      "loss": 0.5605,
+      "step": 5482
+    },
+    {
+      "epoch": 0.5484,
+      "learning_rate": 1.8151277957285538e-05,
+      "loss": 0.0359,
+      "step": 5484
+    },
+    {
+      "epoch": 0.5486,
+      "learning_rate": 1.8155320116040976e-05,
+      "loss": 0.9937,
+      "step": 5486
+    },
+    {
+      "epoch": 0.5488,
+      "learning_rate": 1.8159358299992472e-05,
+      "loss": 1.3703,
+      "step": 5488
+    },
+    {
+      "epoch": 0.549,
+      "learning_rate": 1.8163392507171834e-05,
+      "loss": 0.1649,
+      "step": 5490
+    },
+    {
+      "epoch": 0.5492,
+      "learning_rate": 1.8167422735612866e-05,
+      "loss": 0.4576,
+      "step": 5492
+    },
+    {
+      "epoch": 0.5494,
+      "learning_rate": 1.817144898335129e-05,
+      "loss": 1.8121,
+      "step": 5494
+    },
+    {
+      "epoch": 0.5496,
+      "learning_rate": 1.8175471248424734e-05,
+      "loss": 0.1319,
+      "step": 5496
+    },
+    {
+      "epoch": 0.5498,
+      "learning_rate": 1.8179489528872797e-05,
+      "loss": 0.1953,
+      "step": 5498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8183503822737055e-05,
+      "loss": 0.039,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5502,
+      "learning_rate": 1.818751412806095e-05,
+      "loss": 0.3969,
+      "step": 5502
+    },
+    {
+      "epoch": 0.5504,
+      "learning_rate": 1.819152044288991e-05,
+      "loss": 0.9807,
+      "step": 5504
+    },
+    {
+      "epoch": 0.5506,
+      "learning_rate": 1.819552276527134e-05,
+      "loss": 0.7237,
+      "step": 5506
+    },
+    {
+      "epoch": 0.5508,
+      "learning_rate": 1.8199521093254528e-05,
+      "loss": 0.1236,
+      "step": 5508
+    },
+    {
+      "epoch": 0.551,
+      "learning_rate": 1.8203515424890738e-05,
+      "loss": 0.1113,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5512,
+      "learning_rate": 1.820750575823318e-05,
+      "loss": 1.0425,
+      "step": 5512
+    },
+    {
+      "epoch": 0.5514,
+      "learning_rate": 1.821149209133704e-05,
+      "loss": 0.4554,
+      "step": 5514
+    },
+    {
+      "epoch": 0.5516,
+      "learning_rate": 1.8215474422259403e-05,
+      "loss": 0.4265,
+      "step": 5516
+    },
+    {
+      "epoch": 0.5518,
+      "learning_rate": 1.8219452749059322e-05,
+      "loss": 0.9625,
+      "step": 5518
+    },
+    {
+      "epoch": 0.552,
+      "learning_rate": 1.8223427069797842e-05,
+      "loss": 0.8216,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5522,
+      "learning_rate": 1.82273973825379e-05,
+      "loss": 0.6477,
+      "step": 5522
+    },
+    {
+      "epoch": 0.5524,
+      "learning_rate": 1.823136368534442e-05,
+      "loss": 0.4819,
+      "step": 5524
+    },
+    {
+      "epoch": 0.5526,
+      "learning_rate": 1.8235325976284276e-05,
+      "loss": 0.7189,
+      "step": 5526
+    },
+    {
+      "epoch": 0.5528,
+      "learning_rate": 1.8239284253426294e-05,
+      "loss": 0.5836,
+      "step": 5528
+    },
+    {
+      "epoch": 0.553,
+      "learning_rate": 1.8243238514841258e-05,
+      "loss": 0.1269,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5532,
+      "learning_rate": 1.824718875860191e-05,
+      "loss": 0.4911,
+      "step": 5532
+    },
+    {
+      "epoch": 0.5534,
+      "learning_rate": 1.8251134982782952e-05,
+      "loss": 0.5909,
+      "step": 5534
+    },
+    {
+      "epoch": 0.5536,
+      "learning_rate": 1.8255077185461036e-05,
+      "loss": 0.3688,
+      "step": 5536
+    },
+    {
+      "epoch": 0.5538,
+      "learning_rate": 1.8259015364714786e-05,
+      "loss": 0.3632,
+      "step": 5538
+    },
+    {
+      "epoch": 0.554,
+      "learning_rate": 1.8262949518624777e-05,
+      "loss": 0.5657,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5542,
+      "learning_rate": 1.826687964527355e-05,
+      "loss": 0.9956,
+      "step": 5542
+    },
+    {
+      "epoch": 0.5544,
+      "learning_rate": 1.8270805742745615e-05,
+      "loss": 0.6415,
+      "step": 5544
+    },
+    {
+      "epoch": 0.5546,
+      "learning_rate": 1.8274727809127437e-05,
+      "loss": 0.169,
+      "step": 5546
+    },
+    {
+      "epoch": 0.5548,
+      "learning_rate": 1.8278645842507445e-05,
+      "loss": 0.3446,
+      "step": 5548
+    },
+    {
+      "epoch": 0.555,
+      "learning_rate": 1.828255984097604e-05,
+      "loss": 0.9067,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5552,
+      "learning_rate": 1.8286469802625585e-05,
+      "loss": 1.2265,
+      "step": 5552
+    },
+    {
+      "epoch": 0.5554,
+      "learning_rate": 1.8290375725550413e-05,
+      "loss": 0.611,
+      "step": 5554
+    },
+    {
+      "epoch": 0.5556,
+      "learning_rate": 1.8294277607846827e-05,
+      "loss": 0.6488,
+      "step": 5556
+    },
+    {
+      "epoch": 0.5558,
+      "learning_rate": 1.8298175447613093e-05,
+      "loss": 0.2005,
+      "step": 5558
+    },
+    {
+      "epoch": 0.556,
+      "learning_rate": 1.8302069242949463e-05,
+      "loss": 0.7298,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5562,
+      "learning_rate": 1.8305958991958125e-05,
+      "loss": 0.9016,
+      "step": 5562
+    },
+    {
+      "epoch": 0.5564,
+      "learning_rate": 1.830984469274328e-05,
+      "loss": 0.8848,
+      "step": 5564
+    },
+    {
+      "epoch": 0.5566,
+      "learning_rate": 1.8313726343411092e-05,
+      "loss": 0.9181,
+      "step": 5566
+    },
+    {
+      "epoch": 0.5568,
+      "learning_rate": 1.831760394206967e-05,
+      "loss": 1.2058,
+      "step": 5568
+    },
+    {
+      "epoch": 0.557,
+      "learning_rate": 1.832147748682912e-05,
+      "loss": 1.0491,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5572,
+      "learning_rate": 1.8325346975801543e-05,
+      "loss": 0.787,
+      "step": 5572
+    },
+    {
+      "epoch": 0.5574,
+      "learning_rate": 1.8329212407101e-05,
+      "loss": 0.4276,
+      "step": 5574
+    },
+    {
+      "epoch": 0.5576,
+      "learning_rate": 1.8333073778843493e-05,
+      "loss": 0.2842,
+      "step": 5576
+    },
+    {
+      "epoch": 0.5578,
+      "learning_rate": 1.8336931089147065e-05,
+      "loss": 0.0317,
+      "step": 5578
+    },
+    {
+      "epoch": 0.558,
+      "learning_rate": 1.8340784336131715e-05,
+      "loss": 0.6204,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5582,
+      "learning_rate": 1.8344633517919394e-05,
+      "loss": 1.1998,
+      "step": 5582
+    },
+    {
+      "epoch": 0.5584,
+      "learning_rate": 1.834847863263406e-05,
+      "loss": 0.5242,
+      "step": 5584
+    },
+    {
+      "epoch": 0.5586,
+      "learning_rate": 1.8352319678401677e-05,
+      "loss": 0.1298,
+      "step": 5586
+    },
+    {
+      "epoch": 0.5588,
+      "learning_rate": 1.8356156653350138e-05,
+      "loss": 0.6589,
+      "step": 5588
+    },
+    {
+      "epoch": 0.559,
+      "learning_rate": 1.8359989555609344e-05,
+      "loss": 0.2831,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5592,
+      "learning_rate": 1.8363818383311226e-05,
+      "loss": 0.9864,
+      "step": 5592
+    },
+    {
+      "epoch": 0.5594,
+      "learning_rate": 1.836764313458962e-05,
+      "loss": 0.46,
+      "step": 5594
+    },
+    {
+      "epoch": 0.5596,
+      "learning_rate": 1.83714638075804e-05,
+      "loss": 0.9493,
+      "step": 5596
+    },
+    {
+      "epoch": 0.5598,
+      "learning_rate": 1.8375280400421407e-05,
+      "loss": 0.5981,
+      "step": 5598
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8379092911252515e-05,
+      "loss": 0.7234,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5602,
+      "learning_rate": 1.8382901338215515e-05,
+      "loss": 0.8477,
+      "step": 5602
+    },
+    {
+      "epoch": 0.5604,
+      "learning_rate": 1.8386705679454243e-05,
+      "loss": 1.0927,
+      "step": 5604
+    },
+    {
+      "epoch": 0.5606,
+      "learning_rate": 1.8390505933114503e-05,
+      "loss": 0.4993,
+      "step": 5606
+    },
+    {
+      "epoch": 0.5608,
+      "learning_rate": 1.83943020973441e-05,
+      "loss": 1.0562,
+      "step": 5608
+    },
+    {
+      "epoch": 0.561,
+      "learning_rate": 1.839809417029283e-05,
+      "loss": 0.9893,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5612,
+      "learning_rate": 1.840188215011248e-05,
+      "loss": 0.5998,
+      "step": 5612
+    },
+    {
+      "epoch": 0.5614,
+      "learning_rate": 1.8405666034956842e-05,
+      "loss": 0.8744,
+      "step": 5614
+    },
+    {
+      "epoch": 0.5616,
+      "learning_rate": 1.8409445822981687e-05,
+      "loss": 0.2612,
+      "step": 5616
+    },
+    {
+      "epoch": 0.5618,
+      "learning_rate": 1.8413221512344805e-05,
+      "loss": 0.6146,
+      "step": 5618
+    },
+    {
+      "epoch": 0.562,
+      "learning_rate": 1.8416993101205957e-05,
+      "loss": 0.8302,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5622,
+      "learning_rate": 1.842076058772692e-05,
+      "loss": 0.7388,
+      "step": 5622
+    },
+    {
+      "epoch": 0.5624,
+      "learning_rate": 1.8424523970071475e-05,
+      "loss": 0.755,
+      "step": 5624
+    },
+    {
+      "epoch": 0.5626,
+      "learning_rate": 1.8428283246405386e-05,
+      "loss": 1.4252,
+      "step": 5626
+    },
+    {
+      "epoch": 0.5628,
+      "learning_rate": 1.843203841489643e-05,
+      "loss": 1.3402,
+      "step": 5628
+    },
+    {
+      "epoch": 0.563,
+      "learning_rate": 1.8435789473714384e-05,
+      "loss": 0.369,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5632,
+      "learning_rate": 1.843953642103103e-05,
+      "loss": 1.2148,
+      "step": 5632
+    },
+    {
+      "epoch": 0.5634,
+      "learning_rate": 1.844327925502015e-05,
+      "loss": 0.3141,
+      "step": 5634
+    },
+    {
+      "epoch": 0.5636,
+      "learning_rate": 1.8447017973857528e-05,
+      "loss": 0.7207,
+      "step": 5636
+    },
+    {
+      "epoch": 0.5638,
+      "learning_rate": 1.8450752575720964e-05,
+      "loss": 0.4128,
+      "step": 5638
+    },
+    {
+      "epoch": 0.564,
+      "learning_rate": 1.845448305879026e-05,
+      "loss": 0.5683,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5642,
+      "learning_rate": 1.8458209421247205e-05,
+      "loss": 0.223,
+      "step": 5642
+    },
+    {
+      "epoch": 0.5644,
+      "learning_rate": 1.8461931661275635e-05,
+      "loss": 0.4198,
+      "step": 5644
+    },
+    {
+      "epoch": 0.5646,
+      "learning_rate": 1.8465649777061384e-05,
+      "loss": 1.9991,
+      "step": 5646
+    },
+    {
+      "epoch": 0.5648,
+      "learning_rate": 1.846936376679225e-05,
+      "loss": 0.635,
+      "step": 5648
+    },
+    {
+      "epoch": 0.565,
+      "learning_rate": 1.8473073628658116e-05,
+      "loss": 0.8242,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5652,
+      "learning_rate": 1.847677936085084e-05,
+      "loss": 1.7428,
+      "step": 5652
+    },
+    {
+      "epoch": 0.5654,
+      "learning_rate": 1.8480480961564266e-05,
+      "loss": 0.5888,
+      "step": 5654
+    },
+    {
+      "epoch": 0.5656,
+      "learning_rate": 1.848417842899428e-05,
+      "loss": 1.098,
+      "step": 5656
+    },
+    {
+      "epoch": 0.5658,
+      "learning_rate": 1.848787176133881e-05,
+      "loss": 0.8092,
+      "step": 5658
+    },
+    {
+      "epoch": 0.566,
+      "learning_rate": 1.8491560956797766e-05,
+      "loss": 0.1081,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5662,
+      "learning_rate": 1.8495246013573047e-05,
+      "loss": 0.594,
+      "step": 5662
+    },
+    {
+      "epoch": 0.5664,
+      "learning_rate": 1.8498926929868634e-05,
+      "loss": 0.5405,
+      "step": 5664
+    },
+    {
+      "epoch": 0.5666,
+      "learning_rate": 1.850260370389049e-05,
+      "loss": 1.4108,
+      "step": 5666
+    },
+    {
+      "epoch": 0.5668,
+      "learning_rate": 1.850627633384658e-05,
+      "loss": 0.5565,
+      "step": 5668
+    },
+    {
+      "epoch": 0.567,
+      "learning_rate": 1.850994481794691e-05,
+      "loss": 0.9671,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5672,
+      "learning_rate": 1.851360915440354e-05,
+      "loss": 1.2306,
+      "step": 5672
+    },
+    {
+      "epoch": 0.5674,
+      "learning_rate": 1.851726934143048e-05,
+      "loss": 1.079,
+      "step": 5674
+    },
+    {
+      "epoch": 0.5676,
+      "learning_rate": 1.8520925377243812e-05,
+      "loss": 0.915,
+      "step": 5676
+    },
+    {
+      "epoch": 0.5678,
+      "learning_rate": 1.8524577260061628e-05,
+      "loss": 0.6494,
+      "step": 5678
+    },
+    {
+      "epoch": 0.568,
+      "learning_rate": 1.8528224988104044e-05,
+      "loss": 0.7025,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5682,
+      "learning_rate": 1.8531868559593205e-05,
+      "loss": 0.3215,
+      "step": 5682
+    },
+    {
+      "epoch": 0.5684,
+      "learning_rate": 1.8535507972753265e-05,
+      "loss": 1.4606,
+      "step": 5684
+    },
+    {
+      "epoch": 0.5686,
+      "learning_rate": 1.8539143225810453e-05,
+      "loss": 0.6566,
+      "step": 5686
+    },
+    {
+      "epoch": 0.5688,
+      "learning_rate": 1.8542774316992953e-05,
+      "loss": 0.8378,
+      "step": 5688
+    },
+    {
+      "epoch": 0.569,
+      "learning_rate": 1.8546401244531028e-05,
+      "loss": 0.3057,
+      "step": 5690
+    },
+    {
+      "epoch": 0.5692,
+      "learning_rate": 1.8550024006656967e-05,
+      "loss": 1.1858,
+      "step": 5692
+    },
+    {
+      "epoch": 0.5694,
+      "learning_rate": 1.8553642601605066e-05,
+      "loss": 0.8748,
+      "step": 5694
+    },
+    {
+      "epoch": 0.5696,
+      "learning_rate": 1.8557257027611673e-05,
+      "loss": 1.1315,
+      "step": 5696
+    },
+    {
+      "epoch": 0.5698,
+      "learning_rate": 1.856086728291516e-05,
+      "loss": 0.837,
+      "step": 5698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8564473365755933e-05,
+      "loss": 0.5947,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5702,
+      "learning_rate": 1.856807527437643e-05,
+      "loss": 0.4279,
+      "step": 5702
+    },
+    {
+      "epoch": 0.5704,
+      "learning_rate": 1.8571673007021124e-05,
+      "loss": 0.7164,
+      "step": 5704
+    },
+    {
+      "epoch": 0.5706,
+      "learning_rate": 1.857526656193652e-05,
+      "loss": 0.4415,
+      "step": 5706
+    },
+    {
+      "epoch": 0.5708,
+      "learning_rate": 1.8578855937371172e-05,
+      "loss": 1.0528,
+      "step": 5708
+    },
+    {
+      "epoch": 0.571,
+      "learning_rate": 1.8582441131575658e-05,
+      "loss": 0.9234,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5712,
+      "learning_rate": 1.85860221428026e-05,
+      "loss": 0.703,
+      "step": 5712
+    },
+    {
+      "epoch": 0.5714,
+      "learning_rate": 1.8589598969306643e-05,
+      "loss": 0.4291,
+      "step": 5714
+    },
+    {
+      "epoch": 0.5716,
+      "learning_rate": 1.8593171609344502e-05,
+      "loss": 1.3525,
+      "step": 5716
+    },
+    {
+      "epoch": 0.5718,
+      "learning_rate": 1.859674006117491e-05,
+      "loss": 1.0548,
+      "step": 5718
+    },
+    {
+      "epoch": 0.572,
+      "learning_rate": 1.8600304323058644e-05,
+      "loss": 0.6996,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5722,
+      "learning_rate": 1.860386439325853e-05,
+      "loss": 0.907,
+      "step": 5722
+    },
+    {
+      "epoch": 0.5724,
+      "learning_rate": 1.8607420270039432e-05,
+      "loss": 0.5103,
+      "step": 5724
+    },
+    {
+      "epoch": 0.5726,
+      "learning_rate": 1.8610971951668268e-05,
+      "loss": 0.0866,
+      "step": 5726
+    },
+    {
+      "epoch": 0.5728,
+      "learning_rate": 1.8614519436413965e-05,
+      "loss": 1.2033,
+      "step": 5728
+    },
+    {
+      "epoch": 0.573,
+      "learning_rate": 1.8618062722547544e-05,
+      "loss": 0.1962,
+      "step": 5730
+    },
+    {
+      "epoch": 0.5732,
+      "learning_rate": 1.862160180834206e-05,
+      "loss": 0.201,
+      "step": 5732
+    },
+    {
+      "epoch": 0.5734,
+      "learning_rate": 1.862513669207257e-05,
+      "loss": 1.0934,
+      "step": 5734
+    },
+    {
+      "epoch": 0.5736,
+      "learning_rate": 1.8628667372016243e-05,
+      "loss": 1.0845,
+      "step": 5736
+    },
+    {
+      "epoch": 0.5738,
+      "learning_rate": 1.8632193846452274e-05,
+      "loss": 1.2849,
+      "step": 5738
+    },
+    {
+      "epoch": 0.574,
+      "learning_rate": 1.863571611366188e-05,
+      "loss": 0.5532,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5742,
+      "learning_rate": 1.8639234171928348e-05,
+      "loss": 0.5229,
+      "step": 5742
+    },
+    {
+      "epoch": 0.5744,
+      "learning_rate": 1.864274801953704e-05,
+      "loss": 0.0623,
+      "step": 5744
+    },
+    {
+      "epoch": 0.5746,
+      "learning_rate": 1.8646257654775354e-05,
+      "loss": 0.3962,
+      "step": 5746
+    },
+    {
+      "epoch": 0.5748,
+      "learning_rate": 1.864976307593271e-05,
+      "loss": 1.2473,
+      "step": 5748
+    },
+    {
+      "epoch": 0.575,
+      "learning_rate": 1.8653264281300612e-05,
+      "loss": 0.612,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5752,
+      "learning_rate": 1.8656761269172645e-05,
+      "loss": 0.8552,
+      "step": 5752
+    },
+    {
+      "epoch": 0.5754,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.3856,
+      "step": 5754
+    },
+    {
+      "epoch": 0.5756,
+      "learning_rate": 1.866374258561351e-05,
+      "loss": 1.2221,
+      "step": 5756
+    },
+    {
+      "epoch": 0.5758,
+      "learning_rate": 1.8667226910779767e-05,
+      "loss": 0.4799,
+      "step": 5758
+    },
+    {
+      "epoch": 0.576,
+      "learning_rate": 1.86707070116449e-05,
+      "loss": 1.0471,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5762,
+      "learning_rate": 1.8674182886512776e-05,
+      "loss": 0.4943,
+      "step": 5762
+    },
+    {
+      "epoch": 0.5764,
+      "learning_rate": 1.8677654533689287e-05,
+      "loss": 0.5059,
+      "step": 5764
+    },
+    {
+      "epoch": 0.5766,
+      "learning_rate": 1.8681121951482393e-05,
+      "loss": 0.4225,
+      "step": 5766
+    },
+    {
+      "epoch": 0.5768,
+      "learning_rate": 1.868458513820212e-05,
+      "loss": 0.1684,
+      "step": 5768
+    },
+    {
+      "epoch": 0.577,
+      "learning_rate": 1.8688044092160554e-05,
+      "loss": 0.4428,
+      "step": 5770
+    },
+    {
+      "epoch": 0.5772,
+      "learning_rate": 1.869149881167184e-05,
+      "loss": 0.8819,
+      "step": 5772
+    },
+    {
+      "epoch": 0.5774,
+      "learning_rate": 1.869494929505219e-05,
+      "loss": 1.2407,
+      "step": 5774
+    },
+    {
+      "epoch": 0.5776,
+      "learning_rate": 1.8698395540619883e-05,
+      "loss": 0.8609,
+      "step": 5776
+    },
+    {
+      "epoch": 0.5778,
+      "learning_rate": 1.8701837546695256e-05,
+      "loss": 1.1924,
+      "step": 5778
+    },
+    {
+      "epoch": 0.578,
+      "learning_rate": 1.870527531160072e-05,
+      "loss": 0.9832,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5782,
+      "learning_rate": 1.870870883366075e-05,
+      "loss": 0.6343,
+      "step": 5782
+    },
+    {
+      "epoch": 0.5784,
+      "learning_rate": 1.8712138111201894e-05,
+      "loss": 1.3504,
+      "step": 5784
+    },
+    {
+      "epoch": 0.5786,
+      "learning_rate": 1.871556314255275e-05,
+      "loss": 0.9431,
+      "step": 5786
+    },
+    {
+      "epoch": 0.5788,
+      "learning_rate": 1.871898392604402e-05,
+      "loss": 0.7098,
+      "step": 5788
+    },
+    {
+      "epoch": 0.579,
+      "learning_rate": 1.8722400460008434e-05,
+      "loss": 0.7925,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5792,
+      "learning_rate": 1.872581274278083e-05,
+      "loss": 0.5714,
+      "step": 5792
+    },
+    {
+      "epoch": 0.5794,
+      "learning_rate": 1.8729220772698093e-05,
+      "loss": 0.0303,
+      "step": 5794
+    },
+    {
+      "epoch": 0.5796,
+      "learning_rate": 1.87326245480992e-05,
+      "loss": 0.9228,
+      "step": 5796
+    },
+    {
+      "epoch": 0.5798,
+      "learning_rate": 1.8736024067325195e-05,
+      "loss": 1.0539,
+      "step": 5798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8739419328719166e-05,
+      "loss": 0.2771,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5802,
+      "learning_rate": 1.8742810330626335e-05,
+      "loss": 0.5574,
+      "step": 5802
+    },
+    {
+      "epoch": 0.5804,
+      "learning_rate": 1.8746197071393963e-05,
+      "loss": 0.6233,
+      "step": 5804
+    },
+    {
+      "epoch": 0.5806,
+      "learning_rate": 1.8749579549371373e-05,
+      "loss": 1.0801,
+      "step": 5806
+    },
+    {
+      "epoch": 0.5808,
+      "learning_rate": 1.875295776291001e-05,
+      "loss": 0.9572,
+      "step": 5808
+    },
+    {
+      "epoch": 0.581,
+      "learning_rate": 1.8756331710363368e-05,
+      "loss": 0.8719,
+      "step": 5810
+    },
+    {
+      "epoch": 0.5812,
+      "learning_rate": 1.875970139008703e-05,
+      "loss": 0.2589,
+      "step": 5812
+    },
+    {
+      "epoch": 0.5814,
+      "learning_rate": 1.876306680043863e-05,
+      "loss": 0.2814,
+      "step": 5814
+    },
+    {
+      "epoch": 0.5816,
+      "learning_rate": 1.8766427939777936e-05,
+      "loss": 0.5015,
+      "step": 5816
+    },
+    {
+      "epoch": 0.5818,
+      "learning_rate": 1.876978480646677e-05,
+      "loss": 0.7675,
+      "step": 5818
+    },
+    {
+      "epoch": 0.582,
+      "learning_rate": 1.8773137398869007e-05,
+      "loss": 0.466,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5822,
+      "learning_rate": 1.8776485715350665e-05,
+      "loss": 1.1918,
+      "step": 5822
+    },
+    {
+      "epoch": 0.5824,
+      "learning_rate": 1.877982975427981e-05,
+      "loss": 0.7905,
+      "step": 5824
+    },
+    {
+      "epoch": 0.5826,
+      "learning_rate": 1.878316951402658e-05,
+      "loss": 0.8287,
+      "step": 5826
+    },
+    {
+      "epoch": 0.5828,
+      "learning_rate": 1.878650499296322e-05,
+      "loss": 0.7925,
+      "step": 5828
+    },
+    {
+      "epoch": 0.583,
+      "learning_rate": 1.878983618946409e-05,
+      "loss": 1.0461,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5832,
+      "learning_rate": 1.8793163101905565e-05,
+      "loss": 1.0898,
+      "step": 5832
+    },
+    {
+      "epoch": 0.5834,
+      "learning_rate": 1.879648572866617e-05,
+      "loss": 0.7093,
+      "step": 5834
+    },
+    {
+      "epoch": 0.5836,
+      "learning_rate": 1.879980406812648e-05,
+      "loss": 0.3617,
+      "step": 5836
+    },
+    {
+      "epoch": 0.5838,
+      "learning_rate": 1.8803118118669203e-05,
+      "loss": 0.7123,
+      "step": 5838
+    },
+    {
+      "epoch": 0.584,
+      "learning_rate": 1.880642787867909e-05,
+      "loss": 0.1266,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5842,
+      "learning_rate": 1.8809733346543006e-05,
+      "loss": 0.6751,
+      "step": 5842
+    },
+    {
+      "epoch": 0.5844,
+      "learning_rate": 1.8813034520649923e-05,
+      "loss": 1.0531,
+      "step": 5844
+    },
+    {
+      "epoch": 0.5846,
+      "learning_rate": 1.881633139939087e-05,
+      "loss": 0.5335,
+      "step": 5846
+    },
+    {
+      "epoch": 0.5848,
+      "learning_rate": 1.8819623981158996e-05,
+      "loss": 0.301,
+      "step": 5848
+    },
+    {
+      "epoch": 0.585,
+      "learning_rate": 1.8822912264349532e-05,
+      "loss": 1.2301,
+      "step": 5850
+    },
+    {
+      "epoch": 0.5852,
+      "learning_rate": 1.8826196247359816e-05,
+      "loss": 0.5119,
+      "step": 5852
+    },
+    {
+      "epoch": 0.5854,
+      "learning_rate": 1.882947592858927e-05,
+      "loss": 1.0105,
+      "step": 5854
+    },
+    {
+      "epoch": 0.5856,
+      "learning_rate": 1.883275130643942e-05,
+      "loss": 0.6306,
+      "step": 5856
+    },
+    {
+      "epoch": 0.5858,
+      "learning_rate": 1.8836022379313877e-05,
+      "loss": 1.0729,
+      "step": 5858
+    },
+    {
+      "epoch": 0.586,
+      "learning_rate": 1.8839289145618375e-05,
+      "loss": 1.2791,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5862,
+      "learning_rate": 1.884255160376072e-05,
+      "loss": 1.2814,
+      "step": 5862
+    },
+    {
+      "epoch": 0.5864,
+      "learning_rate": 1.884580975215084e-05,
+      "loss": 0.8639,
+      "step": 5864
+    },
+    {
+      "epoch": 0.5866,
+      "learning_rate": 1.8849063589200744e-05,
+      "loss": 1.1901,
+      "step": 5866
+    },
+    {
+      "epoch": 0.5868,
+      "learning_rate": 1.8852313113324553e-05,
+      "loss": 0.8714,
+      "step": 5868
+    },
+    {
+      "epoch": 0.587,
+      "learning_rate": 1.885555832293849e-05,
+      "loss": 0.8229,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5872,
+      "learning_rate": 1.8858799216460876e-05,
+      "loss": 0.5886,
+      "step": 5872
+    },
+    {
+      "epoch": 0.5874,
+      "learning_rate": 1.8862035792312145e-05,
+      "loss": 0.8358,
+      "step": 5874
+    },
+    {
+      "epoch": 0.5876,
+      "learning_rate": 1.8865268048914828e-05,
+      "loss": 0.4087,
+      "step": 5876
+    },
+    {
+      "epoch": 0.5878,
+      "learning_rate": 1.886849598469356e-05,
+      "loss": 0.4032,
+      "step": 5878
+    },
+    {
+      "epoch": 0.588,
+      "learning_rate": 1.8871719598075076e-05,
+      "loss": 0.7839,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5882,
+      "learning_rate": 1.8874938887488246e-05,
+      "loss": 1.1556,
+      "step": 5882
+    },
+    {
+      "epoch": 0.5884,
+      "learning_rate": 1.887815385136402e-05,
+      "loss": 0.4573,
+      "step": 5884
+    },
+    {
+      "epoch": 0.5886,
+      "learning_rate": 1.888136448813544e-05,
+      "loss": 0.753,
+      "step": 5886
+    },
+    {
+      "epoch": 0.5888,
+      "learning_rate": 1.8884570796237716e-05,
+      "loss": 0.7976,
+      "step": 5888
+    },
+    {
+      "epoch": 0.589,
+      "learning_rate": 1.888777277410812e-05,
+      "loss": 0.8274,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5892,
+      "learning_rate": 1.889097042018603e-05,
+      "loss": 0.7292,
+      "step": 5892
+    },
+    {
+      "epoch": 0.5894,
+      "learning_rate": 1.8894163732912972e-05,
+      "loss": 0.6234,
+      "step": 5894
+    },
+    {
+      "epoch": 0.5896,
+      "learning_rate": 1.889735271073256e-05,
+      "loss": 1.0246,
+      "step": 5896
+    },
+    {
+      "epoch": 0.5898,
+      "learning_rate": 1.890053735209053e-05,
+      "loss": 0.8762,
+      "step": 5898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.89037176554347e-05,
+      "loss": 0.5597,
+      "step": 5900
+    },
+    {
+      "epoch": 0.5902,
+      "learning_rate": 1.890689361921506e-05,
+      "loss": 0.5631,
+      "step": 5902
+    },
+    {
+      "epoch": 0.5904,
+      "learning_rate": 1.8910065241883683e-05,
+      "loss": 1.039,
+      "step": 5904
+    },
+    {
+      "epoch": 0.5906,
+      "learning_rate": 1.8913232521894737e-05,
+      "loss": 0.0514,
+      "step": 5906
+    },
+    {
+      "epoch": 0.5908,
+      "learning_rate": 1.891639545770453e-05,
+      "loss": 0.7188,
+      "step": 5908
+    },
+    {
+      "epoch": 0.591,
+      "learning_rate": 1.891955404777151e-05,
+      "loss": 0.404,
+      "step": 5910
+    },
+    {
+      "epoch": 0.5912,
+      "learning_rate": 1.8922708290556197e-05,
+      "loss": 0.4287,
+      "step": 5912
+    },
+    {
+      "epoch": 0.5914,
+      "learning_rate": 1.8925858184521248e-05,
+      "loss": 0.9644,
+      "step": 5914
+    },
+    {
+      "epoch": 0.5916,
+      "learning_rate": 1.8929003728131473e-05,
+      "loss": 0.2088,
+      "step": 5916
+    },
+    {
+      "epoch": 0.5918,
+      "learning_rate": 1.893214491985374e-05,
+      "loss": 0.9412,
+      "step": 5918
+    },
+    {
+      "epoch": 0.592,
+      "learning_rate": 1.893528175815708e-05,
+      "loss": 0.0864,
+      "step": 5920
+    },
+    {
+      "epoch": 0.5922,
+      "learning_rate": 1.8938414241512634e-05,
+      "loss": 0.4472,
+      "step": 5922
+    },
+    {
+      "epoch": 0.5924,
+      "learning_rate": 1.8941542368393683e-05,
+      "loss": 0.505,
+      "step": 5924
+    },
+    {
+      "epoch": 0.5926,
+      "learning_rate": 1.89446661372756e-05,
+      "loss": 1.075,
+      "step": 5926
+    },
+    {
+      "epoch": 0.5928,
+      "learning_rate": 1.8947785546635895e-05,
+      "loss": 0.0608,
+      "step": 5928
+    },
+    {
+      "epoch": 0.593,
+      "learning_rate": 1.8950900594954226e-05,
+      "loss": 0.149,
+      "step": 5930
+    },
+    {
+      "epoch": 0.5932,
+      "learning_rate": 1.8954011280712338e-05,
+      "loss": 0.5782,
+      "step": 5932
+    },
+    {
+      "epoch": 0.5934,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 0.8727,
+      "step": 5934
+    },
+    {
+      "epoch": 0.5936,
+      "learning_rate": 1.896021955848561e-05,
+      "loss": 0.1822,
+      "step": 5936
+    },
+    {
+      "epoch": 0.5938,
+      "learning_rate": 1.896331714747493e-05,
+      "loss": 0.2741,
+      "step": 5938
+    },
+    {
+      "epoch": 0.594,
+      "learning_rate": 1.8966410367852358e-05,
+      "loss": 0.5928,
+      "step": 5940
+    },
+    {
+      "epoch": 0.5942,
+      "learning_rate": 1.89694992181103e-05,
+      "loss": 0.919,
+      "step": 5942
+    },
+    {
+      "epoch": 0.5944,
+      "learning_rate": 1.8972583696743284e-05,
+      "loss": 0.7504,
+      "step": 5944
+    },
+    {
+      "epoch": 0.5946,
+      "learning_rate": 1.8975663802247975e-05,
+      "loss": 1.2331,
+      "step": 5946
+    },
+    {
+      "epoch": 0.5948,
+      "learning_rate": 1.8978739533123166e-05,
+      "loss": 0.2211,
+      "step": 5948
+    },
+    {
+      "epoch": 0.595,
+      "learning_rate": 1.8981810887869784e-05,
+      "loss": 0.4776,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5952,
+      "learning_rate": 1.8984877864990888e-05,
+      "loss": 0.8871,
+      "step": 5952
+    },
+    {
+      "epoch": 0.5954,
+      "learning_rate": 1.898794046299167e-05,
+      "loss": 0.6649,
+      "step": 5954
+    },
+    {
+      "epoch": 0.5956,
+      "learning_rate": 1.8990998680379455e-05,
+      "loss": 0.3046,
+      "step": 5956
+    },
+    {
+      "epoch": 0.5958,
+      "learning_rate": 1.8994052515663708e-05,
+      "loss": 0.6719,
+      "step": 5958
+    },
+    {
+      "epoch": 0.596,
+      "learning_rate": 1.8997101967356028e-05,
+      "loss": 0.5067,
+      "step": 5960
+    },
+    {
+      "epoch": 0.5962,
+      "learning_rate": 1.9000147033970144e-05,
+      "loss": 1.156,
+      "step": 5962
+    },
+    {
+      "epoch": 0.5964,
+      "learning_rate": 1.9003187714021933e-05,
+      "loss": 0.7789,
+      "step": 5964
+    },
+    {
+      "epoch": 0.5966,
+      "learning_rate": 1.90062240060294e-05,
+      "loss": 0.4279,
+      "step": 5966
+    },
+    {
+      "epoch": 0.5968,
+      "learning_rate": 1.9009255908512697e-05,
+      "loss": 1.1202,
+      "step": 5968
+    },
+    {
+      "epoch": 0.597,
+      "learning_rate": 1.901228341999412e-05,
+      "loss": 0.13,
+      "step": 5970
+    },
+    {
+      "epoch": 0.5972,
+      "learning_rate": 1.9015306538998066e-05,
+      "loss": 0.0766,
+      "step": 5972
+    },
+    {
+      "epoch": 0.5974,
+      "learning_rate": 1.9018325264051136e-05,
+      "loss": 0.5354,
+      "step": 5974
+    },
+    {
+      "epoch": 0.5976,
+      "learning_rate": 1.9021339593682032e-05,
+      "loss": 0.8664,
+      "step": 5976
+    },
+    {
+      "epoch": 0.5978,
+      "learning_rate": 1.9024349526421596e-05,
+      "loss": 0.732,
+      "step": 5978
+    },
+    {
+      "epoch": 0.598,
+      "learning_rate": 1.9027355060802827e-05,
+      "loss": 0.5084,
+      "step": 5980
+    },
+    {
+      "epoch": 0.5982,
+      "learning_rate": 1.9030356195360868e-05,
+      "loss": 0.3877,
+      "step": 5982
+    },
+    {
+      "epoch": 0.5984,
+      "learning_rate": 1.903335292863301e-05,
+      "loss": 1.0426,
+      "step": 5984
+    },
+    {
+      "epoch": 0.5986,
+      "learning_rate": 1.903634525915866e-05,
+      "loss": 0.7692,
+      "step": 5986
+    },
+    {
+      "epoch": 0.5988,
+      "learning_rate": 1.9039333185479415e-05,
+      "loss": 0.5296,
+      "step": 5988
+    },
+    {
+      "epoch": 0.599,
+      "learning_rate": 1.904231670613899e-05,
+      "loss": 0.19,
+      "step": 5990
+    },
+    {
+      "epoch": 0.5992,
+      "learning_rate": 1.9045295819683244e-05,
+      "loss": 0.4952,
+      "step": 5992
+    },
+    {
+      "epoch": 0.5994,
+      "learning_rate": 1.904827052466019e-05,
+      "loss": 0.6625,
+      "step": 5994
+    },
+    {
+      "epoch": 0.5996,
+      "learning_rate": 1.9051240819620018e-05,
+      "loss": 0.3223,
+      "step": 5996
+    },
+    {
+      "epoch": 0.5998,
+      "learning_rate": 1.905420670311502e-05,
+      "loss": 0.8791,
+      "step": 5998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9057168173699658e-05,
+      "loss": 0.8119,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6002,
+      "learning_rate": 1.9060125229930572e-05,
+      "loss": 1.2922,
+      "step": 6002
+    },
+    {
+      "epoch": 0.6004,
+      "learning_rate": 1.9063077870366504e-05,
+      "loss": 0.7707,
+      "step": 6004
+    },
+    {
+      "epoch": 0.6006,
+      "learning_rate": 1.906602609356838e-05,
+      "loss": 0.6367,
+      "step": 6006
+    },
+    {
+      "epoch": 0.6008,
+      "learning_rate": 1.906896989809927e-05,
+      "loss": 1.0335,
+      "step": 6008
+    },
+    {
+      "epoch": 0.601,
+      "learning_rate": 1.907190928252441e-05,
+      "loss": 0.9497,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6012,
+      "learning_rate": 1.907484424541117e-05,
+      "loss": 0.9736,
+      "step": 6012
+    },
+    {
+      "epoch": 0.6014,
+      "learning_rate": 1.9077774785329078e-05,
+      "loss": 0.765,
+      "step": 6014
+    },
+    {
+      "epoch": 0.6016,
+      "learning_rate": 1.9080700900849855e-05,
+      "loss": 0.1885,
+      "step": 6016
+    },
+    {
+      "epoch": 0.6018,
+      "learning_rate": 1.908362259054731e-05,
+      "loss": 1.6732,
+      "step": 6018
+    },
+    {
+      "epoch": 0.602,
+      "learning_rate": 1.908653985299747e-05,
+      "loss": 0.6359,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6022,
+      "learning_rate": 1.9089452686778487e-05,
+      "loss": 0.1056,
+      "step": 6022
+    },
+    {
+      "epoch": 0.6024,
+      "learning_rate": 1.9092361090470685e-05,
+      "loss": 0.4952,
+      "step": 6024
+    },
+    {
+      "epoch": 0.6026,
+      "learning_rate": 1.9095265062656542e-05,
+      "loss": 0.8278,
+      "step": 6026
+    },
+    {
+      "epoch": 0.6028,
+      "learning_rate": 1.9098164601920702e-05,
+      "loss": 0.5883,
+      "step": 6028
+    },
+    {
+      "epoch": 0.603,
+      "learning_rate": 1.9101059706849957e-05,
+      "loss": 0.0838,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6032,
+      "learning_rate": 1.9103950376033272e-05,
+      "loss": 0.6436,
+      "step": 6032
+    },
+    {
+      "epoch": 0.6034,
+      "learning_rate": 1.910683660806177e-05,
+      "loss": 0.5312,
+      "step": 6034
+    },
+    {
+      "epoch": 0.6036,
+      "learning_rate": 1.9109718401528735e-05,
+      "loss": 0.6722,
+      "step": 6036
+    },
+    {
+      "epoch": 0.6038,
+      "learning_rate": 1.911259575502962e-05,
+      "loss": 0.769,
+      "step": 6038
+    },
+    {
+      "epoch": 0.604,
+      "learning_rate": 1.9115468667162034e-05,
+      "loss": 0.4084,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6042,
+      "learning_rate": 1.9118337136525754e-05,
+      "loss": 0.6263,
+      "step": 6042
+    },
+    {
+      "epoch": 0.6044,
+      "learning_rate": 1.9121201161722725e-05,
+      "loss": 0.2408,
+      "step": 6044
+    },
+    {
+      "epoch": 0.6046,
+      "learning_rate": 1.912406074135706e-05,
+      "loss": 0.1126,
+      "step": 6046
+    },
+    {
+      "epoch": 0.6048,
+      "learning_rate": 1.9126915874035025e-05,
+      "loss": 0.8325,
+      "step": 6048
+    },
+    {
+      "epoch": 0.605,
+      "learning_rate": 1.912976655836507e-05,
+      "loss": 0.673,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6052,
+      "learning_rate": 1.9132612792957805e-05,
+      "loss": 0.988,
+      "step": 6052
+    },
+    {
+      "epoch": 0.6054,
+      "learning_rate": 1.9135454576426006e-05,
+      "loss": 1.0916,
+      "step": 6054
+    },
+    {
+      "epoch": 0.6056,
+      "learning_rate": 1.9138291907384636e-05,
+      "loss": 0.5525,
+      "step": 6056
+    },
+    {
+      "epoch": 0.6058,
+      "learning_rate": 1.9141124784450786e-05,
+      "loss": 0.2048,
+      "step": 6058
+    },
+    {
+      "epoch": 0.606,
+      "learning_rate": 1.914395320624377e-05,
+      "loss": 1.1703,
+      "step": 6060
+    },
+    {
+      "epoch": 0.6062,
+      "learning_rate": 1.9146777171385053e-05,
+      "loss": 0.8668,
+      "step": 6062
+    },
+    {
+      "epoch": 0.6064,
+      "learning_rate": 1.914959667849825e-05,
+      "loss": 1.1175,
+      "step": 6064
+    },
+    {
+      "epoch": 0.6066,
+      "learning_rate": 1.9152411726209172e-05,
+      "loss": 1.0327,
+      "step": 6066
+    },
+    {
+      "epoch": 0.6068,
+      "learning_rate": 1.9155222313145817e-05,
+      "loss": 0.6638,
+      "step": 6068
+    },
+    {
+      "epoch": 0.607,
+      "learning_rate": 1.9158028437938316e-05,
+      "loss": 0.5814,
+      "step": 6070
+    },
+    {
+      "epoch": 0.6072,
+      "learning_rate": 1.9160830099219003e-05,
+      "loss": 0.3733,
+      "step": 6072
+    },
+    {
+      "epoch": 0.6074,
+      "learning_rate": 1.916362729562239e-05,
+      "loss": 1.1069,
+      "step": 6074
+    },
+    {
+      "epoch": 0.6076,
+      "learning_rate": 1.9166420025785165e-05,
+      "loss": 0.8687,
+      "step": 6076
+    },
+    {
+      "epoch": 0.6078,
+      "learning_rate": 1.9169208288346168e-05,
+      "loss": 0.9279,
+      "step": 6078
+    },
+    {
+      "epoch": 0.608,
+      "learning_rate": 1.917199208194643e-05,
+      "loss": 0.4845,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6082,
+      "learning_rate": 1.9174771405229187e-05,
+      "loss": 0.5467,
+      "step": 6082
+    },
+    {
+      "epoch": 0.6084,
+      "learning_rate": 1.9177546256839814e-05,
+      "loss": 1.26,
+      "step": 6084
+    },
+    {
+      "epoch": 0.6086,
+      "learning_rate": 1.9180316635425876e-05,
+      "loss": 1.0854,
+      "step": 6086
+    },
+    {
+      "epoch": 0.6088,
+      "learning_rate": 1.918308253963715e-05,
+      "loss": 0.1411,
+      "step": 6088
+    },
+    {
+      "epoch": 0.609,
+      "learning_rate": 1.9185843968125543e-05,
+      "loss": 1.0929,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6092,
+      "learning_rate": 1.9188600919545176e-05,
+      "loss": 0.6674,
+      "step": 6092
+    },
+    {
+      "epoch": 0.6094,
+      "learning_rate": 1.9191353392552346e-05,
+      "loss": 1.2208,
+      "step": 6094
+    },
+    {
+      "epoch": 0.6096,
+      "learning_rate": 1.919410138580553e-05,
+      "loss": 0.4747,
+      "step": 6096
+    },
+    {
+      "epoch": 0.6098,
+      "learning_rate": 1.919684489796539e-05,
+      "loss": 0.5874,
+      "step": 6098
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.919958392769477e-05,
+      "loss": 0.2206,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6102,
+      "learning_rate": 1.9202318473658703e-05,
+      "loss": 0.2277,
+      "step": 6102
+    },
+    {
+      "epoch": 0.6104,
+      "learning_rate": 1.92050485345244e-05,
+      "loss": 0.7216,
+      "step": 6104
+    },
+    {
+      "epoch": 0.6106,
+      "learning_rate": 1.9207774108961273e-05,
+      "loss": 0.5154,
+      "step": 6106
+    },
+    {
+      "epoch": 0.6108,
+      "learning_rate": 1.9210495195640895e-05,
+      "loss": 0.5403,
+      "step": 6108
+    },
+    {
+      "epoch": 0.611,
+      "learning_rate": 1.9213211793237052e-05,
+      "loss": 0.7894,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6112,
+      "learning_rate": 1.9215923900425703e-05,
+      "loss": 0.5595,
+      "step": 6112
+    },
+    {
+      "epoch": 0.6114,
+      "learning_rate": 1.9218631515885004e-05,
+      "loss": 0.1468,
+      "step": 6114
+    },
+    {
+      "epoch": 0.6116,
+      "learning_rate": 1.9221334638295292e-05,
+      "loss": 0.381,
+      "step": 6116
+    },
+    {
+      "epoch": 0.6118,
+      "learning_rate": 1.92240332663391e-05,
+      "loss": 0.6033,
+      "step": 6118
+    },
+    {
+      "epoch": 0.612,
+      "learning_rate": 1.9226727398701146e-05,
+      "loss": 0.7094,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6122,
+      "learning_rate": 1.922941703406835e-05,
+      "loss": 0.1417,
+      "step": 6122
+    },
+    {
+      "epoch": 0.6124,
+      "learning_rate": 1.9232102171129807e-05,
+      "loss": 1.1326,
+      "step": 6124
+    },
+    {
+      "epoch": 0.6126,
+      "learning_rate": 1.923478280857682e-05,
+      "loss": 0.2309,
+      "step": 6126
+    },
+    {
+      "epoch": 0.6128,
+      "learning_rate": 1.9237458945102885e-05,
+      "loss": 1.0785,
+      "step": 6128
+    },
+    {
+      "epoch": 0.613,
+      "learning_rate": 1.9240130579403663e-05,
+      "loss": 0.5818,
+      "step": 6130
+    },
+    {
+      "epoch": 0.6132,
+      "learning_rate": 1.9242797710177058e-05,
+      "loss": 0.6137,
+      "step": 6132
+    },
+    {
+      "epoch": 0.6134,
+      "learning_rate": 1.924546033612313e-05,
+      "loss": 0.9065,
+      "step": 6134
+    },
+    {
+      "epoch": 0.6136,
+      "learning_rate": 1.9248118455944156e-05,
+      "loss": 0.1451,
+      "step": 6136
+    },
+    {
+      "epoch": 0.6138,
+      "learning_rate": 1.9250772068344577e-05,
+      "loss": 0.3826,
+      "step": 6138
+    },
+    {
+      "epoch": 0.614,
+      "learning_rate": 1.9253421172031082e-05,
+      "loss": 0.0709,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6142,
+      "learning_rate": 1.9256065765712524e-05,
+      "loss": 1.1017,
+      "step": 6142
+    },
+    {
+      "epoch": 0.6144,
+      "learning_rate": 1.9258705848099944e-05,
+      "loss": 0.2256,
+      "step": 6144
+    },
+    {
+      "epoch": 0.6146,
+      "learning_rate": 1.9261341417906615e-05,
+      "loss": 0.7808,
+      "step": 6146
+    },
+    {
+      "epoch": 0.6148,
+      "learning_rate": 1.9263972473847995e-05,
+      "loss": 0.7201,
+      "step": 6148
+    },
+    {
+      "epoch": 0.615,
+      "learning_rate": 1.9266599014641724e-05,
+      "loss": 0.7382,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6152,
+      "learning_rate": 1.926922103900766e-05,
+      "loss": 0.7293,
+      "step": 6152
+    },
+    {
+      "epoch": 0.6154,
+      "learning_rate": 1.9271838545667876e-05,
+      "loss": 1.1019,
+      "step": 6154
+    },
+    {
+      "epoch": 0.6156,
+      "learning_rate": 1.9274451533346617e-05,
+      "loss": 0.6996,
+      "step": 6156
+    },
+    {
+      "epoch": 0.6158,
+      "learning_rate": 1.927706000077034e-05,
+      "loss": 0.5503,
+      "step": 6158
+    },
+    {
+      "epoch": 0.616,
+      "learning_rate": 1.9279663946667722e-05,
+      "loss": 0.5471,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6162,
+      "learning_rate": 1.9282263369769633e-05,
+      "loss": 1.4583,
+      "step": 6162
+    },
+    {
+      "epoch": 0.6164,
+      "learning_rate": 1.9284858268809135e-05,
+      "loss": 0.8943,
+      "step": 6164
+    },
+    {
+      "epoch": 0.6166,
+      "learning_rate": 1.9287448642521507e-05,
+      "loss": 1.0679,
+      "step": 6166
+    },
+    {
+      "epoch": 0.6168,
+      "learning_rate": 1.9290034489644247e-05,
+      "loss": 0.825,
+      "step": 6168
+    },
+    {
+      "epoch": 0.617,
+      "learning_rate": 1.9292615808917024e-05,
+      "loss": 0.4981,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6172,
+      "learning_rate": 1.9295192599081747e-05,
+      "loss": 0.6781,
+      "step": 6172
+    },
+    {
+      "epoch": 0.6174,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 0.7183,
+      "step": 6174
+    },
+    {
+      "epoch": 0.6176,
+      "learning_rate": 1.9300332587065644e-05,
+      "loss": 0.4596,
+      "step": 6176
+    },
+    {
+      "epoch": 0.6178,
+      "learning_rate": 1.9302895782379648e-05,
+      "loss": 0.0805,
+      "step": 6178
+    },
+    {
+      "epoch": 0.618,
+      "learning_rate": 1.930545444357526e-05,
+      "loss": 1.1415,
+      "step": 6180
+    },
+    {
+      "epoch": 0.6182,
+      "learning_rate": 1.9308008569405424e-05,
+      "loss": 0.3896,
+      "step": 6182
+    },
+    {
+      "epoch": 0.6184,
+      "learning_rate": 1.9310558158625283e-05,
+      "loss": 0.559,
+      "step": 6184
+    },
+    {
+      "epoch": 0.6186,
+      "learning_rate": 1.9313103209992205e-05,
+      "loss": 0.8923,
+      "step": 6186
+    },
+    {
+      "epoch": 0.6188,
+      "learning_rate": 1.9315643722265753e-05,
+      "loss": 0.6808,
+      "step": 6188
+    },
+    {
+      "epoch": 0.619,
+      "learning_rate": 1.9318179694207722e-05,
+      "loss": 0.7705,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6192,
+      "learning_rate": 1.932071112458211e-05,
+      "loss": 0.6507,
+      "step": 6192
+    },
+    {
+      "epoch": 0.6194,
+      "learning_rate": 1.932323801215512e-05,
+      "loss": 0.4922,
+      "step": 6194
+    },
+    {
+      "epoch": 0.6196,
+      "learning_rate": 1.9325760355695187e-05,
+      "loss": 1.5776,
+      "step": 6196
+    },
+    {
+      "epoch": 0.6198,
+      "learning_rate": 1.9328278153972943e-05,
+      "loss": 0.3535,
+      "step": 6198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9330791405761254e-05,
+      "loss": 1.3755,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6202,
+      "learning_rate": 1.933330010983518e-05,
+      "loss": 0.5374,
+      "step": 6202
+    },
+    {
+      "epoch": 0.6204,
+      "learning_rate": 1.9335804264972018e-05,
+      "loss": 0.7272,
+      "step": 6204
+    },
+    {
+      "epoch": 0.6206,
+      "learning_rate": 1.9338303869951266e-05,
+      "loss": 0.5603,
+      "step": 6206
+    },
+    {
+      "epoch": 0.6208,
+      "learning_rate": 1.934079892355466e-05,
+      "loss": 0.5128,
+      "step": 6208
+    },
+    {
+      "epoch": 0.621,
+      "learning_rate": 1.934328942456612e-05,
+      "loss": 0.8974,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6212,
+      "learning_rate": 1.9345775371771822e-05,
+      "loss": 0.7677,
+      "step": 6212
+    },
+    {
+      "epoch": 0.6214,
+      "learning_rate": 1.934825676396015e-05,
+      "loss": 0.698,
+      "step": 6214
+    },
+    {
+      "epoch": 0.6216,
+      "learning_rate": 1.935073359992168e-05,
+      "loss": 1.0186,
+      "step": 6216
+    },
+    {
+      "epoch": 0.6218,
+      "learning_rate": 1.9353205878449257e-05,
+      "loss": 0.0708,
+      "step": 6218
+    },
+    {
+      "epoch": 0.622,
+      "learning_rate": 1.935567359833791e-05,
+      "loss": 0.9595,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6222,
+      "learning_rate": 1.935813675838491e-05,
+      "loss": 0.3587,
+      "step": 6222
+    },
+    {
+      "epoch": 0.6224,
+      "learning_rate": 1.936059535738973e-05,
+      "loss": 0.5508,
+      "step": 6224
+    },
+    {
+      "epoch": 0.6226,
+      "learning_rate": 1.9363049394154088e-05,
+      "loss": 1.0517,
+      "step": 6226
+    },
+    {
+      "epoch": 0.6228,
+      "learning_rate": 1.9365498867481926e-05,
+      "loss": 0.5895,
+      "step": 6228
+    },
+    {
+      "epoch": 0.623,
+      "learning_rate": 1.9367943776179375e-05,
+      "loss": 1.7026,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6232,
+      "learning_rate": 1.9370384119054838e-05,
+      "loss": 0.2949,
+      "step": 6232
+    },
+    {
+      "epoch": 0.6234,
+      "learning_rate": 1.937281989491892e-05,
+      "loss": 2.1653,
+      "step": 6234
+    },
+    {
+      "epoch": 0.6236,
+      "learning_rate": 1.9375251102584438e-05,
+      "loss": 1.2023,
+      "step": 6236
+    },
+    {
+      "epoch": 0.6238,
+      "learning_rate": 1.9377677740866457e-05,
+      "loss": 0.5306,
+      "step": 6238
+    },
+    {
+      "epoch": 0.624,
+      "learning_rate": 1.9380099808582278e-05,
+      "loss": 1.0926,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6242,
+      "learning_rate": 1.9382517304551397e-05,
+      "loss": 0.2909,
+      "step": 6242
+    },
+    {
+      "epoch": 0.6244,
+      "learning_rate": 1.938493022759556e-05,
+      "loss": 0.8634,
+      "step": 6244
+    },
+    {
+      "epoch": 0.6246,
+      "learning_rate": 1.9387338576538743e-05,
+      "loss": 0.8931,
+      "step": 6246
+    },
+    {
+      "epoch": 0.6248,
+      "learning_rate": 1.938974235020714e-05,
+      "loss": 1.1322,
+      "step": 6248
+    },
+    {
+      "epoch": 0.625,
+      "learning_rate": 1.9392141547429183e-05,
+      "loss": 0.1164,
+      "step": 6250
+    },
+    {
+      "epoch": 0.6252,
+      "learning_rate": 1.9394536167035528e-05,
+      "loss": 0.5416,
+      "step": 6252
+    },
+    {
+      "epoch": 0.6254,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.8492,
+      "step": 6254
+    },
+    {
+      "epoch": 0.6256,
+      "learning_rate": 1.9399311668734957e-05,
+      "loss": 0.5058,
+      "step": 6256
+    },
+    {
+      "epoch": 0.6258,
+      "learning_rate": 1.9401692548500504e-05,
+      "loss": 0.6652,
+      "step": 6258
+    },
+    {
+      "epoch": 0.626,
+      "learning_rate": 1.9404068845995317e-05,
+      "loss": 0.9572,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6262,
+      "learning_rate": 1.9406440560061214e-05,
+      "loss": 0.7985,
+      "step": 6262
+    },
+    {
+      "epoch": 0.6264,
+      "learning_rate": 1.9408807689542253e-05,
+      "loss": 1.2814,
+      "step": 6264
+    },
+    {
+      "epoch": 0.6266,
+      "learning_rate": 1.9411170233284728e-05,
+      "loss": 1.0008,
+      "step": 6266
+    },
+    {
+      "epoch": 0.6268,
+      "learning_rate": 1.9413528190137158e-05,
+      "loss": 1.083,
+      "step": 6268
+    },
+    {
+      "epoch": 0.627,
+      "learning_rate": 1.9415881558950302e-05,
+      "loss": 0.4331,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6272,
+      "learning_rate": 1.9418230338577164e-05,
+      "loss": 0.6064,
+      "step": 6272
+    },
+    {
+      "epoch": 0.6274,
+      "learning_rate": 1.942057452787297e-05,
+      "loss": 0.9403,
+      "step": 6274
+    },
+    {
+      "epoch": 0.6276,
+      "learning_rate": 1.942291412569519e-05,
+      "loss": 0.5391,
+      "step": 6276
+    },
+    {
+      "epoch": 0.6278,
+      "learning_rate": 1.942524913090354e-05,
+      "loss": 0.7813,
+      "step": 6278
+    },
+    {
+      "epoch": 0.628,
+      "learning_rate": 1.9427579542359966e-05,
+      "loss": 0.5151,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6282,
+      "learning_rate": 1.9429905358928645e-05,
+      "loss": 0.6268,
+      "step": 6282
+    },
+    {
+      "epoch": 0.6284,
+      "learning_rate": 1.943222657947601e-05,
+      "loss": 0.3319,
+      "step": 6284
+    },
+    {
+      "epoch": 0.6286,
+      "learning_rate": 1.9434543202870723e-05,
+      "loss": 0.5029,
+      "step": 6286
+    },
+    {
+      "epoch": 0.6288,
+      "learning_rate": 1.9436855227983692e-05,
+      "loss": 0.3287,
+      "step": 6288
+    },
+    {
+      "epoch": 0.629,
+      "learning_rate": 1.9439162653688063e-05,
+      "loss": 0.369,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6292,
+      "learning_rate": 1.9441465478859226e-05,
+      "loss": 0.2065,
+      "step": 6292
+    },
+    {
+      "epoch": 0.6294,
+      "learning_rate": 1.9443763702374815e-05,
+      "loss": 0.7589,
+      "step": 6294
+    },
+    {
+      "epoch": 0.6296,
+      "learning_rate": 1.944605732311469e-05,
+      "loss": 0.9574,
+      "step": 6296
+    },
+    {
+      "epoch": 0.6298,
+      "learning_rate": 1.944834633996098e-05,
+      "loss": 0.9472,
+      "step": 6298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9450630751798053e-05,
+      "loss": 1.1008,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6302,
+      "learning_rate": 1.9452910557512494e-05,
+      "loss": 0.0187,
+      "step": 6302
+    },
+    {
+      "epoch": 0.6304,
+      "learning_rate": 1.9455185755993167e-05,
+      "loss": 1.0381,
+      "step": 6304
+    },
+    {
+      "epoch": 0.6306,
+      "learning_rate": 1.9457456346131172e-05,
+      "loss": 0.6499,
+      "step": 6306
+    },
+    {
+      "epoch": 0.6308,
+      "learning_rate": 1.945972232681984e-05,
+      "loss": 0.5,
+      "step": 6308
+    },
+    {
+      "epoch": 0.631,
+      "learning_rate": 1.9461983696954756e-05,
+      "loss": 0.8968,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6312,
+      "learning_rate": 1.946424045543377e-05,
+      "loss": 1.5028,
+      "step": 6312
+    },
+    {
+      "epoch": 0.6314,
+      "learning_rate": 1.9466492601156964e-05,
+      "loss": 0.285,
+      "step": 6314
+    },
+    {
+      "epoch": 0.6316,
+      "learning_rate": 1.9468740133026655e-05,
+      "loss": 0.8537,
+      "step": 6316
+    },
+    {
+      "epoch": 0.6318,
+      "learning_rate": 1.947098304994744e-05,
+      "loss": 0.1228,
+      "step": 6318
+    },
+    {
+      "epoch": 0.632,
+      "learning_rate": 1.9473221350826145e-05,
+      "loss": 0.746,
+      "step": 6320
+    },
+    {
+      "epoch": 0.6322,
+      "learning_rate": 1.947545503457184e-05,
+      "loss": 0.5654,
+      "step": 6322
+    },
+    {
+      "epoch": 0.6324,
+      "learning_rate": 1.9477684100095856e-05,
+      "loss": 0.7796,
+      "step": 6324
+    },
+    {
+      "epoch": 0.6326,
+      "learning_rate": 1.9479908546311787e-05,
+      "loss": 0.398,
+      "step": 6326
+    },
+    {
+      "epoch": 0.6328,
+      "learning_rate": 1.9482128372135446e-05,
+      "loss": 0.4897,
+      "step": 6328
+    },
+    {
+      "epoch": 0.633,
+      "learning_rate": 1.9484343576484935e-05,
+      "loss": 1.1461,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6332,
+      "learning_rate": 1.9486554158280576e-05,
+      "loss": 0.4995,
+      "step": 6332
+    },
+    {
+      "epoch": 0.6334,
+      "learning_rate": 1.9488760116444966e-05,
+      "loss": 0.6295,
+      "step": 6334
+    },
+    {
+      "epoch": 0.6336,
+      "learning_rate": 1.9490961449902946e-05,
+      "loss": 0.7584,
+      "step": 6336
+    },
+    {
+      "epoch": 0.6338,
+      "learning_rate": 1.949315815758161e-05,
+      "loss": 0.4959,
+      "step": 6338
+    },
+    {
+      "epoch": 0.634,
+      "learning_rate": 1.949535023841032e-05,
+      "loss": 0.5006,
+      "step": 6340
+    },
+    {
+      "epoch": 0.6342,
+      "learning_rate": 1.949753769132067e-05,
+      "loss": 1.069,
+      "step": 6342
+    },
+    {
+      "epoch": 0.6344,
+      "learning_rate": 1.9499720515246524e-05,
+      "loss": 1.0422,
+      "step": 6344
+    },
+    {
+      "epoch": 0.6346,
+      "learning_rate": 1.9501898709124008e-05,
+      "loss": 0.8946,
+      "step": 6346
+    },
+    {
+      "epoch": 0.6348,
+      "learning_rate": 1.9504072271891486e-05,
+      "loss": 0.6873,
+      "step": 6348
+    },
+    {
+      "epoch": 0.635,
+      "learning_rate": 1.95062412024896e-05,
+      "loss": 0.6749,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6352,
+      "learning_rate": 1.9508405499861235e-05,
+      "loss": 0.7899,
+      "step": 6352
+    },
+    {
+      "epoch": 0.6354,
+      "learning_rate": 1.9510565162951534e-05,
+      "loss": 0.7097,
+      "step": 6354
+    },
+    {
+      "epoch": 0.6356,
+      "learning_rate": 1.951272019070791e-05,
+      "loss": 0.6479,
+      "step": 6356
+    },
+    {
+      "epoch": 0.6358,
+      "learning_rate": 1.951487058208003e-05,
+      "loss": 0.6627,
+      "step": 6358
+    },
+    {
+      "epoch": 0.636,
+      "learning_rate": 1.9517016336019817e-05,
+      "loss": 0.2002,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6362,
+      "learning_rate": 1.9519157451481453e-05,
+      "loss": 0.2721,
+      "step": 6362
+    },
+    {
+      "epoch": 0.6364,
+      "learning_rate": 1.9521293927421388e-05,
+      "loss": 0.3261,
+      "step": 6364
+    },
+    {
+      "epoch": 0.6366,
+      "learning_rate": 1.952342576279833e-05,
+      "loss": 1.147,
+      "step": 6366
+    },
+    {
+      "epoch": 0.6368,
+      "learning_rate": 1.952555295657324e-05,
+      "loss": 0.7795,
+      "step": 6368
+    },
+    {
+      "epoch": 0.637,
+      "learning_rate": 1.9527675507709364e-05,
+      "loss": 0.9318,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6372,
+      "learning_rate": 1.9529793415172188e-05,
+      "loss": 0.1812,
+      "step": 6372
+    },
+    {
+      "epoch": 0.6374,
+      "learning_rate": 1.953190667792947e-05,
+      "loss": 1.0547,
+      "step": 6374
+    },
+    {
+      "epoch": 0.6376,
+      "learning_rate": 1.953401529495123e-05,
+      "loss": 1.1365,
+      "step": 6376
+    },
+    {
+      "epoch": 0.6378,
+      "learning_rate": 1.953611926520976e-05,
+      "loss": 0.7905,
+      "step": 6378
+    },
+    {
+      "epoch": 0.638,
+      "learning_rate": 1.953821858767961e-05,
+      "loss": 0.991,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6382,
+      "learning_rate": 1.9540313261337578e-05,
+      "loss": 0.9667,
+      "step": 6382
+    },
+    {
+      "epoch": 0.6384,
+      "learning_rate": 1.9542403285162765e-05,
+      "loss": 0.6979,
+      "step": 6384
+    },
+    {
+      "epoch": 0.6386,
+      "learning_rate": 1.9544488658136522e-05,
+      "loss": 0.6985,
+      "step": 6386
+    },
+    {
+      "epoch": 0.6388,
+      "learning_rate": 1.9546569379242443e-05,
+      "loss": 0.4258,
+      "step": 6388
+    },
+    {
+      "epoch": 0.639,
+      "learning_rate": 1.954864544746643e-05,
+      "loss": 0.9985,
+      "step": 6390
+    },
+    {
+      "epoch": 0.6392,
+      "learning_rate": 1.9550716861796623e-05,
+      "loss": 0.2911,
+      "step": 6392
+    },
+    {
+      "epoch": 0.6394,
+      "learning_rate": 1.955278362122344e-05,
+      "loss": 0.6523,
+      "step": 6394
+    },
+    {
+      "epoch": 0.6396,
+      "learning_rate": 1.955484572473956e-05,
+      "loss": 0.8484,
+      "step": 6396
+    },
+    {
+      "epoch": 0.6398,
+      "learning_rate": 1.955690317133996e-05,
+      "loss": 0.6521,
+      "step": 6398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.955895596002185e-05,
+      "loss": 0.8122,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6402,
+      "learning_rate": 1.9561004089784726e-05,
+      "loss": 1.6795,
+      "step": 6402
+    },
+    {
+      "epoch": 0.6404,
+      "learning_rate": 1.9563047559630353e-05,
+      "loss": 0.5149,
+      "step": 6404
+    },
+    {
+      "epoch": 0.6406,
+      "learning_rate": 1.956508636856278e-05,
+      "loss": 0.5054,
+      "step": 6406
+    },
+    {
+      "epoch": 0.6408,
+      "learning_rate": 1.9567120515588307e-05,
+      "loss": 0.3117,
+      "step": 6408
+    },
+    {
+      "epoch": 0.641,
+      "learning_rate": 1.956914999971551e-05,
+      "loss": 0.2776,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6412,
+      "learning_rate": 1.9571174819955264e-05,
+      "loss": 1.1636,
+      "step": 6412
+    },
+    {
+      "epoch": 0.6414,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 1.4398,
+      "step": 6414
+    },
+    {
+      "epoch": 0.6416,
+      "learning_rate": 1.957521046482715e-05,
+      "loss": 0.5143,
+      "step": 6416
+    },
+    {
+      "epoch": 0.6418,
+      "learning_rate": 1.9577221287492368e-05,
+      "loss": 0.5924,
+      "step": 6418
+    },
+    {
+      "epoch": 0.642,
+      "learning_rate": 1.9579227442336276e-05,
+      "loss": 0.999,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6422,
+      "learning_rate": 1.95812289283811e-05,
+      "loss": 0.2818,
+      "step": 6422
+    },
+    {
+      "epoch": 0.6424,
+      "learning_rate": 1.9583225744651327e-05,
+      "loss": 0.5937,
+      "step": 6424
+    },
+    {
+      "epoch": 0.6426,
+      "learning_rate": 1.958521789017376e-05,
+      "loss": 1.0471,
+      "step": 6426
+    },
+    {
+      "epoch": 0.6428,
+      "learning_rate": 1.9587205363977428e-05,
+      "loss": 0.7323,
+      "step": 6428
+    },
+    {
+      "epoch": 0.643,
+      "learning_rate": 1.958918816509367e-05,
+      "loss": 0.2236,
+      "step": 6430
+    },
+    {
+      "epoch": 0.6432,
+      "learning_rate": 1.959116629255609e-05,
+      "loss": 0.6326,
+      "step": 6432
+    },
+    {
+      "epoch": 0.6434,
+      "learning_rate": 1.9593139745400575e-05,
+      "loss": 0.4995,
+      "step": 6434
+    },
+    {
+      "epoch": 0.6436,
+      "learning_rate": 1.959510852266529e-05,
+      "loss": 0.4601,
+      "step": 6436
+    },
+    {
+      "epoch": 0.6438,
+      "learning_rate": 1.9597072623390668e-05,
+      "loss": 0.145,
+      "step": 6438
+    },
+    {
+      "epoch": 0.644,
+      "learning_rate": 1.9599032046619437e-05,
+      "loss": 0.7884,
+      "step": 6440
+    },
+    {
+      "epoch": 0.6442,
+      "learning_rate": 1.9600986791396597e-05,
+      "loss": 0.5316,
+      "step": 6442
+    },
+    {
+      "epoch": 0.6444,
+      "learning_rate": 1.960293685676943e-05,
+      "loss": 0.4871,
+      "step": 6444
+    },
+    {
+      "epoch": 0.6446,
+      "learning_rate": 1.9604882241787496e-05,
+      "loss": 0.1725,
+      "step": 6446
+    },
+    {
+      "epoch": 0.6448,
+      "learning_rate": 1.9606822945502642e-05,
+      "loss": 0.7875,
+      "step": 6448
+    },
+    {
+      "epoch": 0.645,
+      "learning_rate": 1.9608758966968983e-05,
+      "loss": 0.5781,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6452,
+      "learning_rate": 1.9610690305242943e-05,
+      "loss": 0.8487,
+      "step": 6452
+    },
+    {
+      "epoch": 0.6454,
+      "learning_rate": 1.9612616959383187e-05,
+      "loss": 0.2626,
+      "step": 6454
+    },
+    {
+      "epoch": 0.6456,
+      "learning_rate": 1.9614538928450707e-05,
+      "loss": 0.4516,
+      "step": 6456
+    },
+    {
+      "epoch": 0.6458,
+      "learning_rate": 1.9616456211508752e-05,
+      "loss": 0.2733,
+      "step": 6458
+    },
+    {
+      "epoch": 0.646,
+      "learning_rate": 1.961836880762286e-05,
+      "loss": 1.4636,
+      "step": 6460
+    },
+    {
+      "epoch": 0.6462,
+      "learning_rate": 1.9620276715860856e-05,
+      "loss": 0.4938,
+      "step": 6462
+    },
+    {
+      "epoch": 0.6464,
+      "learning_rate": 1.962217993529285e-05,
+      "loss": 0.4855,
+      "step": 6464
+    },
+    {
+      "epoch": 0.6466,
+      "learning_rate": 1.962407846499124e-05,
+      "loss": 0.803,
+      "step": 6466
+    },
+    {
+      "epoch": 0.6468,
+      "learning_rate": 1.9625972304030694e-05,
+      "loss": 0.9655,
+      "step": 6468
+    },
+    {
+      "epoch": 0.647,
+      "learning_rate": 1.9627861451488187e-05,
+      "loss": 0.3613,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6472,
+      "learning_rate": 1.9629745906442973e-05,
+      "loss": 1.795,
+      "step": 6472
+    },
+    {
+      "epoch": 0.6474,
+      "learning_rate": 1.9631625667976584e-05,
+      "loss": 0.694,
+      "step": 6474
+    },
+    {
+      "epoch": 0.6476,
+      "learning_rate": 1.9633500735172845e-05,
+      "loss": 0.4543,
+      "step": 6476
+    },
+    {
+      "epoch": 0.6478,
+      "learning_rate": 1.963537110711789e-05,
+      "loss": 0.6582,
+      "step": 6478
+    },
+    {
+      "epoch": 0.648,
+      "learning_rate": 1.96372367829001e-05,
+      "loss": 1.0414,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6482,
+      "learning_rate": 1.9639097761610174e-05,
+      "loss": 0.9352,
+      "step": 6482
+    },
+    {
+      "epoch": 0.6484,
+      "learning_rate": 1.9640954042341098e-05,
+      "loss": 0.7426,
+      "step": 6484
+    },
+    {
+      "epoch": 0.6486,
+      "learning_rate": 1.964280562418815e-05,
+      "loss": 0.5398,
+      "step": 6486
+    },
+    {
+      "epoch": 0.6488,
+      "learning_rate": 1.9644652506248872e-05,
+      "loss": 0.5119,
+      "step": 6488
+    },
+    {
+      "epoch": 0.649,
+      "learning_rate": 1.964649468762313e-05,
+      "loss": 1.2723,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6492,
+      "learning_rate": 1.9648332167413067e-05,
+      "loss": 1.1536,
+      "step": 6492
+    },
+    {
+      "epoch": 0.6494,
+      "learning_rate": 1.9650164944723116e-05,
+      "loss": 1.0777,
+      "step": 6494
+    },
+    {
+      "epoch": 0.6496,
+      "learning_rate": 1.9651993018659995e-05,
+      "loss": 0.3607,
+      "step": 6496
+    },
+    {
+      "epoch": 0.6498,
+      "learning_rate": 1.965381638833274e-05,
+      "loss": 0.5597,
+      "step": 6498
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9655635052852648e-05,
+      "loss": 1.4034,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6502,
+      "learning_rate": 1.9657449011333328e-05,
+      "loss": 0.0196,
+      "step": 6502
+    },
+    {
+      "epoch": 0.6504,
+      "learning_rate": 1.9659258262890683e-05,
+      "loss": 1.1992,
+      "step": 6504
+    },
+    {
+      "epoch": 0.6506,
+      "learning_rate": 1.96610628066429e-05,
+      "loss": 0.8725,
+      "step": 6506
+    },
+    {
+      "epoch": 0.6508,
+      "learning_rate": 1.966286264171047e-05,
+      "loss": 1.2445,
+      "step": 6508
+    },
+    {
+      "epoch": 0.651,
+      "learning_rate": 1.9664657767216176e-05,
+      "loss": 0.992,
+      "step": 6510
+    },
+    {
+      "epoch": 0.6512,
+      "learning_rate": 1.9666448182285092e-05,
+      "loss": 0.8649,
+      "step": 6512
+    },
+    {
+      "epoch": 0.6514,
+      "learning_rate": 1.9668233886044594e-05,
+      "loss": 0.3111,
+      "step": 6514
+    },
+    {
+      "epoch": 0.6516,
+      "learning_rate": 1.967001487762435e-05,
+      "loss": 0.6394,
+      "step": 6516
+    },
+    {
+      "epoch": 0.6518,
+      "learning_rate": 1.967179115615633e-05,
+      "loss": 0.6592,
+      "step": 6518
+    },
+    {
+      "epoch": 0.652,
+      "learning_rate": 1.967356272077479e-05,
+      "loss": 0.5307,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6522,
+      "learning_rate": 1.96753295706163e-05,
+      "loss": 0.6874,
+      "step": 6522
+    },
+    {
+      "epoch": 0.6524,
+      "learning_rate": 1.967709170481971e-05,
+      "loss": 0.2837,
+      "step": 6524
+    },
+    {
+      "epoch": 0.6526,
+      "learning_rate": 1.967884912252619e-05,
+      "loss": 0.822,
+      "step": 6526
+    },
+    {
+      "epoch": 0.6528,
+      "learning_rate": 1.9680601822879178e-05,
+      "loss": 0.9464,
+      "step": 6528
+    },
+    {
+      "epoch": 0.653,
+      "learning_rate": 1.9682349805024443e-05,
+      "loss": 0.5862,
+      "step": 6530
+    },
+    {
+      "epoch": 0.6532,
+      "learning_rate": 1.968409306811004e-05,
+      "loss": 0.4214,
+      "step": 6532
+    },
+    {
+      "epoch": 0.6534,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 0.5489,
+      "step": 6534
+    },
+    {
+      "epoch": 0.6536,
+      "learning_rate": 1.9687565433705923e-05,
+      "loss": 0.2758,
+      "step": 6536
+    },
+    {
+      "epoch": 0.6538,
+      "learning_rate": 1.9689294534523833e-05,
+      "loss": 0.5652,
+      "step": 6538
+    },
+    {
+      "epoch": 0.654,
+      "learning_rate": 1.9691018912897285e-05,
+      "loss": 1.0426,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6542,
+      "learning_rate": 1.969273856798585e-05,
+      "loss": 0.8072,
+      "step": 6542
+    },
+    {
+      "epoch": 0.6544,
+      "learning_rate": 1.9694453498951392e-05,
+      "loss": 0.8643,
+      "step": 6544
+    },
+    {
+      "epoch": 0.6546,
+      "learning_rate": 1.969616370495806e-05,
+      "loss": 0.8278,
+      "step": 6546
+    },
+    {
+      "epoch": 0.6548,
+      "learning_rate": 1.969786918517233e-05,
+      "loss": 0.6908,
+      "step": 6548
+    },
+    {
+      "epoch": 0.655,
+      "learning_rate": 1.9699569938762972e-05,
+      "loss": 0.9074,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6552,
+      "learning_rate": 1.9701265964901063e-05,
+      "loss": 0.4432,
+      "step": 6552
+    },
+    {
+      "epoch": 0.6554,
+      "learning_rate": 1.9702957262759964e-05,
+      "loss": 0.0957,
+      "step": 6554
+    },
+    {
+      "epoch": 0.6556,
+      "learning_rate": 1.970464383151537e-05,
+      "loss": 0.8715,
+      "step": 6556
+    },
+    {
+      "epoch": 0.6558,
+      "learning_rate": 1.9706325670345276e-05,
+      "loss": 0.5875,
+      "step": 6558
+    },
+    {
+      "epoch": 0.656,
+      "learning_rate": 1.9708002778429957e-05,
+      "loss": 0.5681,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6562,
+      "learning_rate": 1.9709675154952013e-05,
+      "loss": 0.7997,
+      "step": 6562
+    },
+    {
+      "epoch": 0.6564,
+      "learning_rate": 1.971134279909636e-05,
+      "loss": 0.3837,
+      "step": 6564
+    },
+    {
+      "epoch": 0.6566,
+      "learning_rate": 1.9713005710050203e-05,
+      "loss": 0.6581,
+      "step": 6566
+    },
+    {
+      "epoch": 0.6568,
+      "learning_rate": 1.9714663887003048e-05,
+      "loss": 0.8877,
+      "step": 6568
+    },
+    {
+      "epoch": 0.657,
+      "learning_rate": 1.971631732914674e-05,
+      "loss": 0.7989,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6572,
+      "learning_rate": 1.97179660356754e-05,
+      "loss": 0.8855,
+      "step": 6572
+    },
+    {
+      "epoch": 0.6574,
+      "learning_rate": 1.9719610005785466e-05,
+      "loss": 0.6966,
+      "step": 6574
+    },
+    {
+      "epoch": 0.6576,
+      "learning_rate": 1.9721249238675684e-05,
+      "loss": 0.7784,
+      "step": 6576
+    },
+    {
+      "epoch": 0.6578,
+      "learning_rate": 1.9722883733547128e-05,
+      "loss": 0.8129,
+      "step": 6578
+    },
+    {
+      "epoch": 0.658,
+      "learning_rate": 1.9724513489603153e-05,
+      "loss": 0.5568,
+      "step": 6580
+    },
+    {
+      "epoch": 0.6582,
+      "learning_rate": 1.9726138506049434e-05,
+      "loss": 0.4073,
+      "step": 6582
+    },
+    {
+      "epoch": 0.6584,
+      "learning_rate": 1.9727758782093966e-05,
+      "loss": 1.0809,
+      "step": 6584
+    },
+    {
+      "epoch": 0.6586,
+      "learning_rate": 1.972937431694704e-05,
+      "loss": 0.4216,
+      "step": 6586
+    },
+    {
+      "epoch": 0.6588,
+      "learning_rate": 1.9730985109821268e-05,
+      "loss": 0.4151,
+      "step": 6588
+    },
+    {
+      "epoch": 0.659,
+      "learning_rate": 1.9732591159931564e-05,
+      "loss": 0.5976,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6592,
+      "learning_rate": 1.9734192466495162e-05,
+      "loss": 1.1432,
+      "step": 6592
+    },
+    {
+      "epoch": 0.6594,
+      "learning_rate": 1.9735789028731603e-05,
+      "loss": 0.3082,
+      "step": 6594
+    },
+    {
+      "epoch": 0.6596,
+      "learning_rate": 1.9737380845862745e-05,
+      "loss": 0.9828,
+      "step": 6596
+    },
+    {
+      "epoch": 0.6598,
+      "learning_rate": 1.9738967917112752e-05,
+      "loss": 0.1719,
+      "step": 6598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9740550241708107e-05,
+      "loss": 0.7031,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6602,
+      "learning_rate": 1.9742127818877605e-05,
+      "loss": 0.3148,
+      "step": 6602
+    },
+    {
+      "epoch": 0.6604,
+      "learning_rate": 1.9743700647852352e-05,
+      "loss": 0.3568,
+      "step": 6604
+    },
+    {
+      "epoch": 0.6606,
+      "learning_rate": 1.974526872786577e-05,
+      "loss": 0.0367,
+      "step": 6606
+    },
+    {
+      "epoch": 0.6608,
+      "learning_rate": 1.97468320581536e-05,
+      "loss": 1.178,
+      "step": 6608
+    },
+    {
+      "epoch": 0.661,
+      "learning_rate": 1.974839063795389e-05,
+      "loss": 0.1265,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6612,
+      "learning_rate": 1.9749944466507007e-05,
+      "loss": 0.003,
+      "step": 6612
+    },
+    {
+      "epoch": 0.6614,
+      "learning_rate": 1.9751493543055634e-05,
+      "loss": 0.9209,
+      "step": 6614
+    },
+    {
+      "epoch": 0.6616,
+      "learning_rate": 1.975303786684477e-05,
+      "loss": 0.7785,
+      "step": 6616
+    },
+    {
+      "epoch": 0.6618,
+      "learning_rate": 1.975457743712173e-05,
+      "loss": 0.5712,
+      "step": 6618
+    },
+    {
+      "epoch": 0.662,
+      "learning_rate": 1.9756112253136147e-05,
+      "loss": 0.5378,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6622,
+      "learning_rate": 1.9757642314139977e-05,
+      "loss": 0.8092,
+      "step": 6622
+    },
+    {
+      "epoch": 0.6624,
+      "learning_rate": 1.9759167619387474e-05,
+      "loss": 0.9979,
+      "step": 6624
+    },
+    {
+      "epoch": 0.6626,
+      "learning_rate": 1.976068816813523e-05,
+      "loss": 0.0785,
+      "step": 6626
+    },
+    {
+      "epoch": 0.6628,
+      "learning_rate": 1.976220395964215e-05,
+      "loss": 0.3789,
+      "step": 6628
+    },
+    {
+      "epoch": 0.663,
+      "learning_rate": 1.976371499316945e-05,
+      "loss": 0.6069,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6632,
+      "learning_rate": 1.9765221267980675e-05,
+      "loss": 0.5431,
+      "step": 6632
+    },
+    {
+      "epoch": 0.6634,
+      "learning_rate": 1.9766722783341675e-05,
+      "loss": 0.9178,
+      "step": 6634
+    },
+    {
+      "epoch": 0.6636,
+      "learning_rate": 1.9768219538520648e-05,
+      "loss": 0.6569,
+      "step": 6636
+    },
+    {
+      "epoch": 0.6638,
+      "learning_rate": 1.9769711532788083e-05,
+      "loss": 1.2159,
+      "step": 6638
+    },
+    {
+      "epoch": 0.664,
+      "learning_rate": 1.9771198765416795e-05,
+      "loss": 0.7245,
+      "step": 6640
+    },
+    {
+      "epoch": 0.6642,
+      "learning_rate": 1.9772681235681933e-05,
+      "loss": 1.0867,
+      "step": 6642
+    },
+    {
+      "epoch": 0.6644,
+      "learning_rate": 1.9774158942860962e-05,
+      "loss": 0.6767,
+      "step": 6644
+    },
+    {
+      "epoch": 0.6646,
+      "learning_rate": 1.9775631886233655e-05,
+      "loss": 0.8909,
+      "step": 6646
+    },
+    {
+      "epoch": 0.6648,
+      "learning_rate": 1.9777100065082116e-05,
+      "loss": 1.1508,
+      "step": 6648
+    },
+    {
+      "epoch": 0.665,
+      "learning_rate": 1.977856347869079e-05,
+      "loss": 1.0116,
+      "step": 6650
+    },
+    {
+      "epoch": 0.6652,
+      "learning_rate": 1.9780022126346413e-05,
+      "loss": 1.3053,
+      "step": 6652
+    },
+    {
+      "epoch": 0.6654,
+      "learning_rate": 1.9781476007338054e-05,
+      "loss": 0.8677,
+      "step": 6654
+    },
+    {
+      "epoch": 0.6656,
+      "learning_rate": 1.9782925120957123e-05,
+      "loss": 1.3501,
+      "step": 6656
+    },
+    {
+      "epoch": 0.6658,
+      "learning_rate": 1.9784369466497333e-05,
+      "loss": 0.8147,
+      "step": 6658
+    },
+    {
+      "epoch": 0.666,
+      "learning_rate": 1.978580904325472e-05,
+      "loss": 0.9562,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6662,
+      "learning_rate": 1.978724385052766e-05,
+      "loss": 0.4827,
+      "step": 6662
+    },
+    {
+      "epoch": 0.6664,
+      "learning_rate": 1.9788673887616852e-05,
+      "loss": 0.6929,
+      "step": 6664
+    },
+    {
+      "epoch": 0.6666,
+      "learning_rate": 1.97900991538253e-05,
+      "loss": 1.1944,
+      "step": 6666
+    },
+    {
+      "epoch": 0.6668,
+      "learning_rate": 1.9791519648458352e-05,
+      "loss": 1.1636,
+      "step": 6668
+    },
+    {
+      "epoch": 0.667,
+      "learning_rate": 1.9792935370823673e-05,
+      "loss": 1.2506,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6672,
+      "learning_rate": 1.9794346320231265e-05,
+      "loss": 0.9316,
+      "step": 6672
+    },
+    {
+      "epoch": 0.6674,
+      "learning_rate": 1.979575249599344e-05,
+      "loss": 0.433,
+      "step": 6674
+    },
+    {
+      "epoch": 0.6676,
+      "learning_rate": 1.9797153897424854e-05,
+      "loss": 0.5374,
+      "step": 6676
+    },
+    {
+      "epoch": 0.6678,
+      "learning_rate": 1.979855052384247e-05,
+      "loss": 0.8815,
+      "step": 6678
+    },
+    {
+      "epoch": 0.668,
+      "learning_rate": 1.9799942374565594e-05,
+      "loss": 0.2334,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6682,
+      "learning_rate": 1.980132944891586e-05,
+      "loss": 0.619,
+      "step": 6682
+    },
+    {
+      "epoch": 0.6684,
+      "learning_rate": 1.980271174621722e-05,
+      "loss": 0.0614,
+      "step": 6684
+    },
+    {
+      "epoch": 0.6686,
+      "learning_rate": 1.9804089265795956e-05,
+      "loss": 0.1703,
+      "step": 6686
+    },
+    {
+      "epoch": 0.6688,
+      "learning_rate": 1.9805462006980688e-05,
+      "loss": 0.8638,
+      "step": 6688
+    },
+    {
+      "epoch": 0.669,
+      "learning_rate": 1.9806829969102356e-05,
+      "loss": 0.2094,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6692,
+      "learning_rate": 1.980819315149423e-05,
+      "loss": 1.3413,
+      "step": 6692
+    },
+    {
+      "epoch": 0.6694,
+      "learning_rate": 1.9809551553491918e-05,
+      "loss": 0.4179,
+      "step": 6694
+    },
+    {
+      "epoch": 0.6696,
+      "learning_rate": 1.981090517443334e-05,
+      "loss": 0.6951,
+      "step": 6696
+    },
+    {
+      "epoch": 0.6698,
+      "learning_rate": 1.981225401365877e-05,
+      "loss": 0.9403,
+      "step": 6698
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9813598070510787e-05,
+      "loss": 0.0022,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6702,
+      "learning_rate": 1.981493734433433e-05,
+      "loss": 1.0715,
+      "step": 6702
+    },
+    {
+      "epoch": 0.6704,
+      "learning_rate": 1.9816271834476642e-05,
+      "loss": 1.0681,
+      "step": 6704
+    },
+    {
+      "epoch": 0.6706,
+      "learning_rate": 1.981760154028731e-05,
+      "loss": 0.2236,
+      "step": 6706
+    },
+    {
+      "epoch": 0.6708,
+      "learning_rate": 1.9818926461118247e-05,
+      "loss": 0.7703,
+      "step": 6708
+    },
+    {
+      "epoch": 0.671,
+      "learning_rate": 1.982024659632372e-05,
+      "loss": 0.4585,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6712,
+      "learning_rate": 1.9821561945260292e-05,
+      "loss": 0.8112,
+      "step": 6712
+    },
+    {
+      "epoch": 0.6714,
+      "learning_rate": 1.9822872507286887e-05,
+      "loss": 0.5332,
+      "step": 6714
+    },
+    {
+      "epoch": 0.6716,
+      "learning_rate": 1.9824178281764753e-05,
+      "loss": 0.7011,
+      "step": 6716
+    },
+    {
+      "epoch": 0.6718,
+      "learning_rate": 1.9825479268057472e-05,
+      "loss": 1.2638,
+      "step": 6718
+    },
+    {
+      "epoch": 0.672,
+      "learning_rate": 1.9826775465530947e-05,
+      "loss": 0.5595,
+      "step": 6720
+    },
+    {
+      "epoch": 0.6722,
+      "learning_rate": 1.9828066873553445e-05,
+      "loss": 1.2906,
+      "step": 6722
+    },
+    {
+      "epoch": 0.6724,
+      "learning_rate": 1.9829353491495545e-05,
+      "loss": 0.4404,
+      "step": 6724
+    },
+    {
+      "epoch": 0.6726,
+      "learning_rate": 1.9830635318730155e-05,
+      "loss": 1.1076,
+      "step": 6726
+    },
+    {
+      "epoch": 0.6728,
+      "learning_rate": 1.9831912354632537e-05,
+      "loss": 0.3769,
+      "step": 6728
+    },
+    {
+      "epoch": 0.673,
+      "learning_rate": 1.983318459858028e-05,
+      "loss": 0.1809,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6732,
+      "learning_rate": 1.98344520499533e-05,
+      "loss": 0.7675,
+      "step": 6732
+    },
+    {
+      "epoch": 0.6734,
+      "learning_rate": 1.9835714708133858e-05,
+      "loss": 0.5705,
+      "step": 6734
+    },
+    {
+      "epoch": 0.6736,
+      "learning_rate": 1.9836972572506557e-05,
+      "loss": 0.8599,
+      "step": 6736
+    },
+    {
+      "epoch": 0.6738,
+      "learning_rate": 1.983822564245833e-05,
+      "loss": 0.7648,
+      "step": 6738
+    },
+    {
+      "epoch": 0.674,
+      "learning_rate": 1.9839473917378432e-05,
+      "loss": 0.5204,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6742,
+      "learning_rate": 1.9840717396658483e-05,
+      "loss": 0.0925,
+      "step": 6742
+    },
+    {
+      "epoch": 0.6744,
+      "learning_rate": 1.984195607969242e-05,
+      "loss": 0.8268,
+      "step": 6744
+    },
+    {
+      "epoch": 0.6746,
+      "learning_rate": 1.9843189965876525e-05,
+      "loss": 0.4636,
+      "step": 6746
+    },
+    {
+      "epoch": 0.6748,
+      "learning_rate": 1.9844419054609418e-05,
+      "loss": 0.4959,
+      "step": 6748
+    },
+    {
+      "epoch": 0.675,
+      "learning_rate": 1.9845643345292055e-05,
+      "loss": 1.0395,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6752,
+      "learning_rate": 1.9846862837327733e-05,
+      "loss": 0.417,
+      "step": 6752
+    },
+    {
+      "epoch": 0.6754,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.5235,
+      "step": 6754
+    },
+    {
+      "epoch": 0.6756,
+      "learning_rate": 1.984928742308308e-05,
+      "loss": 0.5351,
+      "step": 6756
+    },
+    {
+      "epoch": 0.6758,
+      "learning_rate": 1.9850492515621038e-05,
+      "loss": 0.9501,
+      "step": 6758
+    },
+    {
+      "epoch": 0.676,
+      "learning_rate": 1.985169280714861e-05,
+      "loss": 0.6793,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6762,
+      "learning_rate": 1.9852888297080785e-05,
+      "loss": 0.2138,
+      "step": 6762
+    },
+    {
+      "epoch": 0.6764,
+      "learning_rate": 1.9854078984834904e-05,
+      "loss": 0.4475,
+      "step": 6764
+    },
+    {
+      "epoch": 0.6766,
+      "learning_rate": 1.985526486983063e-05,
+      "loss": 0.8696,
+      "step": 6766
+    },
+    {
+      "epoch": 0.6768,
+      "learning_rate": 1.985644595148998e-05,
+      "loss": 0.6846,
+      "step": 6768
+    },
+    {
+      "epoch": 0.677,
+      "learning_rate": 1.9857622229237315e-05,
+      "loss": 1.0464,
+      "step": 6770
+    },
+    {
+      "epoch": 0.6772,
+      "learning_rate": 1.9858793702499322e-05,
+      "loss": 0.6252,
+      "step": 6772
+    },
+    {
+      "epoch": 0.6774,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 0.7551,
+      "step": 6774
+    },
+    {
+      "epoch": 0.6776,
+      "learning_rate": 1.9861122233285873e-05,
+      "loss": 0.8845,
+      "step": 6776
+    },
+    {
+      "epoch": 0.6778,
+      "learning_rate": 1.986227928967551e-05,
+      "loss": 0.8624,
+      "step": 6778
+    },
+    {
+      "epoch": 0.678,
+      "learning_rate": 1.986343153931003e-05,
+      "loss": 0.9194,
+      "step": 6780
+    },
+    {
+      "epoch": 0.6782,
+      "learning_rate": 1.9864578981627844e-05,
+      "loss": 1.5121,
+      "step": 6782
+    },
+    {
+      "epoch": 0.6784,
+      "learning_rate": 1.9865721616069695e-05,
+      "loss": 0.7687,
+      "step": 6784
+    },
+    {
+      "epoch": 0.6786,
+      "learning_rate": 1.986685944207868e-05,
+      "loss": 0.977,
+      "step": 6786
+    },
+    {
+      "epoch": 0.6788,
+      "learning_rate": 1.986799245910024e-05,
+      "loss": 0.661,
+      "step": 6788
+    },
+    {
+      "epoch": 0.679,
+      "learning_rate": 1.9869120666582153e-05,
+      "loss": 0.7318,
+      "step": 6790
+    },
+    {
+      "epoch": 0.6792,
+      "learning_rate": 1.987024406397454e-05,
+      "loss": 0.9194,
+      "step": 6792
+    },
+    {
+      "epoch": 0.6794,
+      "learning_rate": 1.9871362650729877e-05,
+      "loss": 0.3918,
+      "step": 6794
+    },
+    {
+      "epoch": 0.6796,
+      "learning_rate": 1.9872476426302983e-05,
+      "loss": 0.6278,
+      "step": 6796
+    },
+    {
+      "epoch": 0.6798,
+      "learning_rate": 1.9873585390151003e-05,
+      "loss": 0.9582,
+      "step": 6798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.987468954173345e-05,
+      "loss": 0.8406,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6802,
+      "learning_rate": 1.9875788880512183e-05,
+      "loss": 0.1217,
+      "step": 6802
+    },
+    {
+      "epoch": 0.6804,
+      "learning_rate": 1.9876883405951378e-05,
+      "loss": 0.114,
+      "step": 6804
+    },
+    {
+      "epoch": 0.6806,
+      "learning_rate": 1.987797311751759e-05,
+      "loss": 0.6275,
+      "step": 6806
+    },
+    {
+      "epoch": 0.6808,
+      "learning_rate": 1.9879058014679704e-05,
+      "loss": 1.016,
+      "step": 6808
+    },
+    {
+      "epoch": 0.681,
+      "learning_rate": 1.9880138096908955e-05,
+      "loss": 0.7902,
+      "step": 6810
+    },
+    {
+      "epoch": 0.6812,
+      "learning_rate": 1.9881213363678916e-05,
+      "loss": 0.2152,
+      "step": 6812
+    },
+    {
+      "epoch": 0.6814,
+      "learning_rate": 1.9882283814465528e-05,
+      "loss": 0.5265,
+      "step": 6814
+    },
+    {
+      "epoch": 0.6816,
+      "learning_rate": 1.988334944874706e-05,
+      "loss": 0.7802,
+      "step": 6816
+    },
+    {
+      "epoch": 0.6818,
+      "learning_rate": 1.9884410266004134e-05,
+      "loss": 0.9112,
+      "step": 6818
+    },
+    {
+      "epoch": 0.682,
+      "learning_rate": 1.988546626571972e-05,
+      "loss": 0.5198,
+      "step": 6820
+    },
+    {
+      "epoch": 0.6822,
+      "learning_rate": 1.988651744737914e-05,
+      "loss": 0.2037,
+      "step": 6822
+    },
+    {
+      "epoch": 0.6824,
+      "learning_rate": 1.988756381047006e-05,
+      "loss": 0.3987,
+      "step": 6824
+    },
+    {
+      "epoch": 0.6826,
+      "learning_rate": 1.9888605354482494e-05,
+      "loss": 0.9532,
+      "step": 6826
+    },
+    {
+      "epoch": 0.6828,
+      "learning_rate": 1.9889642078908805e-05,
+      "loss": 0.3561,
+      "step": 6828
+    },
+    {
+      "epoch": 0.683,
+      "learning_rate": 1.9890673983243704e-05,
+      "loss": 0.4685,
+      "step": 6830
+    },
+    {
+      "epoch": 0.6832,
+      "learning_rate": 1.9891701066984264e-05,
+      "loss": 0.8144,
+      "step": 6832
+    },
+    {
+      "epoch": 0.6834,
+      "learning_rate": 1.9892723329629885e-05,
+      "loss": 0.9034,
+      "step": 6834
+    },
+    {
+      "epoch": 0.6836,
+      "learning_rate": 1.9893740770682334e-05,
+      "loss": 0.3918,
+      "step": 6836
+    },
+    {
+      "epoch": 0.6838,
+      "learning_rate": 1.9894753389645723e-05,
+      "loss": 0.6875,
+      "step": 6838
+    },
+    {
+      "epoch": 0.684,
+      "learning_rate": 1.989576118602651e-05,
+      "loss": 0.7897,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6842,
+      "learning_rate": 1.989676415933351e-05,
+      "loss": 1.0365,
+      "step": 6842
+    },
+    {
+      "epoch": 0.6844,
+      "learning_rate": 1.989776230907789e-05,
+      "loss": 0.1211,
+      "step": 6844
+    },
+    {
+      "epoch": 0.6846,
+      "learning_rate": 1.9898755634773155e-05,
+      "loss": 0.6346,
+      "step": 6846
+    },
+    {
+      "epoch": 0.6848,
+      "learning_rate": 1.989974413593518e-05,
+      "loss": 0.1731,
+      "step": 6848
+    },
+    {
+      "epoch": 0.685,
+      "learning_rate": 1.9900727812082174e-05,
+      "loss": 0.8267,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6852,
+      "learning_rate": 1.990170666273471e-05,
+      "loss": 0.7511,
+      "step": 6852
+    },
+    {
+      "epoch": 0.6854,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 0.3545,
+      "step": 6854
+    },
+    {
+      "epoch": 0.6856,
+      "learning_rate": 1.990364988565043e-05,
+      "loss": 0.3493,
+      "step": 6856
+    },
+    {
+      "epoch": 0.6858,
+      "learning_rate": 1.9904614256966514e-05,
+      "loss": 0.5732,
+      "step": 6858
+    },
+    {
+      "epoch": 0.686,
+      "learning_rate": 1.990557380089393e-05,
+      "loss": 0.5849,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6862,
+      "learning_rate": 1.9906528516965014e-05,
+      "loss": 0.965,
+      "step": 6862
+    },
+    {
+      "epoch": 0.6864,
+      "learning_rate": 1.9907478404714435e-05,
+      "loss": 0.7749,
+      "step": 6864
+    },
+    {
+      "epoch": 0.6866,
+      "learning_rate": 1.9908423463679246e-05,
+      "loss": 0.9902,
+      "step": 6866
+    },
+    {
+      "epoch": 0.6868,
+      "learning_rate": 1.990936369339883e-05,
+      "loss": 0.5605,
+      "step": 6868
+    },
+    {
+      "epoch": 0.687,
+      "learning_rate": 1.9910299093414926e-05,
+      "loss": 0.2175,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6872,
+      "learning_rate": 1.991122966327164e-05,
+      "loss": 0.2762,
+      "step": 6872
+    },
+    {
+      "epoch": 0.6874,
+      "learning_rate": 1.991215540251542e-05,
+      "loss": 1.1122,
+      "step": 6874
+    },
+    {
+      "epoch": 0.6876,
+      "learning_rate": 1.9913076310695068e-05,
+      "loss": 0.013,
+      "step": 6876
+    },
+    {
+      "epoch": 0.6878,
+      "learning_rate": 1.9913992387361744e-05,
+      "loss": 0.6566,
+      "step": 6878
+    },
+    {
+      "epoch": 0.688,
+      "learning_rate": 1.9914903632068975e-05,
+      "loss": 0.8216,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6882,
+      "learning_rate": 1.9915810044372618e-05,
+      "loss": 0.9851,
+      "step": 6882
+    },
+    {
+      "epoch": 0.6884,
+      "learning_rate": 1.9916711623830904e-05,
+      "loss": 0.702,
+      "step": 6884
+    },
+    {
+      "epoch": 0.6886,
+      "learning_rate": 1.9917608370004414e-05,
+      "loss": 1.0064,
+      "step": 6886
+    },
+    {
+      "epoch": 0.6888,
+      "learning_rate": 1.991850028245609e-05,
+      "loss": 0.1018,
+      "step": 6888
+    },
+    {
+      "epoch": 0.689,
+      "learning_rate": 1.9919387360751216e-05,
+      "loss": 0.2712,
+      "step": 6890
+    },
+    {
+      "epoch": 0.6892,
+      "learning_rate": 1.992026960445744e-05,
+      "loss": 0.7227,
+      "step": 6892
+    },
+    {
+      "epoch": 0.6894,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 1.2433,
+      "step": 6894
+    },
+    {
+      "epoch": 0.6896,
+      "learning_rate": 1.9922019586385587e-05,
+      "loss": 0.6799,
+      "step": 6896
+    },
+    {
+      "epoch": 0.6898,
+      "learning_rate": 1.992288732375458e-05,
+      "loss": 0.6123,
+      "step": 6898
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9923750224828833e-05,
+      "loss": 0.7771,
+      "step": 6900
+    },
+    {
+      "epoch": 0.6902,
+      "learning_rate": 1.9924608289187786e-05,
+      "loss": 0.1411,
+      "step": 6902
+    },
+    {
+      "epoch": 0.6904,
+      "learning_rate": 1.9925461516413224e-05,
+      "loss": 0.5962,
+      "step": 6904
+    },
+    {
+      "epoch": 0.6906,
+      "learning_rate": 1.992630990608929e-05,
+      "loss": 0.2455,
+      "step": 6906
+    },
+    {
+      "epoch": 0.6908,
+      "learning_rate": 1.99271534578025e-05,
+      "loss": 0.4825,
+      "step": 6908
+    },
+    {
+      "epoch": 0.691,
+      "learning_rate": 1.9927992171141707e-05,
+      "loss": 1.4423,
+      "step": 6910
+    },
+    {
+      "epoch": 0.6912,
+      "learning_rate": 1.9928826045698138e-05,
+      "loss": 1.1349,
+      "step": 6912
+    },
+    {
+      "epoch": 0.6914,
+      "learning_rate": 1.992965508106537e-05,
+      "loss": 0.9421,
+      "step": 6914
+    },
+    {
+      "epoch": 0.6916,
+      "learning_rate": 1.9930479276839347e-05,
+      "loss": 0.1142,
+      "step": 6916
+    },
+    {
+      "epoch": 0.6918,
+      "learning_rate": 1.9931298632618355e-05,
+      "loss": 0.3521,
+      "step": 6918
+    },
+    {
+      "epoch": 0.692,
+      "learning_rate": 1.9932113148003057e-05,
+      "loss": 0.5372,
+      "step": 6920
+    },
+    {
+      "epoch": 0.6922,
+      "learning_rate": 1.993292282259647e-05,
+      "loss": 0.763,
+      "step": 6922
+    },
+    {
+      "epoch": 0.6924,
+      "learning_rate": 1.9933727656003964e-05,
+      "loss": 0.6847,
+      "step": 6924
+    },
+    {
+      "epoch": 0.6926,
+      "learning_rate": 1.9934527647833276e-05,
+      "loss": 0.634,
+      "step": 6926
+    },
+    {
+      "epoch": 0.6928,
+      "learning_rate": 1.99353227976945e-05,
+      "loss": 0.5538,
+      "step": 6928
+    },
+    {
+      "epoch": 0.693,
+      "learning_rate": 1.9936113105200085e-05,
+      "loss": 0.7917,
+      "step": 6930
+    },
+    {
+      "epoch": 0.6932,
+      "learning_rate": 1.993689856996485e-05,
+      "loss": 0.5613,
+      "step": 6932
+    },
+    {
+      "epoch": 0.6934,
+      "learning_rate": 1.9937679191605964e-05,
+      "loss": 0.8852,
+      "step": 6934
+    },
+    {
+      "epoch": 0.6936,
+      "learning_rate": 1.9938454969742966e-05,
+      "loss": 0.3456,
+      "step": 6936
+    },
+    {
+      "epoch": 0.6938,
+      "learning_rate": 1.9939225903997748e-05,
+      "loss": 1.6097,
+      "step": 6938
+    },
+    {
+      "epoch": 0.694,
+      "learning_rate": 1.993999199399457e-05,
+      "loss": 0.9667,
+      "step": 6940
+    },
+    {
+      "epoch": 0.6942,
+      "learning_rate": 1.9940753239360047e-05,
+      "loss": 0.1725,
+      "step": 6942
+    },
+    {
+      "epoch": 0.6944,
+      "learning_rate": 1.9941509639723155e-05,
+      "loss": 0.3501,
+      "step": 6944
+    },
+    {
+      "epoch": 0.6946,
+      "learning_rate": 1.9942261194715236e-05,
+      "loss": 0.223,
+      "step": 6946
+    },
+    {
+      "epoch": 0.6948,
+      "learning_rate": 1.994300790396999e-05,
+      "loss": 1.0827,
+      "step": 6948
+    },
+    {
+      "epoch": 0.695,
+      "learning_rate": 1.994374976712348e-05,
+      "loss": 0.9194,
+      "step": 6950
+    },
+    {
+      "epoch": 0.6952,
+      "learning_rate": 1.9944486783814132e-05,
+      "loss": 0.3464,
+      "step": 6952
+    },
+    {
+      "epoch": 0.6954,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 0.6408,
+      "step": 6954
+    },
+    {
+      "epoch": 0.6956,
+      "learning_rate": 1.9945946276372435e-05,
+      "loss": 0.5527,
+      "step": 6956
+    },
+    {
+      "epoch": 0.6958,
+      "learning_rate": 1.994666875152874e-05,
+      "loss": 0.7097,
+      "step": 6958
+    },
+    {
+      "epoch": 0.696,
+      "learning_rate": 1.9947386378799534e-05,
+      "loss": 0.5205,
+      "step": 6960
+    },
+    {
+      "epoch": 0.6962,
+      "learning_rate": 1.994809915783505e-05,
+      "loss": 1.137,
+      "step": 6962
+    },
+    {
+      "epoch": 0.6964,
+      "learning_rate": 1.9948807088287884e-05,
+      "loss": 0.8416,
+      "step": 6964
+    },
+    {
+      "epoch": 0.6966,
+      "learning_rate": 1.9949510169813003e-05,
+      "loss": 0.4677,
+      "step": 6966
+    },
+    {
+      "epoch": 0.6968,
+      "learning_rate": 1.9950208402067735e-05,
+      "loss": 0.5142,
+      "step": 6968
+    },
+    {
+      "epoch": 0.697,
+      "learning_rate": 1.9950901784711768e-05,
+      "loss": 1.049,
+      "step": 6970
+    },
+    {
+      "epoch": 0.6972,
+      "learning_rate": 1.9951590317407152e-05,
+      "loss": 0.3917,
+      "step": 6972
+    },
+    {
+      "epoch": 0.6974,
+      "learning_rate": 1.9952273999818312e-05,
+      "loss": 0.5157,
+      "step": 6974
+    },
+    {
+      "epoch": 0.6976,
+      "learning_rate": 1.9952952831612027e-05,
+      "loss": 0.6691,
+      "step": 6976
+    },
+    {
+      "epoch": 0.6978,
+      "learning_rate": 1.995362681245744e-05,
+      "loss": 1.0186,
+      "step": 6978
+    },
+    {
+      "epoch": 0.698,
+      "learning_rate": 1.9954295942026065e-05,
+      "loss": 0.4972,
+      "step": 6980
+    },
+    {
+      "epoch": 0.6982,
+      "learning_rate": 1.995496021999177e-05,
+      "loss": 0.857,
+      "step": 6982
+    },
+    {
+      "epoch": 0.6984,
+      "learning_rate": 1.99556196460308e-05,
+      "loss": 0.7218,
+      "step": 6984
+    },
+    {
+      "epoch": 0.6986,
+      "learning_rate": 1.995627421982176e-05,
+      "loss": 0.4133,
+      "step": 6986
+    },
+    {
+      "epoch": 0.6988,
+      "learning_rate": 1.9956923941045613e-05,
+      "loss": 0.8857,
+      "step": 6988
+    },
+    {
+      "epoch": 0.699,
+      "learning_rate": 1.9957568809385693e-05,
+      "loss": 0.3304,
+      "step": 6990
+    },
+    {
+      "epoch": 0.6992,
+      "learning_rate": 1.9958208824527702e-05,
+      "loss": 0.9353,
+      "step": 6992
+    },
+    {
+      "epoch": 0.6994,
+      "learning_rate": 1.9958843986159705e-05,
+      "loss": 0.3261,
+      "step": 6994
+    },
+    {
+      "epoch": 0.6996,
+      "learning_rate": 1.995947429397213e-05,
+      "loss": 0.1307,
+      "step": 6996
+    },
+    {
+      "epoch": 0.6998,
+      "learning_rate": 1.9960099747657774e-05,
+      "loss": 0.5843,
+      "step": 6998
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9960720346911798e-05,
+      "loss": 0.5831,
+      "step": 7000
+    },
+    {
+      "epoch": 0.7002,
+      "learning_rate": 1.9961336091431725e-05,
+      "loss": 0.0771,
+      "step": 7002
+    },
+    {
+      "epoch": 0.7004,
+      "learning_rate": 1.9961946980917457e-05,
+      "loss": 0.424,
+      "step": 7004
+    },
+    {
+      "epoch": 0.7006,
+      "learning_rate": 1.996255301507125e-05,
+      "loss": 0.9768,
+      "step": 7006
+    },
+    {
+      "epoch": 0.7008,
+      "learning_rate": 1.9963154193597725e-05,
+      "loss": 0.1042,
+      "step": 7008
+    },
+    {
+      "epoch": 0.701,
+      "learning_rate": 1.9963750516203884e-05,
+      "loss": 0.0978,
+      "step": 7010
+    },
+    {
+      "epoch": 0.7012,
+      "learning_rate": 1.996434198259908e-05,
+      "loss": 0.9171,
+      "step": 7012
+    },
+    {
+      "epoch": 0.7014,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 0.63,
+      "step": 7014
+    },
+    {
+      "epoch": 0.7016,
+      "learning_rate": 1.9965510345605866e-05,
+      "loss": 0.7486,
+      "step": 7016
+    },
+    {
+      "epoch": 0.7018,
+      "learning_rate": 1.996608724164801e-05,
+      "loss": 1.2362,
+      "step": 7018
+    },
+    {
+      "epoch": 0.702,
+      "learning_rate": 1.99666592803403e-05,
+      "loss": 0.5678,
+      "step": 7020
+    },
+    {
+      "epoch": 0.7022,
+      "learning_rate": 1.9967226461403934e-05,
+      "loss": 0.4287,
+      "step": 7022
+    },
+    {
+      "epoch": 0.7024,
+      "learning_rate": 1.9967788784562474e-05,
+      "loss": 0.875,
+      "step": 7024
+    },
+    {
+      "epoch": 0.7026,
+      "learning_rate": 1.9968346249541848e-05,
+      "loss": 0.0529,
+      "step": 7026
+    },
+    {
+      "epoch": 0.7028,
+      "learning_rate": 1.9968898856070358e-05,
+      "loss": 0.626,
+      "step": 7028
+    },
+    {
+      "epoch": 0.703,
+      "learning_rate": 1.996944660387867e-05,
+      "loss": 0.449,
+      "step": 7030
+    },
+    {
+      "epoch": 0.7032,
+      "learning_rate": 1.996998949269982e-05,
+      "loss": 0.9187,
+      "step": 7032
+    },
+    {
+      "epoch": 0.7034,
+      "learning_rate": 1.9970527522269204e-05,
+      "loss": 1.0984,
+      "step": 7034
+    },
+    {
+      "epoch": 0.7036,
+      "learning_rate": 1.99710606923246e-05,
+      "loss": 0.5674,
+      "step": 7036
+    },
+    {
+      "epoch": 0.7038,
+      "learning_rate": 1.997158900260614e-05,
+      "loss": 1.1272,
+      "step": 7038
+    },
+    {
+      "epoch": 0.704,
+      "learning_rate": 1.997211245285634e-05,
+      "loss": 0.7716,
+      "step": 7040
+    },
+    {
+      "epoch": 0.7042,
+      "learning_rate": 1.997263104282007e-05,
+      "loss": 0.8213,
+      "step": 7042
+    },
+    {
+      "epoch": 0.7044,
+      "learning_rate": 1.997314477224458e-05,
+      "loss": 0.7241,
+      "step": 7044
+    },
+    {
+      "epoch": 0.7046,
+      "learning_rate": 1.9973653640879486e-05,
+      "loss": 0.9405,
+      "step": 7046
+    },
+    {
+      "epoch": 0.7048,
+      "learning_rate": 1.9974157648476768e-05,
+      "loss": 0.868,
+      "step": 7048
+    },
+    {
+      "epoch": 0.705,
+      "learning_rate": 1.9974656794790777e-05,
+      "loss": 0.6496,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7052,
+      "learning_rate": 1.9975151079578238e-05,
+      "loss": 0.6579,
+      "step": 7052
+    },
+    {
+      "epoch": 0.7054,
+      "learning_rate": 1.9975640502598246e-05,
+      "loss": 0.9862,
+      "step": 7054
+    },
+    {
+      "epoch": 0.7056,
+      "learning_rate": 1.9976125063612254e-05,
+      "loss": 0.6279,
+      "step": 7056
+    },
+    {
+      "epoch": 0.7058,
+      "learning_rate": 1.99766047623841e-05,
+      "loss": 0.9815,
+      "step": 7058
+    },
+    {
+      "epoch": 0.706,
+      "learning_rate": 1.9977079598679978e-05,
+      "loss": 0.1822,
+      "step": 7060
+    },
+    {
+      "epoch": 0.7062,
+      "learning_rate": 1.997754957226847e-05,
+      "loss": 0.1134,
+      "step": 7062
+    },
+    {
+      "epoch": 0.7064,
+      "learning_rate": 1.9978014682920503e-05,
+      "loss": 0.6503,
+      "step": 7064
+    },
+    {
+      "epoch": 0.7066,
+      "learning_rate": 1.9978474930409396e-05,
+      "loss": 0.914,
+      "step": 7066
+    },
+    {
+      "epoch": 0.7068,
+      "learning_rate": 1.9978930314510826e-05,
+      "loss": 0.0561,
+      "step": 7068
+    },
+    {
+      "epoch": 0.707,
+      "learning_rate": 1.9979380835002846e-05,
+      "loss": 0.4571,
+      "step": 7070
+    },
+    {
+      "epoch": 0.7072,
+      "learning_rate": 1.997982649166588e-05,
+      "loss": 1.3046,
+      "step": 7072
+    },
+    {
+      "epoch": 0.7074,
+      "learning_rate": 1.9980267284282718e-05,
+      "loss": 0.7619,
+      "step": 7074
+    },
+    {
+      "epoch": 0.7076,
+      "learning_rate": 1.9980703212638522e-05,
+      "loss": 0.8201,
+      "step": 7076
+    },
+    {
+      "epoch": 0.7078,
+      "learning_rate": 1.9981134276520828e-05,
+      "loss": 0.7744,
+      "step": 7078
+    },
+    {
+      "epoch": 0.708,
+      "learning_rate": 1.998156047571954e-05,
+      "loss": 1.2521,
+      "step": 7080
+    },
+    {
+      "epoch": 0.7082,
+      "learning_rate": 1.9981981810026932e-05,
+      "loss": 0.4835,
+      "step": 7082
+    },
+    {
+      "epoch": 0.7084,
+      "learning_rate": 1.9982398279237653e-05,
+      "loss": 1.1536,
+      "step": 7084
+    },
+    {
+      "epoch": 0.7086,
+      "learning_rate": 1.998280988314872e-05,
+      "loss": 0.5196,
+      "step": 7086
+    },
+    {
+      "epoch": 0.7088,
+      "learning_rate": 1.9983216621559525e-05,
+      "loss": 0.3287,
+      "step": 7088
+    },
+    {
+      "epoch": 0.709,
+      "learning_rate": 1.9983618494271825e-05,
+      "loss": 0.6897,
+      "step": 7090
+    },
+    {
+      "epoch": 0.7092,
+      "learning_rate": 1.998401550108975e-05,
+      "loss": 0.6928,
+      "step": 7092
+    },
+    {
+      "epoch": 0.7094,
+      "learning_rate": 1.998440764181981e-05,
+      "loss": 0.7548,
+      "step": 7094
+    },
+    {
+      "epoch": 0.7096,
+      "learning_rate": 1.9984794916270876e-05,
+      "loss": 1.1239,
+      "step": 7096
+    },
+    {
+      "epoch": 0.7098,
+      "learning_rate": 1.99851773242542e-05,
+      "loss": 1.1257,
+      "step": 7098
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9985554865583394e-05,
+      "loss": 0.0095,
+      "step": 7100
+    },
+    {
+      "epoch": 0.7102,
+      "learning_rate": 1.9985927540074453e-05,
+      "loss": 0.6562,
+      "step": 7102
+    },
+    {
+      "epoch": 0.7104,
+      "learning_rate": 1.9986295347545738e-05,
+      "loss": 0.3062,
+      "step": 7104
+    },
+    {
+      "epoch": 0.7106,
+      "learning_rate": 1.9986658287817992e-05,
+      "loss": 0.9622,
+      "step": 7106
+    },
+    {
+      "epoch": 0.7108,
+      "learning_rate": 1.9987016360714307e-05,
+      "loss": 0.4788,
+      "step": 7108
+    },
+    {
+      "epoch": 0.711,
+      "learning_rate": 1.998736956606018e-05,
+      "loss": 1.2692,
+      "step": 7110
+    },
+    {
+      "epoch": 0.7112,
+      "learning_rate": 1.9987717903683444e-05,
+      "loss": 0.1109,
+      "step": 7112
+    },
+    {
+      "epoch": 0.7114,
+      "learning_rate": 1.9988061373414342e-05,
+      "loss": 0.9367,
+      "step": 7114
+    },
+    {
+      "epoch": 0.7116,
+      "learning_rate": 1.998839997508546e-05,
+      "loss": 0.806,
+      "step": 7116
+    },
+    {
+      "epoch": 0.7118,
+      "learning_rate": 1.9988733708531772e-05,
+      "loss": 0.3678,
+      "step": 7118
+    },
+    {
+      "epoch": 0.712,
+      "learning_rate": 1.9989062573590618e-05,
+      "loss": 0.7623,
+      "step": 7120
+    },
+    {
+      "epoch": 0.7122,
+      "learning_rate": 1.9989386570101712e-05,
+      "loss": 0.1465,
+      "step": 7122
+    },
+    {
+      "epoch": 0.7124,
+      "learning_rate": 1.9989705697907145e-05,
+      "loss": 0.6628,
+      "step": 7124
+    },
+    {
+      "epoch": 0.7126,
+      "learning_rate": 1.9990019956851384e-05,
+      "loss": 0.478,
+      "step": 7126
+    },
+    {
+      "epoch": 0.7128,
+      "learning_rate": 1.999032934678125e-05,
+      "loss": 0.4507,
+      "step": 7128
+    },
+    {
+      "epoch": 0.713,
+      "learning_rate": 1.9990633867545956e-05,
+      "loss": 0.8876,
+      "step": 7130
+    },
+    {
+      "epoch": 0.7132,
+      "learning_rate": 1.9990933518997086e-05,
+      "loss": 0.6577,
+      "step": 7132
+    },
+    {
+      "epoch": 0.7134,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 0.3324,
+      "step": 7134
+    },
+    {
+      "epoch": 0.7136,
+      "learning_rate": 1.9991518213376787e-05,
+      "loss": 1.7486,
+      "step": 7136
+    },
+    {
+      "epoch": 0.7138,
+      "learning_rate": 1.9991803256020393e-05,
+      "loss": 0.6257,
+      "step": 7138
+    },
+    {
+      "epoch": 0.714,
+      "learning_rate": 1.999208342878047e-05,
+      "loss": 0.8487,
+      "step": 7140
+    },
+    {
+      "epoch": 0.7142,
+      "learning_rate": 1.999235873152047e-05,
+      "loss": 0.2073,
+      "step": 7142
+    },
+    {
+      "epoch": 0.7144,
+      "learning_rate": 1.999262916410621e-05,
+      "loss": 0.5018,
+      "step": 7144
+    },
+    {
+      "epoch": 0.7146,
+      "learning_rate": 1.9992894726405894e-05,
+      "loss": 0.4427,
+      "step": 7146
+    },
+    {
+      "epoch": 0.7148,
+      "learning_rate": 1.999315541829008e-05,
+      "loss": 0.4558,
+      "step": 7148
+    },
+    {
+      "epoch": 0.715,
+      "learning_rate": 1.9993411239631713e-05,
+      "loss": 0.4222,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7152,
+      "learning_rate": 1.999366219030611e-05,
+      "loss": 0.718,
+      "step": 7152
+    },
+    {
+      "epoch": 0.7154,
+      "learning_rate": 1.999390827019096e-05,
+      "loss": 1.009,
+      "step": 7154
+    },
+    {
+      "epoch": 0.7156,
+      "learning_rate": 1.9994149479166324e-05,
+      "loss": 0.4107,
+      "step": 7156
+    },
+    {
+      "epoch": 0.7158,
+      "learning_rate": 1.9994385817114644e-05,
+      "loss": 0.5853,
+      "step": 7158
+    },
+    {
+      "epoch": 0.716,
+      "learning_rate": 1.999461728392073e-05,
+      "loss": 0.3272,
+      "step": 7160
+    },
+    {
+      "epoch": 0.7162,
+      "learning_rate": 1.999484387947177e-05,
+      "loss": 0.853,
+      "step": 7162
+    },
+    {
+      "epoch": 0.7164,
+      "learning_rate": 1.9995065603657317e-05,
+      "loss": 0.7078,
+      "step": 7164
+    },
+    {
+      "epoch": 0.7166,
+      "learning_rate": 1.9995282456369313e-05,
+      "loss": 0.8885,
+      "step": 7166
+    },
+    {
+      "epoch": 0.7168,
+      "learning_rate": 1.9995494437502064e-05,
+      "loss": 0.608,
+      "step": 7168
+    },
+    {
+      "epoch": 0.717,
+      "learning_rate": 1.9995701546952252e-05,
+      "loss": 0.6496,
+      "step": 7170
+    },
+    {
+      "epoch": 0.7172,
+      "learning_rate": 1.9995903784618936e-05,
+      "loss": 0.2054,
+      "step": 7172
+    },
+    {
+      "epoch": 0.7174,
+      "learning_rate": 1.9996101150403543e-05,
+      "loss": 0.3647,
+      "step": 7174
+    },
+    {
+      "epoch": 0.7176,
+      "learning_rate": 1.9996293644209886e-05,
+      "loss": 0.8468,
+      "step": 7176
+    },
+    {
+      "epoch": 0.7178,
+      "learning_rate": 1.9996481265944146e-05,
+      "loss": 0.7716,
+      "step": 7178
+    },
+    {
+      "epoch": 0.718,
+      "learning_rate": 1.999666401551487e-05,
+      "loss": 0.4426,
+      "step": 7180
+    },
+    {
+      "epoch": 0.7182,
+      "learning_rate": 1.9996841892833e-05,
+      "loss": 1.0422,
+      "step": 7182
+    },
+    {
+      "epoch": 0.7184,
+      "learning_rate": 1.9997014897811834e-05,
+      "loss": 0.6698,
+      "step": 7184
+    },
+    {
+      "epoch": 0.7186,
+      "learning_rate": 1.999718303036705e-05,
+      "loss": 0.6885,
+      "step": 7186
+    },
+    {
+      "epoch": 0.7188,
+      "learning_rate": 1.9997346290416703e-05,
+      "loss": 0.6627,
+      "step": 7188
+    },
+    {
+      "epoch": 0.719,
+      "learning_rate": 1.9997504677881224e-05,
+      "loss": 0.5177,
+      "step": 7190
+    },
+    {
+      "epoch": 0.7192,
+      "learning_rate": 1.9997658192683412e-05,
+      "loss": 0.7463,
+      "step": 7192
+    },
+    {
+      "epoch": 0.7194,
+      "learning_rate": 1.9997806834748455e-05,
+      "loss": 1.248,
+      "step": 7194
+    },
+    {
+      "epoch": 0.7196,
+      "learning_rate": 1.99979506040039e-05,
+      "loss": 0.4744,
+      "step": 7196
+    },
+    {
+      "epoch": 0.7198,
+      "learning_rate": 1.999808950037968e-05,
+      "loss": 0.9891,
+      "step": 7198
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9998223523808092e-05,
+      "loss": 0.5114,
+      "step": 7200
+    },
+    {
+      "epoch": 0.7202,
+      "learning_rate": 1.9998352674223816e-05,
+      "loss": 1.6737,
+      "step": 7202
+    },
+    {
+      "epoch": 0.7204,
+      "learning_rate": 1.9998476951563914e-05,
+      "loss": 0.7339,
+      "step": 7204
+    },
+    {
+      "epoch": 0.7206,
+      "learning_rate": 1.9998596355767805e-05,
+      "loss": 0.0918,
+      "step": 7206
+    },
+    {
+      "epoch": 0.7208,
+      "learning_rate": 1.9998710886777298e-05,
+      "loss": 0.7508,
+      "step": 7208
+    },
+    {
+      "epoch": 0.721,
+      "learning_rate": 1.999882054453657e-05,
+      "loss": 0.9352,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7212,
+      "learning_rate": 1.9998925328992175e-05,
+      "loss": 0.6728,
+      "step": 7212
+    },
+    {
+      "epoch": 0.7214,
+      "learning_rate": 1.9999025240093045e-05,
+      "loss": 1.0143,
+      "step": 7214
+    },
+    {
+      "epoch": 0.7216,
+      "learning_rate": 1.9999120277790477e-05,
+      "loss": 0.4119,
+      "step": 7216
+    },
+    {
+      "epoch": 0.7218,
+      "learning_rate": 1.9999210442038164e-05,
+      "loss": 0.5239,
+      "step": 7218
+    },
+    {
+      "epoch": 0.722,
+      "learning_rate": 1.9999295732792146e-05,
+      "loss": 1.1772,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7222,
+      "learning_rate": 1.9999376150010868e-05,
+      "loss": 0.3645,
+      "step": 7222
+    },
+    {
+      "epoch": 0.7224,
+      "learning_rate": 1.9999451693655125e-05,
+      "loss": 0.5275,
+      "step": 7224
+    },
+    {
+      "epoch": 0.7226,
+      "learning_rate": 1.99995223636881e-05,
+      "loss": 0.5509,
+      "step": 7226
+    },
+    {
+      "epoch": 0.7228,
+      "learning_rate": 1.999958816007535e-05,
+      "loss": 0.5985,
+      "step": 7228
+    },
+    {
+      "epoch": 0.723,
+      "learning_rate": 1.9999649082784807e-05,
+      "loss": 0.7189,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7232,
+      "learning_rate": 1.999970513178678e-05,
+      "loss": 0.5055,
+      "step": 7232
+    },
+    {
+      "epoch": 0.7234,
+      "learning_rate": 1.9999756307053947e-05,
+      "loss": 1.181,
+      "step": 7234
+    },
+    {
+      "epoch": 0.7236,
+      "learning_rate": 1.999980260856137e-05,
+      "loss": 1.1309,
+      "step": 7236
+    },
+    {
+      "epoch": 0.7238,
+      "learning_rate": 1.9999844036286483e-05,
+      "loss": 0.5001,
+      "step": 7238
+    },
+    {
+      "epoch": 0.724,
+      "learning_rate": 1.999988059020909e-05,
+      "loss": 1.0979,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7242,
+      "learning_rate": 1.9999912270311376e-05,
+      "loss": 1.1635,
+      "step": 7242
+    },
+    {
+      "epoch": 0.7244,
+      "learning_rate": 1.9999939076577906e-05,
+      "loss": 0.5458,
+      "step": 7244
+    },
+    {
+      "epoch": 0.7246,
+      "learning_rate": 1.9999961008995607e-05,
+      "loss": 1.0416,
+      "step": 7246
+    },
+    {
+      "epoch": 0.7248,
+      "learning_rate": 1.9999978067553796e-05,
+      "loss": 0.6695,
+      "step": 7248
+    },
+    {
+      "epoch": 0.725,
+      "learning_rate": 1.9999990252244153e-05,
+      "loss": 0.1871,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7252,
+      "learning_rate": 1.9999997563060744e-05,
+      "loss": 0.0988,
+      "step": 7252
+    },
+    {
+      "epoch": 0.7254,
+      "learning_rate": 2e-05,
+      "loss": 0.2883,
+      "step": 7254
+    },
+    {
+      "epoch": 0.7256,
+      "learning_rate": 1.9999997563060744e-05,
+      "loss": 0.2656,
+      "step": 7256
+    },
+    {
+      "epoch": 0.7258,
+      "learning_rate": 1.9999990252244153e-05,
+      "loss": 0.9291,
+      "step": 7258
+    },
+    {
+      "epoch": 0.726,
+      "learning_rate": 1.9999978067553796e-05,
+      "loss": 0.9024,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7262,
+      "learning_rate": 1.9999961008995607e-05,
+      "loss": 1.4474,
+      "step": 7262
+    },
+    {
+      "epoch": 0.7264,
+      "learning_rate": 1.9999939076577906e-05,
+      "loss": 0.918,
+      "step": 7264
+    },
+    {
+      "epoch": 0.7266,
+      "learning_rate": 1.9999912270311376e-05,
+      "loss": 0.5325,
+      "step": 7266
+    },
+    {
+      "epoch": 0.7268,
+      "learning_rate": 1.999988059020909e-05,
+      "loss": 0.3783,
+      "step": 7268
+    },
+    {
+      "epoch": 0.727,
+      "learning_rate": 1.9999844036286483e-05,
+      "loss": 0.4475,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7272,
+      "learning_rate": 1.999980260856137e-05,
+      "loss": 0.0561,
+      "step": 7272
+    },
+    {
+      "epoch": 0.7274,
+      "learning_rate": 1.9999756307053947e-05,
+      "loss": 0.7844,
+      "step": 7274
+    },
+    {
+      "epoch": 0.7276,
+      "learning_rate": 1.999970513178678e-05,
+      "loss": 0.5793,
+      "step": 7276
+    },
+    {
+      "epoch": 0.7278,
+      "learning_rate": 1.9999649082784807e-05,
+      "loss": 0.2174,
+      "step": 7278
+    },
+    {
+      "epoch": 0.728,
+      "learning_rate": 1.999958816007535e-05,
+      "loss": 0.8375,
+      "step": 7280
+    },
+    {
+      "epoch": 0.7282,
+      "learning_rate": 1.99995223636881e-05,
+      "loss": 0.1488,
+      "step": 7282
+    },
+    {
+      "epoch": 0.7284,
+      "learning_rate": 1.9999451693655125e-05,
+      "loss": 1.0267,
+      "step": 7284
+    },
+    {
+      "epoch": 0.7286,
+      "learning_rate": 1.9999376150010868e-05,
+      "loss": 0.9295,
+      "step": 7286
+    },
+    {
+      "epoch": 0.7288,
+      "learning_rate": 1.9999295732792146e-05,
+      "loss": 0.9631,
+      "step": 7288
+    },
+    {
+      "epoch": 0.729,
+      "learning_rate": 1.9999210442038164e-05,
+      "loss": 0.836,
+      "step": 7290
+    },
+    {
+      "epoch": 0.7292,
+      "learning_rate": 1.9999120277790477e-05,
+      "loss": 0.5002,
+      "step": 7292
+    },
+    {
+      "epoch": 0.7294,
+      "learning_rate": 1.9999025240093045e-05,
+      "loss": 0.9911,
+      "step": 7294
+    },
+    {
+      "epoch": 0.7296,
+      "learning_rate": 1.9998925328992175e-05,
+      "loss": 0.5347,
+      "step": 7296
+    },
+    {
+      "epoch": 0.7298,
+      "learning_rate": 1.999882054453657e-05,
+      "loss": 0.6656,
+      "step": 7298
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9998710886777298e-05,
+      "loss": 0.089,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7302,
+      "learning_rate": 1.9998596355767805e-05,
+      "loss": 0.484,
+      "step": 7302
+    },
+    {
+      "epoch": 0.7304,
+      "learning_rate": 1.9998476951563914e-05,
+      "loss": 0.8971,
+      "step": 7304
+    },
+    {
+      "epoch": 0.7306,
+      "learning_rate": 1.9998352674223816e-05,
+      "loss": 0.6416,
+      "step": 7306
+    },
+    {
+      "epoch": 0.7308,
+      "learning_rate": 1.9998223523808092e-05,
+      "loss": 0.9128,
+      "step": 7308
+    },
+    {
+      "epoch": 0.731,
+      "learning_rate": 1.999808950037968e-05,
+      "loss": 1.2295,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7312,
+      "learning_rate": 1.99979506040039e-05,
+      "loss": 0.1299,
+      "step": 7312
+    },
+    {
+      "epoch": 0.7314,
+      "learning_rate": 1.9997806834748455e-05,
+      "loss": 0.4139,
+      "step": 7314
+    },
+    {
+      "epoch": 0.7316,
+      "learning_rate": 1.9997658192683416e-05,
+      "loss": 0.3148,
+      "step": 7316
+    },
+    {
+      "epoch": 0.7318,
+      "learning_rate": 1.9997504677881224e-05,
+      "loss": 0.1594,
+      "step": 7318
+    },
+    {
+      "epoch": 0.732,
+      "learning_rate": 1.9997346290416703e-05,
+      "loss": 0.2855,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7322,
+      "learning_rate": 1.999718303036705e-05,
+      "loss": 0.946,
+      "step": 7322
+    },
+    {
+      "epoch": 0.7324,
+      "learning_rate": 1.9997014897811834e-05,
+      "loss": 0.6601,
+      "step": 7324
+    },
+    {
+      "epoch": 0.7326,
+      "learning_rate": 1.9996841892833e-05,
+      "loss": 0.3891,
+      "step": 7326
+    },
+    {
+      "epoch": 0.7328,
+      "learning_rate": 1.9996664015514875e-05,
+      "loss": 0.9702,
+      "step": 7328
+    },
+    {
+      "epoch": 0.733,
+      "learning_rate": 1.9996481265944146e-05,
+      "loss": 0.6478,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7332,
+      "learning_rate": 1.9996293644209886e-05,
+      "loss": 0.9529,
+      "step": 7332
+    },
+    {
+      "epoch": 0.7334,
+      "learning_rate": 1.9996101150403547e-05,
+      "loss": 0.463,
+      "step": 7334
+    },
+    {
+      "epoch": 0.7336,
+      "learning_rate": 1.9995903784618936e-05,
+      "loss": 1.1707,
+      "step": 7336
+    },
+    {
+      "epoch": 0.7338,
+      "learning_rate": 1.9995701546952252e-05,
+      "loss": 0.884,
+      "step": 7338
+    },
+    {
+      "epoch": 0.734,
+      "learning_rate": 1.9995494437502064e-05,
+      "loss": 0.6654,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7342,
+      "learning_rate": 1.9995282456369313e-05,
+      "loss": 0.0054,
+      "step": 7342
+    },
+    {
+      "epoch": 0.7344,
+      "learning_rate": 1.9995065603657317e-05,
+      "loss": 1.03,
+      "step": 7344
+    },
+    {
+      "epoch": 0.7346,
+      "learning_rate": 1.9994843879471766e-05,
+      "loss": 0.1367,
+      "step": 7346
+    },
+    {
+      "epoch": 0.7348,
+      "learning_rate": 1.999461728392073e-05,
+      "loss": 0.1107,
+      "step": 7348
+    },
+    {
+      "epoch": 0.735,
+      "learning_rate": 1.9994385817114644e-05,
+      "loss": 0.9882,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7352,
+      "learning_rate": 1.999414947916632e-05,
+      "loss": 0.2818,
+      "step": 7352
+    },
+    {
+      "epoch": 0.7354,
+      "learning_rate": 1.999390827019096e-05,
+      "loss": 0.8254,
+      "step": 7354
+    },
+    {
+      "epoch": 0.7356,
+      "learning_rate": 1.999366219030611e-05,
+      "loss": 0.6489,
+      "step": 7356
+    },
+    {
+      "epoch": 0.7358,
+      "learning_rate": 1.9993411239631713e-05,
+      "loss": 1.1641,
+      "step": 7358
+    },
+    {
+      "epoch": 0.736,
+      "learning_rate": 1.999315541829008e-05,
+      "loss": 0.4053,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7362,
+      "learning_rate": 1.9992894726405898e-05,
+      "loss": 0.6736,
+      "step": 7362
+    },
+    {
+      "epoch": 0.7364,
+      "learning_rate": 1.999262916410621e-05,
+      "loss": 1.3859,
+      "step": 7364
+    },
+    {
+      "epoch": 0.7366,
+      "learning_rate": 1.999235873152047e-05,
+      "loss": 1.0703,
+      "step": 7366
+    },
+    {
+      "epoch": 0.7368,
+      "learning_rate": 1.999208342878047e-05,
+      "loss": 0.7208,
+      "step": 7368
+    },
+    {
+      "epoch": 0.737,
+      "learning_rate": 1.9991803256020393e-05,
+      "loss": 0.9932,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7372,
+      "learning_rate": 1.9991518213376787e-05,
+      "loss": 0.2869,
+      "step": 7372
+    },
+    {
+      "epoch": 0.7374,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 0.1345,
+      "step": 7374
+    },
+    {
+      "epoch": 0.7376,
+      "learning_rate": 1.9990933518997086e-05,
+      "loss": 0.7576,
+      "step": 7376
+    },
+    {
+      "epoch": 0.7378,
+      "learning_rate": 1.9990633867545956e-05,
+      "loss": 0.3413,
+      "step": 7378
+    },
+    {
+      "epoch": 0.738,
+      "learning_rate": 1.999032934678125e-05,
+      "loss": 0.72,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7382,
+      "learning_rate": 1.9990019956851384e-05,
+      "loss": 0.9196,
+      "step": 7382
+    },
+    {
+      "epoch": 0.7384,
+      "learning_rate": 1.998970569790715e-05,
+      "loss": 0.9724,
+      "step": 7384
+    },
+    {
+      "epoch": 0.7386,
+      "learning_rate": 1.9989386570101716e-05,
+      "loss": 0.7721,
+      "step": 7386
+    },
+    {
+      "epoch": 0.7388,
+      "learning_rate": 1.9989062573590618e-05,
+      "loss": 0.6618,
+      "step": 7388
+    },
+    {
+      "epoch": 0.739,
+      "learning_rate": 1.9988733708531772e-05,
+      "loss": 0.8882,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7392,
+      "learning_rate": 1.998839997508546e-05,
+      "loss": 0.5459,
+      "step": 7392
+    },
+    {
+      "epoch": 0.7394,
+      "learning_rate": 1.9988061373414342e-05,
+      "loss": 0.5751,
+      "step": 7394
+    },
+    {
+      "epoch": 0.7396,
+      "learning_rate": 1.9987717903683447e-05,
+      "loss": 0.9109,
+      "step": 7396
+    },
+    {
+      "epoch": 0.7398,
+      "learning_rate": 1.998736956606018e-05,
+      "loss": 0.2336,
+      "step": 7398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9987016360714307e-05,
+      "loss": 0.6632,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7402,
+      "learning_rate": 1.998665828781799e-05,
+      "loss": 0.7177,
+      "step": 7402
+    },
+    {
+      "epoch": 0.7404,
+      "learning_rate": 1.9986295347545738e-05,
+      "loss": 0.1546,
+      "step": 7404
+    },
+    {
+      "epoch": 0.7406,
+      "learning_rate": 1.9985927540074453e-05,
+      "loss": 0.6584,
+      "step": 7406
+    },
+    {
+      "epoch": 0.7408,
+      "learning_rate": 1.9985554865583397e-05,
+      "loss": 0.9847,
+      "step": 7408
+    },
+    {
+      "epoch": 0.741,
+      "learning_rate": 1.99851773242542e-05,
+      "loss": 1.1196,
+      "step": 7410
+    },
+    {
+      "epoch": 0.7412,
+      "learning_rate": 1.9984794916270876e-05,
+      "loss": 0.8385,
+      "step": 7412
+    },
+    {
+      "epoch": 0.7414,
+      "learning_rate": 1.9984407641819812e-05,
+      "loss": 0.7453,
+      "step": 7414
+    },
+    {
+      "epoch": 0.7416,
+      "learning_rate": 1.998401550108975e-05,
+      "loss": 0.5709,
+      "step": 7416
+    },
+    {
+      "epoch": 0.7418,
+      "learning_rate": 1.9983618494271825e-05,
+      "loss": 0.6397,
+      "step": 7418
+    },
+    {
+      "epoch": 0.742,
+      "learning_rate": 1.998321662155953e-05,
+      "loss": 0.5439,
+      "step": 7420
+    },
+    {
+      "epoch": 0.7422,
+      "learning_rate": 1.998280988314872e-05,
+      "loss": 0.8437,
+      "step": 7422
+    },
+    {
+      "epoch": 0.7424,
+      "learning_rate": 1.9982398279237653e-05,
+      "loss": 1.0988,
+      "step": 7424
+    },
+    {
+      "epoch": 0.7426,
+      "learning_rate": 1.9981981810026932e-05,
+      "loss": 0.5878,
+      "step": 7426
+    },
+    {
+      "epoch": 0.7428,
+      "learning_rate": 1.998156047571954e-05,
+      "loss": 0.8309,
+      "step": 7428
+    },
+    {
+      "epoch": 0.743,
+      "learning_rate": 1.9981134276520828e-05,
+      "loss": 0.7538,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7432,
+      "learning_rate": 1.9980703212638522e-05,
+      "loss": 0.7106,
+      "step": 7432
+    },
+    {
+      "epoch": 0.7434,
+      "learning_rate": 1.9980267284282718e-05,
+      "loss": 0.0512,
+      "step": 7434
+    },
+    {
+      "epoch": 0.7436,
+      "learning_rate": 1.997982649166588e-05,
+      "loss": 0.1271,
+      "step": 7436
+    },
+    {
+      "epoch": 0.7438,
+      "learning_rate": 1.9979380835002846e-05,
+      "loss": 1.0385,
+      "step": 7438
+    },
+    {
+      "epoch": 0.744,
+      "learning_rate": 1.9978930314510826e-05,
+      "loss": 1.2298,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7442,
+      "learning_rate": 1.9978474930409396e-05,
+      "loss": 1.0534,
+      "step": 7442
+    },
+    {
+      "epoch": 0.7444,
+      "learning_rate": 1.9978014682920503e-05,
+      "loss": 0.3628,
+      "step": 7444
+    },
+    {
+      "epoch": 0.7446,
+      "learning_rate": 1.9977549572268467e-05,
+      "loss": 0.3939,
+      "step": 7446
+    },
+    {
+      "epoch": 0.7448,
+      "learning_rate": 1.997707959867998e-05,
+      "loss": 0.6731,
+      "step": 7448
+    },
+    {
+      "epoch": 0.745,
+      "learning_rate": 1.99766047623841e-05,
+      "loss": 0.7762,
+      "step": 7450
+    },
+    {
+      "epoch": 0.7452,
+      "learning_rate": 1.9976125063612254e-05,
+      "loss": 0.5412,
+      "step": 7452
+    },
+    {
+      "epoch": 0.7454,
+      "learning_rate": 1.9975640502598246e-05,
+      "loss": 0.0782,
+      "step": 7454
+    },
+    {
+      "epoch": 0.7456,
+      "learning_rate": 1.997515107957824e-05,
+      "loss": 0.1293,
+      "step": 7456
+    },
+    {
+      "epoch": 0.7458,
+      "learning_rate": 1.9974656794790777e-05,
+      "loss": 0.3809,
+      "step": 7458
+    },
+    {
+      "epoch": 0.746,
+      "learning_rate": 1.9974157648476768e-05,
+      "loss": 0.6587,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7462,
+      "learning_rate": 1.9973653640879486e-05,
+      "loss": 0.1829,
+      "step": 7462
+    },
+    {
+      "epoch": 0.7464,
+      "learning_rate": 1.997314477224458e-05,
+      "loss": 0.929,
+      "step": 7464
+    },
+    {
+      "epoch": 0.7466,
+      "learning_rate": 1.9972631042820074e-05,
+      "loss": 0.4898,
+      "step": 7466
+    },
+    {
+      "epoch": 0.7468,
+      "learning_rate": 1.9972112452856337e-05,
+      "loss": 1.1078,
+      "step": 7468
+    },
+    {
+      "epoch": 0.747,
+      "learning_rate": 1.997158900260614e-05,
+      "loss": 0.9325,
+      "step": 7470
+    },
+    {
+      "epoch": 0.7472,
+      "learning_rate": 1.99710606923246e-05,
+      "loss": 1.15,
+      "step": 7472
+    },
+    {
+      "epoch": 0.7474,
+      "learning_rate": 1.9970527522269204e-05,
+      "loss": 1.4019,
+      "step": 7474
+    },
+    {
+      "epoch": 0.7476,
+      "learning_rate": 1.996998949269982e-05,
+      "loss": 0.785,
+      "step": 7476
+    },
+    {
+      "epoch": 0.7478,
+      "learning_rate": 1.9969446603878673e-05,
+      "loss": 0.5979,
+      "step": 7478
+    },
+    {
+      "epoch": 0.748,
+      "learning_rate": 1.996889885607036e-05,
+      "loss": 0.5565,
+      "step": 7480
+    },
+    {
+      "epoch": 0.7482,
+      "learning_rate": 1.9968346249541848e-05,
+      "loss": 0.9905,
+      "step": 7482
+    },
+    {
+      "epoch": 0.7484,
+      "learning_rate": 1.996778878456247e-05,
+      "loss": 0.142,
+      "step": 7484
+    },
+    {
+      "epoch": 0.7486,
+      "learning_rate": 1.9967226461403934e-05,
+      "loss": 0.8332,
+      "step": 7486
+    },
+    {
+      "epoch": 0.7488,
+      "learning_rate": 1.99666592803403e-05,
+      "loss": 0.3418,
+      "step": 7488
+    },
+    {
+      "epoch": 0.749,
+      "learning_rate": 1.996608724164801e-05,
+      "loss": 0.7715,
+      "step": 7490
+    },
+    {
+      "epoch": 0.7492,
+      "learning_rate": 1.9965510345605866e-05,
+      "loss": 1.5115,
+      "step": 7492
+    },
+    {
+      "epoch": 0.7494,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 1.0176,
+      "step": 7494
+    },
+    {
+      "epoch": 0.7496,
+      "learning_rate": 1.9964341982599083e-05,
+      "loss": 0.9807,
+      "step": 7496
+    },
+    {
+      "epoch": 0.7498,
+      "learning_rate": 1.9963750516203887e-05,
+      "loss": 1.0599,
+      "step": 7498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9963154193597728e-05,
+      "loss": 0.5075,
+      "step": 7500
+    },
+    {
+      "epoch": 0.7502,
+      "learning_rate": 1.996255301507125e-05,
+      "loss": 0.2574,
+      "step": 7502
+    },
+    {
+      "epoch": 0.7504,
+      "learning_rate": 1.9961946980917457e-05,
+      "loss": 0.0926,
+      "step": 7504
+    },
+    {
+      "epoch": 0.7506,
+      "learning_rate": 1.9961336091431728e-05,
+      "loss": 0.941,
+      "step": 7506
+    },
+    {
+      "epoch": 0.7508,
+      "learning_rate": 1.9960720346911798e-05,
+      "loss": 0.621,
+      "step": 7508
+    },
+    {
+      "epoch": 0.751,
+      "learning_rate": 1.9960099747657774e-05,
+      "loss": 0.4453,
+      "step": 7510
+    },
+    {
+      "epoch": 0.7512,
+      "learning_rate": 1.995947429397213e-05,
+      "loss": 1.0801,
+      "step": 7512
+    },
+    {
+      "epoch": 0.7514,
+      "learning_rate": 1.9958843986159705e-05,
+      "loss": 0.9479,
+      "step": 7514
+    },
+    {
+      "epoch": 0.7516,
+      "learning_rate": 1.9958208824527702e-05,
+      "loss": 1.27,
+      "step": 7516
+    },
+    {
+      "epoch": 0.7518,
+      "learning_rate": 1.9957568809385693e-05,
+      "loss": 1.1774,
+      "step": 7518
+    },
+    {
+      "epoch": 0.752,
+      "learning_rate": 1.9956923941045613e-05,
+      "loss": 0.6013,
+      "step": 7520
+    },
+    {
+      "epoch": 0.7522,
+      "learning_rate": 1.995627421982176e-05,
+      "loss": 0.7981,
+      "step": 7522
+    },
+    {
+      "epoch": 0.7524,
+      "learning_rate": 1.99556196460308e-05,
+      "loss": 0.4691,
+      "step": 7524
+    },
+    {
+      "epoch": 0.7526,
+      "learning_rate": 1.995496021999177e-05,
+      "loss": 0.5169,
+      "step": 7526
+    },
+    {
+      "epoch": 0.7528,
+      "learning_rate": 1.9954295942026065e-05,
+      "loss": 1.2294,
+      "step": 7528
+    },
+    {
+      "epoch": 0.753,
+      "learning_rate": 1.995362681245744e-05,
+      "loss": 0.6729,
+      "step": 7530
+    },
+    {
+      "epoch": 0.7532,
+      "learning_rate": 1.9952952831612027e-05,
+      "loss": 0.9754,
+      "step": 7532
+    },
+    {
+      "epoch": 0.7534,
+      "learning_rate": 1.9952273999818312e-05,
+      "loss": 0.125,
+      "step": 7534
+    },
+    {
+      "epoch": 0.7536,
+      "learning_rate": 1.9951590317407152e-05,
+      "loss": 1.0141,
+      "step": 7536
+    },
+    {
+      "epoch": 0.7538,
+      "learning_rate": 1.9950901784711768e-05,
+      "loss": 0.8736,
+      "step": 7538
+    },
+    {
+      "epoch": 0.754,
+      "learning_rate": 1.9950208402067735e-05,
+      "loss": 0.5157,
+      "step": 7540
+    },
+    {
+      "epoch": 0.7542,
+      "learning_rate": 1.9949510169813006e-05,
+      "loss": 0.9804,
+      "step": 7542
+    },
+    {
+      "epoch": 0.7544,
+      "learning_rate": 1.9948807088287884e-05,
+      "loss": 0.9804,
+      "step": 7544
+    },
+    {
+      "epoch": 0.7546,
+      "learning_rate": 1.994809915783505e-05,
+      "loss": 0.2224,
+      "step": 7546
+    },
+    {
+      "epoch": 0.7548,
+      "learning_rate": 1.9947386378799534e-05,
+      "loss": 0.87,
+      "step": 7548
+    },
+    {
+      "epoch": 0.755,
+      "learning_rate": 1.9946668751528745e-05,
+      "loss": 0.5764,
+      "step": 7550
+    },
+    {
+      "epoch": 0.7552,
+      "learning_rate": 1.9945946276372435e-05,
+      "loss": 0.7188,
+      "step": 7552
+    },
+    {
+      "epoch": 0.7554,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 0.2794,
+      "step": 7554
+    },
+    {
+      "epoch": 0.7556,
+      "learning_rate": 1.9944486783814135e-05,
+      "loss": 0.5213,
+      "step": 7556
+    },
+    {
+      "epoch": 0.7558,
+      "learning_rate": 1.994374976712348e-05,
+      "loss": 1.1621,
+      "step": 7558
+    },
+    {
+      "epoch": 0.756,
+      "learning_rate": 1.994300790396999e-05,
+      "loss": 0.862,
+      "step": 7560
+    },
+    {
+      "epoch": 0.7562,
+      "learning_rate": 1.9942261194715236e-05,
+      "loss": 0.3651,
+      "step": 7562
+    },
+    {
+      "epoch": 0.7564,
+      "learning_rate": 1.9941509639723155e-05,
+      "loss": 0.5021,
+      "step": 7564
+    },
+    {
+      "epoch": 0.7566,
+      "learning_rate": 1.9940753239360047e-05,
+      "loss": 1.1564,
+      "step": 7566
+    },
+    {
+      "epoch": 0.7568,
+      "learning_rate": 1.993999199399457e-05,
+      "loss": 0.5801,
+      "step": 7568
+    },
+    {
+      "epoch": 0.757,
+      "learning_rate": 1.9939225903997748e-05,
+      "loss": 0.4574,
+      "step": 7570
+    },
+    {
+      "epoch": 0.7572,
+      "learning_rate": 1.993845496974297e-05,
+      "loss": 0.5487,
+      "step": 7572
+    },
+    {
+      "epoch": 0.7574,
+      "learning_rate": 1.9937679191605964e-05,
+      "loss": 0.7643,
+      "step": 7574
+    },
+    {
+      "epoch": 0.7576,
+      "learning_rate": 1.993689856996485e-05,
+      "loss": 0.5937,
+      "step": 7576
+    },
+    {
+      "epoch": 0.7578,
+      "learning_rate": 1.993611310520009e-05,
+      "loss": 0.8572,
+      "step": 7578
+    },
+    {
+      "epoch": 0.758,
+      "learning_rate": 1.99353227976945e-05,
+      "loss": 0.4907,
+      "step": 7580
+    },
+    {
+      "epoch": 0.7582,
+      "learning_rate": 1.993452764783328e-05,
+      "loss": 0.4774,
+      "step": 7582
+    },
+    {
+      "epoch": 0.7584,
+      "learning_rate": 1.9933727656003964e-05,
+      "loss": 0.4811,
+      "step": 7584
+    },
+    {
+      "epoch": 0.7586,
+      "learning_rate": 1.993292282259647e-05,
+      "loss": 0.7931,
+      "step": 7586
+    },
+    {
+      "epoch": 0.7588,
+      "learning_rate": 1.993211314800306e-05,
+      "loss": 0.7409,
+      "step": 7588
+    },
+    {
+      "epoch": 0.759,
+      "learning_rate": 1.9931298632618352e-05,
+      "loss": 0.6584,
+      "step": 7590
+    },
+    {
+      "epoch": 0.7592,
+      "learning_rate": 1.9930479276839347e-05,
+      "loss": 0.6088,
+      "step": 7592
+    },
+    {
+      "epoch": 0.7594,
+      "learning_rate": 1.9929655081065373e-05,
+      "loss": 0.1561,
+      "step": 7594
+    },
+    {
+      "epoch": 0.7596,
+      "learning_rate": 1.9928826045698138e-05,
+      "loss": 0.0887,
+      "step": 7596
+    },
+    {
+      "epoch": 0.7598,
+      "learning_rate": 1.9927992171141707e-05,
+      "loss": 0.7145,
+      "step": 7598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.99271534578025e-05,
+      "loss": 0.1774,
+      "step": 7600
+    },
+    {
+      "epoch": 0.7602,
+      "learning_rate": 1.992630990608929e-05,
+      "loss": 1.0981,
+      "step": 7602
+    },
+    {
+      "epoch": 0.7604,
+      "learning_rate": 1.9925461516413224e-05,
+      "loss": 1.1392,
+      "step": 7604
+    },
+    {
+      "epoch": 0.7606,
+      "learning_rate": 1.9924608289187786e-05,
+      "loss": 0.6236,
+      "step": 7606
+    },
+    {
+      "epoch": 0.7608,
+      "learning_rate": 1.9923750224828833e-05,
+      "loss": 0.4287,
+      "step": 7608
+    },
+    {
+      "epoch": 0.761,
+      "learning_rate": 1.992288732375458e-05,
+      "loss": 1.1596,
+      "step": 7610
+    },
+    {
+      "epoch": 0.7612,
+      "learning_rate": 1.9922019586385587e-05,
+      "loss": 0.7066,
+      "step": 7612
+    },
+    {
+      "epoch": 0.7614,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.6066,
+      "step": 7614
+    },
+    {
+      "epoch": 0.7616,
+      "learning_rate": 1.9920269604457444e-05,
+      "loss": 0.2537,
+      "step": 7616
+    },
+    {
+      "epoch": 0.7618,
+      "learning_rate": 1.9919387360751216e-05,
+      "loss": 0.5234,
+      "step": 7618
+    },
+    {
+      "epoch": 0.762,
+      "learning_rate": 1.991850028245609e-05,
+      "loss": 0.891,
+      "step": 7620
+    },
+    {
+      "epoch": 0.7622,
+      "learning_rate": 1.9917608370004417e-05,
+      "loss": 0.5325,
+      "step": 7622
+    },
+    {
+      "epoch": 0.7624,
+      "learning_rate": 1.9916711623830904e-05,
+      "loss": 0.6853,
+      "step": 7624
+    },
+    {
+      "epoch": 0.7626,
+      "learning_rate": 1.991581004437262e-05,
+      "loss": 0.9111,
+      "step": 7626
+    },
+    {
+      "epoch": 0.7628,
+      "learning_rate": 1.9914903632068975e-05,
+      "loss": 1.0847,
+      "step": 7628
+    },
+    {
+      "epoch": 0.763,
+      "learning_rate": 1.9913992387361747e-05,
+      "loss": 0.6685,
+      "step": 7630
+    },
+    {
+      "epoch": 0.7632,
+      "learning_rate": 1.9913076310695068e-05,
+      "loss": 0.4339,
+      "step": 7632
+    },
+    {
+      "epoch": 0.7634,
+      "learning_rate": 1.991215540251542e-05,
+      "loss": 1.1663,
+      "step": 7634
+    },
+    {
+      "epoch": 0.7636,
+      "learning_rate": 1.991122966327164e-05,
+      "loss": 0.9285,
+      "step": 7636
+    },
+    {
+      "epoch": 0.7638,
+      "learning_rate": 1.9910299093414932e-05,
+      "loss": 0.3214,
+      "step": 7638
+    },
+    {
+      "epoch": 0.764,
+      "learning_rate": 1.9909363693398828e-05,
+      "loss": 0.5238,
+      "step": 7640
+    },
+    {
+      "epoch": 0.7642,
+      "learning_rate": 1.9908423463679246e-05,
+      "loss": 1.4681,
+      "step": 7642
+    },
+    {
+      "epoch": 0.7644,
+      "learning_rate": 1.990747840471444e-05,
+      "loss": 1.3744,
+      "step": 7644
+    },
+    {
+      "epoch": 0.7646,
+      "learning_rate": 1.990652851696501e-05,
+      "loss": 0.8364,
+      "step": 7646
+    },
+    {
+      "epoch": 0.7648,
+      "learning_rate": 1.990557380089393e-05,
+      "loss": 1.1081,
+      "step": 7648
+    },
+    {
+      "epoch": 0.765,
+      "learning_rate": 1.9904614256966517e-05,
+      "loss": 0.661,
+      "step": 7650
+    },
+    {
+      "epoch": 0.7652,
+      "learning_rate": 1.990364988565043e-05,
+      "loss": 0.9825,
+      "step": 7652
+    },
+    {
+      "epoch": 0.7654,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 0.9151,
+      "step": 7654
+    },
+    {
+      "epoch": 0.7656,
+      "learning_rate": 1.9901706662734708e-05,
+      "loss": 0.8994,
+      "step": 7656
+    },
+    {
+      "epoch": 0.7658,
+      "learning_rate": 1.9900727812082177e-05,
+      "loss": 1.2283,
+      "step": 7658
+    },
+    {
+      "epoch": 0.766,
+      "learning_rate": 1.9899744135935184e-05,
+      "loss": 1.454,
+      "step": 7660
+    },
+    {
+      "epoch": 0.7662,
+      "learning_rate": 1.9898755634773155e-05,
+      "loss": 0.9741,
+      "step": 7662
+    },
+    {
+      "epoch": 0.7664,
+      "learning_rate": 1.989776230907789e-05,
+      "loss": 0.8373,
+      "step": 7664
+    },
+    {
+      "epoch": 0.7666,
+      "learning_rate": 1.9896764159333518e-05,
+      "loss": 1.0294,
+      "step": 7666
+    },
+    {
+      "epoch": 0.7668,
+      "learning_rate": 1.989576118602651e-05,
+      "loss": 0.842,
+      "step": 7668
+    },
+    {
+      "epoch": 0.767,
+      "learning_rate": 1.9894753389645723e-05,
+      "loss": 0.5704,
+      "step": 7670
+    },
+    {
+      "epoch": 0.7672,
+      "learning_rate": 1.9893740770682338e-05,
+      "loss": 0.2812,
+      "step": 7672
+    },
+    {
+      "epoch": 0.7674,
+      "learning_rate": 1.9892723329629885e-05,
+      "loss": 0.1427,
+      "step": 7674
+    },
+    {
+      "epoch": 0.7676,
+      "learning_rate": 1.9891701066984264e-05,
+      "loss": 0.0067,
+      "step": 7676
+    },
+    {
+      "epoch": 0.7678,
+      "learning_rate": 1.9890673983243704e-05,
+      "loss": 0.9678,
+      "step": 7678
+    },
+    {
+      "epoch": 0.768,
+      "learning_rate": 1.9889642078908805e-05,
+      "loss": 0.4865,
+      "step": 7680
+    },
+    {
+      "epoch": 0.7682,
+      "learning_rate": 1.9888605354482494e-05,
+      "loss": 0.5029,
+      "step": 7682
+    },
+    {
+      "epoch": 0.7684,
+      "learning_rate": 1.988756381047006e-05,
+      "loss": 1.0885,
+      "step": 7684
+    },
+    {
+      "epoch": 0.7686,
+      "learning_rate": 1.9886517447379143e-05,
+      "loss": 0.3052,
+      "step": 7686
+    },
+    {
+      "epoch": 0.7688,
+      "learning_rate": 1.9885466265719723e-05,
+      "loss": 0.6889,
+      "step": 7688
+    },
+    {
+      "epoch": 0.769,
+      "learning_rate": 1.9884410266004134e-05,
+      "loss": 0.6428,
+      "step": 7690
+    },
+    {
+      "epoch": 0.7692,
+      "learning_rate": 1.988334944874706e-05,
+      "loss": 1.3165,
+      "step": 7692
+    },
+    {
+      "epoch": 0.7694,
+      "learning_rate": 1.988228381446553e-05,
+      "loss": 0.5966,
+      "step": 7694
+    },
+    {
+      "epoch": 0.7696,
+      "learning_rate": 1.988121336367892e-05,
+      "loss": 0.7879,
+      "step": 7696
+    },
+    {
+      "epoch": 0.7698,
+      "learning_rate": 1.9880138096908955e-05,
+      "loss": 0.9454,
+      "step": 7698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9879058014679707e-05,
+      "loss": 0.9208,
+      "step": 7700
+    },
+    {
+      "epoch": 0.7702,
+      "learning_rate": 1.987797311751759e-05,
+      "loss": 0.8658,
+      "step": 7702
+    },
+    {
+      "epoch": 0.7704,
+      "learning_rate": 1.987688340595138e-05,
+      "loss": 0.8758,
+      "step": 7704
+    },
+    {
+      "epoch": 0.7706,
+      "learning_rate": 1.9875788880512183e-05,
+      "loss": 0.1012,
+      "step": 7706
+    },
+    {
+      "epoch": 0.7708,
+      "learning_rate": 1.9874689541733455e-05,
+      "loss": 0.3366,
+      "step": 7708
+    },
+    {
+      "epoch": 0.771,
+      "learning_rate": 1.9873585390151007e-05,
+      "loss": 0.5943,
+      "step": 7710
+    },
+    {
+      "epoch": 0.7712,
+      "learning_rate": 1.987247642630298e-05,
+      "loss": 0.3728,
+      "step": 7712
+    },
+    {
+      "epoch": 0.7714,
+      "learning_rate": 1.987136265072988e-05,
+      "loss": 1.1002,
+      "step": 7714
+    },
+    {
+      "epoch": 0.7716,
+      "learning_rate": 1.9870244063974544e-05,
+      "loss": 0.8429,
+      "step": 7716
+    },
+    {
+      "epoch": 0.7718,
+      "learning_rate": 1.9869120666582153e-05,
+      "loss": 0.7057,
+      "step": 7718
+    },
+    {
+      "epoch": 0.772,
+      "learning_rate": 1.986799245910024e-05,
+      "loss": 0.4487,
+      "step": 7720
+    },
+    {
+      "epoch": 0.7722,
+      "learning_rate": 1.9866859442078685e-05,
+      "loss": 0.2313,
+      "step": 7722
+    },
+    {
+      "epoch": 0.7724,
+      "learning_rate": 1.9865721616069695e-05,
+      "loss": 1.0077,
+      "step": 7724
+    },
+    {
+      "epoch": 0.7726,
+      "learning_rate": 1.9864578981627844e-05,
+      "loss": 0.0893,
+      "step": 7726
+    },
+    {
+      "epoch": 0.7728,
+      "learning_rate": 1.986343153931003e-05,
+      "loss": 0.83,
+      "step": 7728
+    },
+    {
+      "epoch": 0.773,
+      "learning_rate": 1.986227928967551e-05,
+      "loss": 0.4653,
+      "step": 7730
+    },
+    {
+      "epoch": 0.7732,
+      "learning_rate": 1.9861122233285876e-05,
+      "loss": 1.0956,
+      "step": 7732
+    },
+    {
+      "epoch": 0.7734,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 1.0626,
+      "step": 7734
+    },
+    {
+      "epoch": 0.7736,
+      "learning_rate": 1.9858793702499326e-05,
+      "loss": 0.7638,
+      "step": 7736
+    },
+    {
+      "epoch": 0.7738,
+      "learning_rate": 1.985762222923732e-05,
+      "loss": 0.1432,
+      "step": 7738
+    },
+    {
+      "epoch": 0.774,
+      "learning_rate": 1.9856445951489984e-05,
+      "loss": 0.494,
+      "step": 7740
+    },
+    {
+      "epoch": 0.7742,
+      "learning_rate": 1.985526486983063e-05,
+      "loss": 0.5896,
+      "step": 7742
+    },
+    {
+      "epoch": 0.7744,
+      "learning_rate": 1.9854078984834908e-05,
+      "loss": 0.5324,
+      "step": 7744
+    },
+    {
+      "epoch": 0.7746,
+      "learning_rate": 1.985288829708079e-05,
+      "loss": 0.1827,
+      "step": 7746
+    },
+    {
+      "epoch": 0.7748,
+      "learning_rate": 1.9851692807148612e-05,
+      "loss": 1.0772,
+      "step": 7748
+    },
+    {
+      "epoch": 0.775,
+      "learning_rate": 1.9850492515621038e-05,
+      "loss": 0.3188,
+      "step": 7750
+    },
+    {
+      "epoch": 0.7752,
+      "learning_rate": 1.9849287423083082e-05,
+      "loss": 1.0576,
+      "step": 7752
+    },
+    {
+      "epoch": 0.7754,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.1904,
+      "step": 7754
+    },
+    {
+      "epoch": 0.7756,
+      "learning_rate": 1.984686283732773e-05,
+      "loss": 0.5452,
+      "step": 7756
+    },
+    {
+      "epoch": 0.7758,
+      "learning_rate": 1.9845643345292055e-05,
+      "loss": 0.5827,
+      "step": 7758
+    },
+    {
+      "epoch": 0.776,
+      "learning_rate": 1.984441905460942e-05,
+      "loss": 0.7955,
+      "step": 7760
+    },
+    {
+      "epoch": 0.7762,
+      "learning_rate": 1.9843189965876525e-05,
+      "loss": 0.5688,
+      "step": 7762
+    },
+    {
+      "epoch": 0.7764,
+      "learning_rate": 1.9841956079692424e-05,
+      "loss": 0.4709,
+      "step": 7764
+    },
+    {
+      "epoch": 0.7766,
+      "learning_rate": 1.9840717396658483e-05,
+      "loss": 0.7181,
+      "step": 7766
+    },
+    {
+      "epoch": 0.7768,
+      "learning_rate": 1.9839473917378432e-05,
+      "loss": 0.4789,
+      "step": 7768
+    },
+    {
+      "epoch": 0.777,
+      "learning_rate": 1.983822564245833e-05,
+      "loss": 2.1085,
+      "step": 7770
+    },
+    {
+      "epoch": 0.7772,
+      "learning_rate": 1.9836972572506564e-05,
+      "loss": 0.7911,
+      "step": 7772
+    },
+    {
+      "epoch": 0.7774,
+      "learning_rate": 1.983571470813386e-05,
+      "loss": 0.9938,
+      "step": 7774
+    },
+    {
+      "epoch": 0.7776,
+      "learning_rate": 1.98344520499533e-05,
+      "loss": 0.6705,
+      "step": 7776
+    },
+    {
+      "epoch": 0.7778,
+      "learning_rate": 1.9833184598580276e-05,
+      "loss": 0.9577,
+      "step": 7778
+    },
+    {
+      "epoch": 0.778,
+      "learning_rate": 1.9831912354632537e-05,
+      "loss": 0.5588,
+      "step": 7780
+    },
+    {
+      "epoch": 0.7782,
+      "learning_rate": 1.983063531873016e-05,
+      "loss": 0.7439,
+      "step": 7782
+    },
+    {
+      "epoch": 0.7784,
+      "learning_rate": 1.982935349149554e-05,
+      "loss": 1.1386,
+      "step": 7784
+    },
+    {
+      "epoch": 0.7786,
+      "learning_rate": 1.982806687355345e-05,
+      "loss": 0.7437,
+      "step": 7786
+    },
+    {
+      "epoch": 0.7788,
+      "learning_rate": 1.9826775465530954e-05,
+      "loss": 0.5568,
+      "step": 7788
+    },
+    {
+      "epoch": 0.779,
+      "learning_rate": 1.982547926805747e-05,
+      "loss": 0.807,
+      "step": 7790
+    },
+    {
+      "epoch": 0.7792,
+      "learning_rate": 1.9824178281764757e-05,
+      "loss": 1.2333,
+      "step": 7792
+    },
+    {
+      "epoch": 0.7794,
+      "learning_rate": 1.982287250728689e-05,
+      "loss": 0.1299,
+      "step": 7794
+    },
+    {
+      "epoch": 0.7796,
+      "learning_rate": 1.9821561945260292e-05,
+      "loss": 0.2063,
+      "step": 7796
+    },
+    {
+      "epoch": 0.7798,
+      "learning_rate": 1.9820246596323724e-05,
+      "loss": 0.8336,
+      "step": 7798
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9818926461118254e-05,
+      "loss": 0.9771,
+      "step": 7800
+    },
+    {
+      "epoch": 0.7802,
+      "learning_rate": 1.981760154028731e-05,
+      "loss": 0.7736,
+      "step": 7802
+    },
+    {
+      "epoch": 0.7804,
+      "learning_rate": 1.9816271834476642e-05,
+      "loss": 0.7351,
+      "step": 7804
+    },
+    {
+      "epoch": 0.7806,
+      "learning_rate": 1.981493734433433e-05,
+      "loss": 0.4353,
+      "step": 7806
+    },
+    {
+      "epoch": 0.7808,
+      "learning_rate": 1.9813598070510794e-05,
+      "loss": 0.6942,
+      "step": 7808
+    },
+    {
+      "epoch": 0.781,
+      "learning_rate": 1.9812254013658773e-05,
+      "loss": 1.0758,
+      "step": 7810
+    },
+    {
+      "epoch": 0.7812,
+      "learning_rate": 1.981090517443334e-05,
+      "loss": 1.2862,
+      "step": 7812
+    },
+    {
+      "epoch": 0.7814,
+      "learning_rate": 1.9809551553491918e-05,
+      "loss": 0.7897,
+      "step": 7814
+    },
+    {
+      "epoch": 0.7816,
+      "learning_rate": 1.9808193151494237e-05,
+      "loss": 0.9382,
+      "step": 7816
+    },
+    {
+      "epoch": 0.7818,
+      "learning_rate": 1.9806829969102356e-05,
+      "loss": 1.1242,
+      "step": 7818
+    },
+    {
+      "epoch": 0.782,
+      "learning_rate": 1.9805462006980688e-05,
+      "loss": 0.735,
+      "step": 7820
+    },
+    {
+      "epoch": 0.7822,
+      "learning_rate": 1.9804089265795963e-05,
+      "loss": 1.1687,
+      "step": 7822
+    },
+    {
+      "epoch": 0.7824,
+      "learning_rate": 1.9802711746217222e-05,
+      "loss": 0.6669,
+      "step": 7824
+    },
+    {
+      "epoch": 0.7826,
+      "learning_rate": 1.9801329448915863e-05,
+      "loss": 1.2934,
+      "step": 7826
+    },
+    {
+      "epoch": 0.7828,
+      "learning_rate": 1.9799942374565594e-05,
+      "loss": 0.4907,
+      "step": 7828
+    },
+    {
+      "epoch": 0.783,
+      "learning_rate": 1.979855052384247e-05,
+      "loss": 1.0273,
+      "step": 7830
+    },
+    {
+      "epoch": 0.7832,
+      "learning_rate": 1.9797153897424854e-05,
+      "loss": 0.9751,
+      "step": 7832
+    },
+    {
+      "epoch": 0.7834,
+      "learning_rate": 1.979575249599344e-05,
+      "loss": 0.8235,
+      "step": 7834
+    },
+    {
+      "epoch": 0.7836,
+      "learning_rate": 1.979434632023127e-05,
+      "loss": 0.8248,
+      "step": 7836
+    },
+    {
+      "epoch": 0.7838,
+      "learning_rate": 1.979293537082368e-05,
+      "loss": 1.0417,
+      "step": 7838
+    },
+    {
+      "epoch": 0.784,
+      "learning_rate": 1.979151964845835e-05,
+      "loss": 0.6114,
+      "step": 7840
+    },
+    {
+      "epoch": 0.7842,
+      "learning_rate": 1.9790099153825303e-05,
+      "loss": 1.0652,
+      "step": 7842
+    },
+    {
+      "epoch": 0.7844,
+      "learning_rate": 1.978867388761686e-05,
+      "loss": 0.2119,
+      "step": 7844
+    },
+    {
+      "epoch": 0.7846,
+      "learning_rate": 1.9787243850527663e-05,
+      "loss": 0.6605,
+      "step": 7846
+    },
+    {
+      "epoch": 0.7848,
+      "learning_rate": 1.9785809043254724e-05,
+      "loss": 0.3529,
+      "step": 7848
+    },
+    {
+      "epoch": 0.785,
+      "learning_rate": 1.978436946649733e-05,
+      "loss": 0.5195,
+      "step": 7850
+    },
+    {
+      "epoch": 0.7852,
+      "learning_rate": 1.978292512095713e-05,
+      "loss": 0.9894,
+      "step": 7852
+    },
+    {
+      "epoch": 0.7854,
+      "learning_rate": 1.978147600733806e-05,
+      "loss": 0.4273,
+      "step": 7854
+    },
+    {
+      "epoch": 0.7856,
+      "learning_rate": 1.9780022126346413e-05,
+      "loss": 0.5317,
+      "step": 7856
+    },
+    {
+      "epoch": 0.7858,
+      "learning_rate": 1.9778563478690793e-05,
+      "loss": 0.8866,
+      "step": 7858
+    },
+    {
+      "epoch": 0.786,
+      "learning_rate": 1.9777100065082123e-05,
+      "loss": 0.3619,
+      "step": 7860
+    },
+    {
+      "epoch": 0.7862,
+      "learning_rate": 1.977563188623365e-05,
+      "loss": 0.3709,
+      "step": 7862
+    },
+    {
+      "epoch": 0.7864,
+      "learning_rate": 1.9774158942860962e-05,
+      "loss": 0.4778,
+      "step": 7864
+    },
+    {
+      "epoch": 0.7866,
+      "learning_rate": 1.977268123568194e-05,
+      "loss": 0.6722,
+      "step": 7866
+    },
+    {
+      "epoch": 0.7868,
+      "learning_rate": 1.9771198765416795e-05,
+      "loss": 0.656,
+      "step": 7868
+    },
+    {
+      "epoch": 0.787,
+      "learning_rate": 1.9769711532788086e-05,
+      "loss": 0.3995,
+      "step": 7870
+    },
+    {
+      "epoch": 0.7872,
+      "learning_rate": 1.976821953852065e-05,
+      "loss": 0.1479,
+      "step": 7872
+    },
+    {
+      "epoch": 0.7874,
+      "learning_rate": 1.9766722783341682e-05,
+      "loss": 0.9349,
+      "step": 7874
+    },
+    {
+      "epoch": 0.7876,
+      "learning_rate": 1.9765221267980678e-05,
+      "loss": 0.507,
+      "step": 7876
+    },
+    {
+      "epoch": 0.7878,
+      "learning_rate": 1.9763714993169448e-05,
+      "loss": 0.9296,
+      "step": 7878
+    },
+    {
+      "epoch": 0.788,
+      "learning_rate": 1.976220395964215e-05,
+      "loss": 0.1644,
+      "step": 7880
+    },
+    {
+      "epoch": 0.7882,
+      "learning_rate": 1.9760688168135236e-05,
+      "loss": 0.7834,
+      "step": 7882
+    },
+    {
+      "epoch": 0.7884,
+      "learning_rate": 1.9759167619387474e-05,
+      "loss": 0.7755,
+      "step": 7884
+    },
+    {
+      "epoch": 0.7886,
+      "learning_rate": 1.9757642314139977e-05,
+      "loss": 0.834,
+      "step": 7886
+    },
+    {
+      "epoch": 0.7888,
+      "learning_rate": 1.9756112253136157e-05,
+      "loss": 0.4884,
+      "step": 7888
+    },
+    {
+      "epoch": 0.789,
+      "learning_rate": 1.9754577437121733e-05,
+      "loss": 0.8348,
+      "step": 7890
+    },
+    {
+      "epoch": 0.7892,
+      "learning_rate": 1.975303786684477e-05,
+      "loss": 0.8031,
+      "step": 7892
+    },
+    {
+      "epoch": 0.7894,
+      "learning_rate": 1.9751493543055638e-05,
+      "loss": 0.6015,
+      "step": 7894
+    },
+    {
+      "epoch": 0.7896,
+      "learning_rate": 1.9749944466507007e-05,
+      "loss": 0.6112,
+      "step": 7896
+    },
+    {
+      "epoch": 0.7898,
+      "learning_rate": 1.974839063795389e-05,
+      "loss": 0.6202,
+      "step": 7898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.97468320581536e-05,
+      "loss": 0.724,
+      "step": 7900
+    },
+    {
+      "epoch": 0.7902,
+      "learning_rate": 1.9745268727865774e-05,
+      "loss": 1.0302,
+      "step": 7902
+    },
+    {
+      "epoch": 0.7904,
+      "learning_rate": 1.974370064785236e-05,
+      "loss": 1.2383,
+      "step": 7904
+    },
+    {
+      "epoch": 0.7906,
+      "learning_rate": 1.97421278188776e-05,
+      "loss": 0.5936,
+      "step": 7906
+    },
+    {
+      "epoch": 0.7908,
+      "learning_rate": 1.974055024170811e-05,
+      "loss": 0.7086,
+      "step": 7908
+    },
+    {
+      "epoch": 0.791,
+      "learning_rate": 1.973896791711276e-05,
+      "loss": 1.0987,
+      "step": 7910
+    },
+    {
+      "epoch": 0.7912,
+      "learning_rate": 1.9737380845862745e-05,
+      "loss": 0.263,
+      "step": 7912
+    },
+    {
+      "epoch": 0.7914,
+      "learning_rate": 1.9735789028731607e-05,
+      "loss": 1.0184,
+      "step": 7914
+    },
+    {
+      "epoch": 0.7916,
+      "learning_rate": 1.973419246649517e-05,
+      "loss": 0.5324,
+      "step": 7916
+    },
+    {
+      "epoch": 0.7918,
+      "learning_rate": 1.9732591159931567e-05,
+      "loss": 0.4339,
+      "step": 7918
+    },
+    {
+      "epoch": 0.792,
+      "learning_rate": 1.9730985109821268e-05,
+      "loss": 0.7232,
+      "step": 7920
+    },
+    {
+      "epoch": 0.7922,
+      "learning_rate": 1.9729374316947037e-05,
+      "loss": 0.429,
+      "step": 7922
+    },
+    {
+      "epoch": 0.7924,
+      "learning_rate": 1.972775878209397e-05,
+      "loss": 0.6186,
+      "step": 7924
+    },
+    {
+      "epoch": 0.7926,
+      "learning_rate": 1.972613850604944e-05,
+      "loss": 0.5469,
+      "step": 7926
+    },
+    {
+      "epoch": 0.7928,
+      "learning_rate": 1.972451348960315e-05,
+      "loss": 0.7405,
+      "step": 7928
+    },
+    {
+      "epoch": 0.793,
+      "learning_rate": 1.972288373354713e-05,
+      "loss": 0.2898,
+      "step": 7930
+    },
+    {
+      "epoch": 0.7932,
+      "learning_rate": 1.972124923867569e-05,
+      "loss": 0.4116,
+      "step": 7932
+    },
+    {
+      "epoch": 0.7934,
+      "learning_rate": 1.9719610005785463e-05,
+      "loss": 0.1168,
+      "step": 7934
+    },
+    {
+      "epoch": 0.7936,
+      "learning_rate": 1.97179660356754e-05,
+      "loss": 0.1958,
+      "step": 7936
+    },
+    {
+      "epoch": 0.7938,
+      "learning_rate": 1.9716317329146743e-05,
+      "loss": 0.5753,
+      "step": 7938
+    },
+    {
+      "epoch": 0.794,
+      "learning_rate": 1.971466388700305e-05,
+      "loss": 0.8319,
+      "step": 7940
+    },
+    {
+      "epoch": 0.7942,
+      "learning_rate": 1.9713005710050206e-05,
+      "loss": 0.6294,
+      "step": 7942
+    },
+    {
+      "epoch": 0.7944,
+      "learning_rate": 1.971134279909636e-05,
+      "loss": 0.3257,
+      "step": 7944
+    },
+    {
+      "epoch": 0.7946,
+      "learning_rate": 1.9709675154952017e-05,
+      "loss": 0.9196,
+      "step": 7946
+    },
+    {
+      "epoch": 0.7948,
+      "learning_rate": 1.970800277842996e-05,
+      "loss": 0.8395,
+      "step": 7948
+    },
+    {
+      "epoch": 0.795,
+      "learning_rate": 1.9706325670345276e-05,
+      "loss": 0.631,
+      "step": 7950
+    },
+    {
+      "epoch": 0.7952,
+      "learning_rate": 1.9704643831515377e-05,
+      "loss": 0.388,
+      "step": 7952
+    },
+    {
+      "epoch": 0.7954,
+      "learning_rate": 1.970295726275997e-05,
+      "loss": 0.2922,
+      "step": 7954
+    },
+    {
+      "epoch": 0.7956,
+      "learning_rate": 1.970126596490106e-05,
+      "loss": 0.9204,
+      "step": 7956
+    },
+    {
+      "epoch": 0.7958,
+      "learning_rate": 1.9699569938762975e-05,
+      "loss": 1.0726,
+      "step": 7958
+    },
+    {
+      "epoch": 0.796,
+      "learning_rate": 1.9697869185172334e-05,
+      "loss": 0.6046,
+      "step": 7960
+    },
+    {
+      "epoch": 0.7962,
+      "learning_rate": 1.969616370495806e-05,
+      "loss": 0.2351,
+      "step": 7962
+    },
+    {
+      "epoch": 0.7964,
+      "learning_rate": 1.9694453498951396e-05,
+      "loss": 0.241,
+      "step": 7964
+    },
+    {
+      "epoch": 0.7966,
+      "learning_rate": 1.969273856798586e-05,
+      "loss": 1.0779,
+      "step": 7966
+    },
+    {
+      "epoch": 0.7968,
+      "learning_rate": 1.9691018912897285e-05,
+      "loss": 0.8246,
+      "step": 7968
+    },
+    {
+      "epoch": 0.797,
+      "learning_rate": 1.9689294534523836e-05,
+      "loss": 0.7008,
+      "step": 7970
+    },
+    {
+      "epoch": 0.7972,
+      "learning_rate": 1.9687565433705926e-05,
+      "loss": 0.6866,
+      "step": 7972
+    },
+    {
+      "epoch": 0.7974,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 0.454,
+      "step": 7974
+    },
+    {
+      "epoch": 0.7976,
+      "learning_rate": 1.9684093068110042e-05,
+      "loss": 0.7798,
+      "step": 7976
+    },
+    {
+      "epoch": 0.7978,
+      "learning_rate": 1.9682349805024447e-05,
+      "loss": 0.018,
+      "step": 7978
+    },
+    {
+      "epoch": 0.798,
+      "learning_rate": 1.968060182287918e-05,
+      "loss": 1.2888,
+      "step": 7980
+    },
+    {
+      "epoch": 0.7982,
+      "learning_rate": 1.9678849122526195e-05,
+      "loss": 0.6905,
+      "step": 7982
+    },
+    {
+      "epoch": 0.7984,
+      "learning_rate": 1.9677091704819714e-05,
+      "loss": 1.3352,
+      "step": 7984
+    },
+    {
+      "epoch": 0.7986,
+      "learning_rate": 1.9675329570616302e-05,
+      "loss": 1.0061,
+      "step": 7986
+    },
+    {
+      "epoch": 0.7988,
+      "learning_rate": 1.96735627207748e-05,
+      "loss": 0.9271,
+      "step": 7988
+    },
+    {
+      "epoch": 0.799,
+      "learning_rate": 1.967179115615633e-05,
+      "loss": 0.0598,
+      "step": 7990
+    },
+    {
+      "epoch": 0.7992,
+      "learning_rate": 1.9670014877624353e-05,
+      "loss": 0.7695,
+      "step": 7992
+    },
+    {
+      "epoch": 0.7994,
+      "learning_rate": 1.966823388604459e-05,
+      "loss": 0.8895,
+      "step": 7994
+    },
+    {
+      "epoch": 0.7996,
+      "learning_rate": 1.9666448182285095e-05,
+      "loss": 0.4303,
+      "step": 7996
+    },
+    {
+      "epoch": 0.7998,
+      "learning_rate": 1.966465776721618e-05,
+      "loss": 0.3796,
+      "step": 7998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966286264171047e-05,
+      "loss": 0.8889,
+      "step": 8000
+    },
+    {
+      "epoch": 0.8002,
+      "learning_rate": 1.9661062806642906e-05,
+      "loss": 0.3293,
+      "step": 8002
+    },
+    {
+      "epoch": 0.8004,
+      "learning_rate": 1.9659258262890687e-05,
+      "loss": 0.8586,
+      "step": 8004
+    },
+    {
+      "epoch": 0.8006,
+      "learning_rate": 1.9657449011333328e-05,
+      "loss": 0.5408,
+      "step": 8006
+    },
+    {
+      "epoch": 0.8008,
+      "learning_rate": 1.965563505285265e-05,
+      "loss": 0.6614,
+      "step": 8008
+    },
+    {
+      "epoch": 0.801,
+      "learning_rate": 1.9653816388332743e-05,
+      "loss": 0.7186,
+      "step": 8010
+    },
+    {
+      "epoch": 0.8012,
+      "learning_rate": 1.965199301866e-05,
+      "loss": 0.6962,
+      "step": 8012
+    },
+    {
+      "epoch": 0.8014,
+      "learning_rate": 1.965016494472312e-05,
+      "loss": 0.4745,
+      "step": 8014
+    },
+    {
+      "epoch": 0.8016,
+      "learning_rate": 1.9648332167413074e-05,
+      "loss": 0.8267,
+      "step": 8016
+    },
+    {
+      "epoch": 0.8018,
+      "learning_rate": 1.964649468762313e-05,
+      "loss": 0.9258,
+      "step": 8018
+    },
+    {
+      "epoch": 0.802,
+      "learning_rate": 1.964465250624888e-05,
+      "loss": 0.4387,
+      "step": 8020
+    },
+    {
+      "epoch": 0.8022,
+      "learning_rate": 1.964280562418815e-05,
+      "loss": 0.4178,
+      "step": 8022
+    },
+    {
+      "epoch": 0.8024,
+      "learning_rate": 1.96409540423411e-05,
+      "loss": 0.5309,
+      "step": 8024
+    },
+    {
+      "epoch": 0.8026,
+      "learning_rate": 1.963909776161018e-05,
+      "loss": 0.1216,
+      "step": 8026
+    },
+    {
+      "epoch": 0.8028,
+      "learning_rate": 1.96372367829001e-05,
+      "loss": 0.7272,
+      "step": 8028
+    },
+    {
+      "epoch": 0.803,
+      "learning_rate": 1.963537110711789e-05,
+      "loss": 0.265,
+      "step": 8030
+    },
+    {
+      "epoch": 0.8032,
+      "learning_rate": 1.9633500735172856e-05,
+      "loss": 0.7192,
+      "step": 8032
+    },
+    {
+      "epoch": 0.8034,
+      "learning_rate": 1.9631625667976584e-05,
+      "loss": 0.4044,
+      "step": 8034
+    },
+    {
+      "epoch": 0.8036,
+      "learning_rate": 1.9629745906442977e-05,
+      "loss": 0.626,
+      "step": 8036
+    },
+    {
+      "epoch": 0.8038,
+      "learning_rate": 1.9627861451488194e-05,
+      "loss": 0.7822,
+      "step": 8038
+    },
+    {
+      "epoch": 0.804,
+      "learning_rate": 1.9625972304030697e-05,
+      "loss": 1.5276,
+      "step": 8040
+    },
+    {
+      "epoch": 0.8042,
+      "learning_rate": 1.9624078464991246e-05,
+      "loss": 0.1861,
+      "step": 8042
+    },
+    {
+      "epoch": 0.8044,
+      "learning_rate": 1.9622179935292855e-05,
+      "loss": 1.0433,
+      "step": 8044
+    },
+    {
+      "epoch": 0.8046,
+      "learning_rate": 1.962027671586086e-05,
+      "loss": 0.0276,
+      "step": 8046
+    },
+    {
+      "epoch": 0.8048,
+      "learning_rate": 1.9618368807622867e-05,
+      "loss": 0.6553,
+      "step": 8048
+    },
+    {
+      "epoch": 0.805,
+      "learning_rate": 1.9616456211508756e-05,
+      "loss": 3.8095,
+      "step": 8050
+    },
+    {
+      "epoch": 0.8052,
+      "learning_rate": 1.961453892845071e-05,
+      "loss": 0.3045,
+      "step": 8052
+    },
+    {
+      "epoch": 0.8054,
+      "learning_rate": 1.9612616959383194e-05,
+      "loss": 0.9311,
+      "step": 8054
+    },
+    {
+      "epoch": 0.8056,
+      "learning_rate": 1.961069030524294e-05,
+      "loss": 0.7183,
+      "step": 8056
+    },
+    {
+      "epoch": 0.8058,
+      "learning_rate": 1.9608758966968987e-05,
+      "loss": 1.2392,
+      "step": 8058
+    },
+    {
+      "epoch": 0.806,
+      "learning_rate": 1.960682294550265e-05,
+      "loss": 0.1933,
+      "step": 8060
+    },
+    {
+      "epoch": 0.8062,
+      "learning_rate": 1.96048822417875e-05,
+      "loss": 1.1359,
+      "step": 8062
+    },
+    {
+      "epoch": 0.8064,
+      "learning_rate": 1.9602936856769435e-05,
+      "loss": 0.4746,
+      "step": 8064
+    },
+    {
+      "epoch": 0.8066,
+      "learning_rate": 1.9600986791396597e-05,
+      "loss": 0.7606,
+      "step": 8066
+    },
+    {
+      "epoch": 0.8068,
+      "learning_rate": 1.959903204661944e-05,
+      "loss": 0.814,
+      "step": 8068
+    },
+    {
+      "epoch": 0.807,
+      "learning_rate": 1.9597072623390668e-05,
+      "loss": 0.81,
+      "step": 8070
+    },
+    {
+      "epoch": 0.8072,
+      "learning_rate": 1.9595108522665285e-05,
+      "loss": 0.5159,
+      "step": 8072
+    },
+    {
+      "epoch": 0.8074,
+      "learning_rate": 1.9593139745400578e-05,
+      "loss": 0.7015,
+      "step": 8074
+    },
+    {
+      "epoch": 0.8076,
+      "learning_rate": 1.9591166292556093e-05,
+      "loss": 0.9511,
+      "step": 8076
+    },
+    {
+      "epoch": 0.8078,
+      "learning_rate": 1.9589188165093666e-05,
+      "loss": 0.8208,
+      "step": 8078
+    },
+    {
+      "epoch": 0.808,
+      "learning_rate": 1.958720536397743e-05,
+      "loss": 0.7538,
+      "step": 8080
+    },
+    {
+      "epoch": 0.8082,
+      "learning_rate": 1.9585217890173765e-05,
+      "loss": 1.1652,
+      "step": 8082
+    },
+    {
+      "epoch": 0.8084,
+      "learning_rate": 1.958322574465133e-05,
+      "loss": 0.4986,
+      "step": 8084
+    },
+    {
+      "epoch": 0.8086,
+      "learning_rate": 1.95812289283811e-05,
+      "loss": 0.5137,
+      "step": 8086
+    },
+    {
+      "epoch": 0.8088,
+      "learning_rate": 1.9579227442336283e-05,
+      "loss": 0.8867,
+      "step": 8088
+    },
+    {
+      "epoch": 0.809,
+      "learning_rate": 1.957722128749237e-05,
+      "loss": 1.3153,
+      "step": 8090
+    },
+    {
+      "epoch": 0.8092,
+      "learning_rate": 1.9575210464827154e-05,
+      "loss": 1.0713,
+      "step": 8092
+    },
+    {
+      "epoch": 0.8094,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 0.7994,
+      "step": 8094
+    },
+    {
+      "epoch": 0.8096,
+      "learning_rate": 1.9571174819955264e-05,
+      "loss": 0.692,
+      "step": 8096
+    },
+    {
+      "epoch": 0.8098,
+      "learning_rate": 1.9569149999715518e-05,
+      "loss": 0.5534,
+      "step": 8098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9567120515588304e-05,
+      "loss": 0.8386,
+      "step": 8100
+    },
+    {
+      "epoch": 0.8102,
+      "learning_rate": 1.9565086368562784e-05,
+      "loss": 1.0349,
+      "step": 8102
+    },
+    {
+      "epoch": 0.8104,
+      "learning_rate": 1.9563047559630363e-05,
+      "loss": 1.092,
+      "step": 8104
+    },
+    {
+      "epoch": 0.8106,
+      "learning_rate": 1.9561004089784722e-05,
+      "loss": 0.27,
+      "step": 8106
+    },
+    {
+      "epoch": 0.8108,
+      "learning_rate": 1.9558955960021854e-05,
+      "loss": 0.6844,
+      "step": 8108
+    },
+    {
+      "epoch": 0.811,
+      "learning_rate": 1.9556903171339966e-05,
+      "loss": 1.0912,
+      "step": 8110
+    },
+    {
+      "epoch": 0.8112,
+      "learning_rate": 1.9554845724739565e-05,
+      "loss": 0.9603,
+      "step": 8112
+    },
+    {
+      "epoch": 0.8114,
+      "learning_rate": 1.955278362122344e-05,
+      "loss": 0.8549,
+      "step": 8114
+    },
+    {
+      "epoch": 0.8116,
+      "learning_rate": 1.9550716861796623e-05,
+      "loss": 0.7135,
+      "step": 8116
+    },
+    {
+      "epoch": 0.8118,
+      "learning_rate": 1.954864544746643e-05,
+      "loss": 0.7793,
+      "step": 8118
+    },
+    {
+      "epoch": 0.812,
+      "learning_rate": 1.954656937924245e-05,
+      "loss": 0.5295,
+      "step": 8120
+    },
+    {
+      "epoch": 0.8122,
+      "learning_rate": 1.954448865813652e-05,
+      "loss": 0.4491,
+      "step": 8122
+    },
+    {
+      "epoch": 0.8124,
+      "learning_rate": 1.954240328516277e-05,
+      "loss": 0.7948,
+      "step": 8124
+    },
+    {
+      "epoch": 0.8126,
+      "learning_rate": 1.9540313261337585e-05,
+      "loss": 1.3837,
+      "step": 8126
+    },
+    {
+      "epoch": 0.8128,
+      "learning_rate": 1.9538218587679605e-05,
+      "loss": 0.5281,
+      "step": 8128
+    },
+    {
+      "epoch": 0.813,
+      "learning_rate": 1.9536119265209763e-05,
+      "loss": 2.2118,
+      "step": 8130
+    },
+    {
+      "epoch": 0.8132,
+      "learning_rate": 1.953401529495124e-05,
+      "loss": 0.5778,
+      "step": 8132
+    },
+    {
+      "epoch": 0.8134,
+      "learning_rate": 1.9531906677929472e-05,
+      "loss": 0.0896,
+      "step": 8134
+    },
+    {
+      "epoch": 0.8136,
+      "learning_rate": 1.952979341517219e-05,
+      "loss": 1.0871,
+      "step": 8136
+    },
+    {
+      "epoch": 0.8138,
+      "learning_rate": 1.9527675507709364e-05,
+      "loss": 1.001,
+      "step": 8138
+    },
+    {
+      "epoch": 0.814,
+      "learning_rate": 1.9525552956573244e-05,
+      "loss": 0.9954,
+      "step": 8140
+    },
+    {
+      "epoch": 0.8142,
+      "learning_rate": 1.9523425762798335e-05,
+      "loss": 0.8357,
+      "step": 8142
+    },
+    {
+      "epoch": 0.8144,
+      "learning_rate": 1.952129392742138e-05,
+      "loss": 0.7118,
+      "step": 8144
+    },
+    {
+      "epoch": 0.8146,
+      "learning_rate": 1.9519157451481456e-05,
+      "loss": 0.7181,
+      "step": 8146
+    },
+    {
+      "epoch": 0.8148,
+      "learning_rate": 1.9517016336019824e-05,
+      "loss": 0.8796,
+      "step": 8148
+    },
+    {
+      "epoch": 0.815,
+      "learning_rate": 1.9514870582080035e-05,
+      "loss": 0.4423,
+      "step": 8150
+    },
+    {
+      "epoch": 0.8152,
+      "learning_rate": 1.9512720190707918e-05,
+      "loss": 0.3449,
+      "step": 8152
+    },
+    {
+      "epoch": 0.8154,
+      "learning_rate": 1.9510565162951545e-05,
+      "loss": 0.848,
+      "step": 8154
+    },
+    {
+      "epoch": 0.8156,
+      "learning_rate": 1.950840549986124e-05,
+      "loss": 0.3787,
+      "step": 8156
+    },
+    {
+      "epoch": 0.8158,
+      "learning_rate": 1.95062412024896e-05,
+      "loss": 0.7327,
+      "step": 8158
+    },
+    {
+      "epoch": 0.816,
+      "learning_rate": 1.9504072271891496e-05,
+      "loss": 1.1862,
+      "step": 8160
+    },
+    {
+      "epoch": 0.8162,
+      "learning_rate": 1.950189870912401e-05,
+      "loss": 0.4965,
+      "step": 8162
+    },
+    {
+      "epoch": 0.8164,
+      "learning_rate": 1.949972051524653e-05,
+      "loss": 0.6134,
+      "step": 8164
+    },
+    {
+      "epoch": 0.8166,
+      "learning_rate": 1.9497537691320667e-05,
+      "loss": 0.4291,
+      "step": 8166
+    },
+    {
+      "epoch": 0.8168,
+      "learning_rate": 1.949535023841032e-05,
+      "loss": 0.297,
+      "step": 8168
+    },
+    {
+      "epoch": 0.817,
+      "learning_rate": 1.9493158157581617e-05,
+      "loss": 0.5995,
+      "step": 8170
+    },
+    {
+      "epoch": 0.8172,
+      "learning_rate": 1.9490961449902946e-05,
+      "loss": 0.7799,
+      "step": 8172
+    },
+    {
+      "epoch": 0.8174,
+      "learning_rate": 1.948876011644497e-05,
+      "loss": 1.1717,
+      "step": 8174
+    },
+    {
+      "epoch": 0.8176,
+      "learning_rate": 1.948655415828058e-05,
+      "loss": 0.4617,
+      "step": 8176
+    },
+    {
+      "epoch": 0.8178,
+      "learning_rate": 1.948434357648493e-05,
+      "loss": 1.0519,
+      "step": 8178
+    },
+    {
+      "epoch": 0.818,
+      "learning_rate": 1.9482128372135453e-05,
+      "loss": 0.7344,
+      "step": 8180
+    },
+    {
+      "epoch": 0.8182,
+      "learning_rate": 1.9479908546311787e-05,
+      "loss": 0.9752,
+      "step": 8182
+    },
+    {
+      "epoch": 0.8184,
+      "learning_rate": 1.9477684100095856e-05,
+      "loss": 1.3197,
+      "step": 8184
+    },
+    {
+      "epoch": 0.8186,
+      "learning_rate": 1.9475455034571843e-05,
+      "loss": 0.5623,
+      "step": 8186
+    },
+    {
+      "epoch": 0.8188,
+      "learning_rate": 1.947322135082614e-05,
+      "loss": 0.4345,
+      "step": 8188
+    },
+    {
+      "epoch": 0.819,
+      "learning_rate": 1.9470983049947443e-05,
+      "loss": 0.4468,
+      "step": 8190
+    },
+    {
+      "epoch": 0.8192,
+      "learning_rate": 1.9468740133026665e-05,
+      "loss": 0.1889,
+      "step": 8192
+    },
+    {
+      "epoch": 0.8194,
+      "learning_rate": 1.9466492601156964e-05,
+      "loss": 0.6017,
+      "step": 8194
+    },
+    {
+      "epoch": 0.8196,
+      "learning_rate": 1.9464240455433775e-05,
+      "loss": 0.4727,
+      "step": 8196
+    },
+    {
+      "epoch": 0.8198,
+      "learning_rate": 1.9461983696954767e-05,
+      "loss": 0.4484,
+      "step": 8198
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9459722326819836e-05,
+      "loss": 0.8762,
+      "step": 8200
+    },
+    {
+      "epoch": 0.8202,
+      "learning_rate": 1.9457456346131175e-05,
+      "loss": 0.9365,
+      "step": 8202
+    },
+    {
+      "epoch": 0.8204,
+      "learning_rate": 1.9455185755993177e-05,
+      "loss": 0.2134,
+      "step": 8204
+    },
+    {
+      "epoch": 0.8206,
+      "learning_rate": 1.9452910557512494e-05,
+      "loss": 0.9749,
+      "step": 8206
+    },
+    {
+      "epoch": 0.8208,
+      "learning_rate": 1.9450630751798053e-05,
+      "loss": 0.3651,
+      "step": 8208
+    },
+    {
+      "epoch": 0.821,
+      "learning_rate": 1.9448346339960984e-05,
+      "loss": 0.718,
+      "step": 8210
+    },
+    {
+      "epoch": 0.8212,
+      "learning_rate": 1.944605732311469e-05,
+      "loss": 1.1936,
+      "step": 8212
+    },
+    {
+      "epoch": 0.8214,
+      "learning_rate": 1.9443763702374818e-05,
+      "loss": 0.8166,
+      "step": 8214
+    },
+    {
+      "epoch": 0.8216,
+      "learning_rate": 1.944146547885923e-05,
+      "loss": 0.5868,
+      "step": 8216
+    },
+    {
+      "epoch": 0.8218,
+      "learning_rate": 1.9439162653688066e-05,
+      "loss": 0.5868,
+      "step": 8218
+    },
+    {
+      "epoch": 0.822,
+      "learning_rate": 1.94368552279837e-05,
+      "loss": 0.4875,
+      "step": 8220
+    },
+    {
+      "epoch": 0.8222,
+      "learning_rate": 1.9434543202870726e-05,
+      "loss": 0.9684,
+      "step": 8222
+    },
+    {
+      "epoch": 0.8224,
+      "learning_rate": 1.9432226579476013e-05,
+      "loss": 0.5831,
+      "step": 8224
+    },
+    {
+      "epoch": 0.8226,
+      "learning_rate": 1.9429905358928655e-05,
+      "loss": 0.0884,
+      "step": 8226
+    },
+    {
+      "epoch": 0.8228,
+      "learning_rate": 1.9427579542359966e-05,
+      "loss": 1.0151,
+      "step": 8228
+    },
+    {
+      "epoch": 0.823,
+      "learning_rate": 1.9425249130903544e-05,
+      "loss": 0.9444,
+      "step": 8230
+    },
+    {
+      "epoch": 0.8232,
+      "learning_rate": 1.9422914125695202e-05,
+      "loss": 0.8997,
+      "step": 8232
+    },
+    {
+      "epoch": 0.8234,
+      "learning_rate": 1.942057452787297e-05,
+      "loss": 0.184,
+      "step": 8234
+    },
+    {
+      "epoch": 0.8236,
+      "learning_rate": 1.9418230338577164e-05,
+      "loss": 0.3853,
+      "step": 8236
+    },
+    {
+      "epoch": 0.8238,
+      "learning_rate": 1.94158815589503e-05,
+      "loss": 0.6064,
+      "step": 8238
+    },
+    {
+      "epoch": 0.824,
+      "learning_rate": 1.9413528190137158e-05,
+      "loss": 0.4453,
+      "step": 8240
+    },
+    {
+      "epoch": 0.8242,
+      "learning_rate": 1.941117023328473e-05,
+      "loss": 1.1748,
+      "step": 8242
+    },
+    {
+      "epoch": 0.8244,
+      "learning_rate": 1.9408807689542253e-05,
+      "loss": 0.9002,
+      "step": 8244
+    },
+    {
+      "epoch": 0.8246,
+      "learning_rate": 1.940644056006122e-05,
+      "loss": 0.4639,
+      "step": 8246
+    },
+    {
+      "epoch": 0.8248,
+      "learning_rate": 1.940406884599532e-05,
+      "loss": 0.1476,
+      "step": 8248
+    },
+    {
+      "epoch": 0.825,
+      "learning_rate": 1.94016925485005e-05,
+      "loss": 0.6821,
+      "step": 8250
+    },
+    {
+      "epoch": 0.8252,
+      "learning_rate": 1.939931166873496e-05,
+      "loss": 0.5149,
+      "step": 8252
+    },
+    {
+      "epoch": 0.8254,
+      "learning_rate": 1.939692620785909e-05,
+      "loss": 0.4869,
+      "step": 8254
+    },
+    {
+      "epoch": 0.8256,
+      "learning_rate": 1.939453616703553e-05,
+      "loss": 1.3341,
+      "step": 8256
+    },
+    {
+      "epoch": 0.8258,
+      "learning_rate": 1.939214154742919e-05,
+      "loss": 0.446,
+      "step": 8258
+    },
+    {
+      "epoch": 0.826,
+      "learning_rate": 1.9389742350207137e-05,
+      "loss": 0.948,
+      "step": 8260
+    },
+    {
+      "epoch": 0.8262,
+      "learning_rate": 1.9387338576538746e-05,
+      "loss": 1.303,
+      "step": 8262
+    },
+    {
+      "epoch": 0.8264,
+      "learning_rate": 1.9384930227595564e-05,
+      "loss": 0.528,
+      "step": 8264
+    },
+    {
+      "epoch": 0.8266,
+      "learning_rate": 1.9382517304551393e-05,
+      "loss": 0.8607,
+      "step": 8266
+    },
+    {
+      "epoch": 0.8268,
+      "learning_rate": 1.938009980858228e-05,
+      "loss": 0.6379,
+      "step": 8268
+    },
+    {
+      "epoch": 0.827,
+      "learning_rate": 1.9377677740866464e-05,
+      "loss": 0.6935,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8272,
+      "learning_rate": 1.9375251102584435e-05,
+      "loss": 0.7261,
+      "step": 8272
+    },
+    {
+      "epoch": 0.8274,
+      "learning_rate": 1.9372819894918922e-05,
+      "loss": 1.4702,
+      "step": 8274
+    },
+    {
+      "epoch": 0.8276,
+      "learning_rate": 1.9370384119054848e-05,
+      "loss": 0.5576,
+      "step": 8276
+    },
+    {
+      "epoch": 0.8278,
+      "learning_rate": 1.936794377617938e-05,
+      "loss": 1.2698,
+      "step": 8278
+    },
+    {
+      "epoch": 0.828,
+      "learning_rate": 1.9365498867481933e-05,
+      "loss": 0.9623,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8282,
+      "learning_rate": 1.9363049394154102e-05,
+      "loss": 0.7908,
+      "step": 8282
+    },
+    {
+      "epoch": 0.8284,
+      "learning_rate": 1.9360595357389735e-05,
+      "loss": 1.0491,
+      "step": 8284
+    },
+    {
+      "epoch": 0.8286,
+      "learning_rate": 1.9358136758384917e-05,
+      "loss": 0.8105,
+      "step": 8286
+    },
+    {
+      "epoch": 0.8288,
+      "learning_rate": 1.9355673598337912e-05,
+      "loss": 0.0161,
+      "step": 8288
+    },
+    {
+      "epoch": 0.829,
+      "learning_rate": 1.935320587844926e-05,
+      "loss": 0.8789,
+      "step": 8290
+    },
+    {
+      "epoch": 0.8292,
+      "learning_rate": 1.935073359992169e-05,
+      "loss": 0.4091,
+      "step": 8292
+    },
+    {
+      "epoch": 0.8294,
+      "learning_rate": 1.9348256763960146e-05,
+      "loss": 1.0054,
+      "step": 8294
+    },
+    {
+      "epoch": 0.8296,
+      "learning_rate": 1.9345775371771826e-05,
+      "loss": 0.9319,
+      "step": 8296
+    },
+    {
+      "epoch": 0.8298,
+      "learning_rate": 1.934328942456613e-05,
+      "loss": 1.1199,
+      "step": 8298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9340798923554657e-05,
+      "loss": 0.6293,
+      "step": 8300
+    },
+    {
+      "epoch": 0.8302,
+      "learning_rate": 1.9338303869951273e-05,
+      "loss": 0.6623,
+      "step": 8302
+    },
+    {
+      "epoch": 0.8304,
+      "learning_rate": 1.9335804264972028e-05,
+      "loss": 0.5776,
+      "step": 8304
+    },
+    {
+      "epoch": 0.8306,
+      "learning_rate": 1.9333300109835186e-05,
+      "loss": 0.6192,
+      "step": 8306
+    },
+    {
+      "epoch": 0.8308,
+      "learning_rate": 1.9330791405761257e-05,
+      "loss": 0.5549,
+      "step": 8308
+    },
+    {
+      "epoch": 0.831,
+      "learning_rate": 1.9328278153972943e-05,
+      "loss": 0.4031,
+      "step": 8310
+    },
+    {
+      "epoch": 0.8312,
+      "learning_rate": 1.932576035569519e-05,
+      "loss": 1.0657,
+      "step": 8312
+    },
+    {
+      "epoch": 0.8314,
+      "learning_rate": 1.9323238012155125e-05,
+      "loss": 1.4379,
+      "step": 8314
+    },
+    {
+      "epoch": 0.8316,
+      "learning_rate": 1.9320711124582107e-05,
+      "loss": 0.77,
+      "step": 8316
+    },
+    {
+      "epoch": 0.8318,
+      "learning_rate": 1.931817969420773e-05,
+      "loss": 0.9196,
+      "step": 8318
+    },
+    {
+      "epoch": 0.832,
+      "learning_rate": 1.9315643722265766e-05,
+      "loss": 1.0314,
+      "step": 8320
+    },
+    {
+      "epoch": 0.8322,
+      "learning_rate": 1.93131032099922e-05,
+      "loss": 0.4906,
+      "step": 8322
+    },
+    {
+      "epoch": 0.8324,
+      "learning_rate": 1.9310558158625286e-05,
+      "loss": 0.4606,
+      "step": 8324
+    },
+    {
+      "epoch": 0.8326,
+      "learning_rate": 1.930800856940543e-05,
+      "loss": 0.4994,
+      "step": 8326
+    },
+    {
+      "epoch": 0.8328,
+      "learning_rate": 1.9305454443575265e-05,
+      "loss": 0.2008,
+      "step": 8328
+    },
+    {
+      "epoch": 0.833,
+      "learning_rate": 1.9302895782379648e-05,
+      "loss": 0.264,
+      "step": 8330
+    },
+    {
+      "epoch": 0.8332,
+      "learning_rate": 1.9300332587065637e-05,
+      "loss": 0.1178,
+      "step": 8332
+    },
+    {
+      "epoch": 0.8334,
+      "learning_rate": 1.929776485888252e-05,
+      "loss": 0.5186,
+      "step": 8334
+    },
+    {
+      "epoch": 0.8336,
+      "learning_rate": 1.929519259908175e-05,
+      "loss": 0.4259,
+      "step": 8336
+    },
+    {
+      "epoch": 0.8338,
+      "learning_rate": 1.9292615808917024e-05,
+      "loss": 1.0896,
+      "step": 8338
+    },
+    {
+      "epoch": 0.834,
+      "learning_rate": 1.929003448964425e-05,
+      "loss": 0.3151,
+      "step": 8340
+    },
+    {
+      "epoch": 0.8342,
+      "learning_rate": 1.9287448642521517e-05,
+      "loss": 0.5834,
+      "step": 8342
+    },
+    {
+      "epoch": 0.8344,
+      "learning_rate": 1.9284858268809135e-05,
+      "loss": 0.9797,
+      "step": 8344
+    },
+    {
+      "epoch": 0.8346,
+      "learning_rate": 1.9282263369769637e-05,
+      "loss": 0.5581,
+      "step": 8346
+    },
+    {
+      "epoch": 0.8348,
+      "learning_rate": 1.9279663946667732e-05,
+      "loss": 0.566,
+      "step": 8348
+    },
+    {
+      "epoch": 0.835,
+      "learning_rate": 1.9277060000770342e-05,
+      "loss": 0.7293,
+      "step": 8350
+    },
+    {
+      "epoch": 0.8352,
+      "learning_rate": 1.927445153334662e-05,
+      "loss": 1.0508,
+      "step": 8352
+    },
+    {
+      "epoch": 0.8354,
+      "learning_rate": 1.927183854566788e-05,
+      "loss": 0.0734,
+      "step": 8354
+    },
+    {
+      "epoch": 0.8356,
+      "learning_rate": 1.9269221039007662e-05,
+      "loss": 0.0843,
+      "step": 8356
+    },
+    {
+      "epoch": 0.8358,
+      "learning_rate": 1.9266599014641727e-05,
+      "loss": 0.0417,
+      "step": 8358
+    },
+    {
+      "epoch": 0.836,
+      "learning_rate": 1.9263972473847992e-05,
+      "loss": 0.0311,
+      "step": 8360
+    },
+    {
+      "epoch": 0.8362,
+      "learning_rate": 1.9261341417906622e-05,
+      "loss": 0.6853,
+      "step": 8362
+    },
+    {
+      "epoch": 0.8364,
+      "learning_rate": 1.9258705848099954e-05,
+      "loss": 0.553,
+      "step": 8364
+    },
+    {
+      "epoch": 0.8366,
+      "learning_rate": 1.925606576571252e-05,
+      "loss": 0.6863,
+      "step": 8366
+    },
+    {
+      "epoch": 0.8368,
+      "learning_rate": 1.9253421172031086e-05,
+      "loss": 1.2282,
+      "step": 8368
+    },
+    {
+      "epoch": 0.837,
+      "learning_rate": 1.925077206834459e-05,
+      "loss": 1.073,
+      "step": 8370
+    },
+    {
+      "epoch": 0.8372,
+      "learning_rate": 1.9248118455944153e-05,
+      "loss": 1.1925,
+      "step": 8372
+    },
+    {
+      "epoch": 0.8374,
+      "learning_rate": 1.9245460336123136e-05,
+      "loss": 0.8366,
+      "step": 8374
+    },
+    {
+      "epoch": 0.8376,
+      "learning_rate": 1.9242797710177068e-05,
+      "loss": 0.2722,
+      "step": 8376
+    },
+    {
+      "epoch": 0.8378,
+      "learning_rate": 1.924013057940367e-05,
+      "loss": 0.252,
+      "step": 8378
+    },
+    {
+      "epoch": 0.838,
+      "learning_rate": 1.9237458945102888e-05,
+      "loss": 0.1563,
+      "step": 8380
+    },
+    {
+      "epoch": 0.8382,
+      "learning_rate": 1.923478280857682e-05,
+      "loss": 0.8239,
+      "step": 8382
+    },
+    {
+      "epoch": 0.8384,
+      "learning_rate": 1.923210217112981e-05,
+      "loss": 0.8617,
+      "step": 8384
+    },
+    {
+      "epoch": 0.8386,
+      "learning_rate": 1.922941703406836e-05,
+      "loss": 0.5875,
+      "step": 8386
+    },
+    {
+      "epoch": 0.8388,
+      "learning_rate": 1.922672739870115e-05,
+      "loss": 0.1122,
+      "step": 8388
+    },
+    {
+      "epoch": 0.839,
+      "learning_rate": 1.9224033266339103e-05,
+      "loss": 0.6988,
+      "step": 8390
+    },
+    {
+      "epoch": 0.8392,
+      "learning_rate": 1.9221334638295303e-05,
+      "loss": 0.1031,
+      "step": 8392
+    },
+    {
+      "epoch": 0.8394,
+      "learning_rate": 1.9218631515885007e-05,
+      "loss": 0.1027,
+      "step": 8394
+    },
+    {
+      "epoch": 0.8396,
+      "learning_rate": 1.921592390042571e-05,
+      "loss": 0.8382,
+      "step": 8396
+    },
+    {
+      "epoch": 0.8398,
+      "learning_rate": 1.9213211793237066e-05,
+      "loss": 1.2563,
+      "step": 8398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9210495195640902e-05,
+      "loss": 0.2221,
+      "step": 8400
+    },
+    {
+      "epoch": 0.8402,
+      "learning_rate": 1.9207774108961276e-05,
+      "loss": 0.771,
+      "step": 8402
+    },
+    {
+      "epoch": 0.8404,
+      "learning_rate": 1.9205048534524398e-05,
+      "loss": 0.6143,
+      "step": 8404
+    },
+    {
+      "epoch": 0.8406,
+      "learning_rate": 1.9202318473658707e-05,
+      "loss": 1.1039,
+      "step": 8406
+    },
+    {
+      "epoch": 0.8408,
+      "learning_rate": 1.9199583927694775e-05,
+      "loss": 1.1163,
+      "step": 8408
+    },
+    {
+      "epoch": 0.841,
+      "learning_rate": 1.9196844897965387e-05,
+      "loss": 0.5282,
+      "step": 8410
+    },
+    {
+      "epoch": 0.8412,
+      "learning_rate": 1.9194101385805536e-05,
+      "loss": 0.9093,
+      "step": 8412
+    },
+    {
+      "epoch": 0.8414,
+      "learning_rate": 1.919135339255235e-05,
+      "loss": 0.5538,
+      "step": 8414
+    },
+    {
+      "epoch": 0.8416,
+      "learning_rate": 1.918860091954517e-05,
+      "loss": 0.6432,
+      "step": 8416
+    },
+    {
+      "epoch": 0.8418,
+      "learning_rate": 1.9185843968125546e-05,
+      "loss": 0.1527,
+      "step": 8418
+    },
+    {
+      "epoch": 0.842,
+      "learning_rate": 1.9183082539637152e-05,
+      "loss": 1.0098,
+      "step": 8420
+    },
+    {
+      "epoch": 0.8422,
+      "learning_rate": 1.918031663542588e-05,
+      "loss": 1.3172,
+      "step": 8422
+    },
+    {
+      "epoch": 0.8424,
+      "learning_rate": 1.9177546256839817e-05,
+      "loss": 0.8535,
+      "step": 8424
+    },
+    {
+      "epoch": 0.8426,
+      "learning_rate": 1.917477140522919e-05,
+      "loss": 0.721,
+      "step": 8426
+    },
+    {
+      "epoch": 0.8428,
+      "learning_rate": 1.9171992081946433e-05,
+      "loss": 1.3764,
+      "step": 8428
+    },
+    {
+      "epoch": 0.843,
+      "learning_rate": 1.916920828834617e-05,
+      "loss": 2.1118,
+      "step": 8430
+    },
+    {
+      "epoch": 0.8432,
+      "learning_rate": 1.916642002578516e-05,
+      "loss": 0.2692,
+      "step": 8432
+    },
+    {
+      "epoch": 0.8434,
+      "learning_rate": 1.9163627295622394e-05,
+      "loss": 0.5626,
+      "step": 8434
+    },
+    {
+      "epoch": 0.8436,
+      "learning_rate": 1.9160830099219014e-05,
+      "loss": 0.3413,
+      "step": 8436
+    },
+    {
+      "epoch": 0.8438,
+      "learning_rate": 1.9158028437938313e-05,
+      "loss": 0.74,
+      "step": 8438
+    },
+    {
+      "epoch": 0.844,
+      "learning_rate": 1.915522231314582e-05,
+      "loss": 0.8584,
+      "step": 8440
+    },
+    {
+      "epoch": 0.8442,
+      "learning_rate": 1.9152411726209183e-05,
+      "loss": 1.1133,
+      "step": 8442
+    },
+    {
+      "epoch": 0.8444,
+      "learning_rate": 1.914959667849825e-05,
+      "loss": 0.3382,
+      "step": 8444
+    },
+    {
+      "epoch": 0.8446,
+      "learning_rate": 1.9146777171385057e-05,
+      "loss": 1.0629,
+      "step": 8446
+    },
+    {
+      "epoch": 0.8448,
+      "learning_rate": 1.914395320624378e-05,
+      "loss": 0.3939,
+      "step": 8448
+    },
+    {
+      "epoch": 0.845,
+      "learning_rate": 1.914112478445079e-05,
+      "loss": 0.4972,
+      "step": 8450
+    },
+    {
+      "epoch": 0.8452,
+      "learning_rate": 1.913829190738464e-05,
+      "loss": 0.8952,
+      "step": 8452
+    },
+    {
+      "epoch": 0.8454,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.7359,
+      "step": 8454
+    },
+    {
+      "epoch": 0.8456,
+      "learning_rate": 1.9132612792957808e-05,
+      "loss": 0.3491,
+      "step": 8456
+    },
+    {
+      "epoch": 0.8458,
+      "learning_rate": 1.9129766558365082e-05,
+      "loss": 0.5619,
+      "step": 8458
+    },
+    {
+      "epoch": 0.846,
+      "learning_rate": 1.912691587403503e-05,
+      "loss": 0.2067,
+      "step": 8460
+    },
+    {
+      "epoch": 0.8462,
+      "learning_rate": 1.9124060741357065e-05,
+      "loss": 0.8542,
+      "step": 8462
+    },
+    {
+      "epoch": 0.8464,
+      "learning_rate": 1.912120116172274e-05,
+      "loss": 0.3124,
+      "step": 8464
+    },
+    {
+      "epoch": 0.8466,
+      "learning_rate": 1.911833713652576e-05,
+      "loss": 0.1271,
+      "step": 8466
+    },
+    {
+      "epoch": 0.8468,
+      "learning_rate": 1.9115468667162038e-05,
+      "loss": 1.0833,
+      "step": 8468
+    },
+    {
+      "epoch": 0.847,
+      "learning_rate": 1.911259575502963e-05,
+      "loss": 0.9207,
+      "step": 8470
+    },
+    {
+      "epoch": 0.8472,
+      "learning_rate": 1.9109718401528742e-05,
+      "loss": 0.2435,
+      "step": 8472
+    },
+    {
+      "epoch": 0.8474,
+      "learning_rate": 1.9106836608061775e-05,
+      "loss": 0.8342,
+      "step": 8474
+    },
+    {
+      "epoch": 0.8476,
+      "learning_rate": 1.910395037603327e-05,
+      "loss": 0.6172,
+      "step": 8476
+    },
+    {
+      "epoch": 0.8478,
+      "learning_rate": 1.910105970684996e-05,
+      "loss": 0.9112,
+      "step": 8478
+    },
+    {
+      "epoch": 0.848,
+      "learning_rate": 1.9098164601920706e-05,
+      "loss": 1.1939,
+      "step": 8480
+    },
+    {
+      "epoch": 0.8482,
+      "learning_rate": 1.909526506265654e-05,
+      "loss": 0.2814,
+      "step": 8482
+    },
+    {
+      "epoch": 0.8484,
+      "learning_rate": 1.9092361090470688e-05,
+      "loss": 0.3382,
+      "step": 8484
+    },
+    {
+      "epoch": 0.8486,
+      "learning_rate": 1.908945268677849e-05,
+      "loss": 0.7439,
+      "step": 8486
+    },
+    {
+      "epoch": 0.8488,
+      "learning_rate": 1.9086539852997466e-05,
+      "loss": 0.3167,
+      "step": 8488
+    },
+    {
+      "epoch": 0.849,
+      "learning_rate": 1.9083622590547313e-05,
+      "loss": 0.5457,
+      "step": 8490
+    },
+    {
+      "epoch": 0.8492,
+      "learning_rate": 1.9080700900849858e-05,
+      "loss": 1.377,
+      "step": 8492
+    },
+    {
+      "epoch": 0.8494,
+      "learning_rate": 1.9077774785329085e-05,
+      "loss": 1.0325,
+      "step": 8494
+    },
+    {
+      "epoch": 0.8496,
+      "learning_rate": 1.9074844245411174e-05,
+      "loss": 0.9749,
+      "step": 8496
+    },
+    {
+      "epoch": 0.8498,
+      "learning_rate": 1.9071909282524422e-05,
+      "loss": 1.3956,
+      "step": 8498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9068969898099278e-05,
+      "loss": 0.4294,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8502,
+      "learning_rate": 1.9066026093568383e-05,
+      "loss": 1.0925,
+      "step": 8502
+    },
+    {
+      "epoch": 0.8504,
+      "learning_rate": 1.9063077870366497e-05,
+      "loss": 0.6431,
+      "step": 8504
+    },
+    {
+      "epoch": 0.8506,
+      "learning_rate": 1.9060125229930576e-05,
+      "loss": 1.0403,
+      "step": 8506
+    },
+    {
+      "epoch": 0.8508,
+      "learning_rate": 1.9057168173699668e-05,
+      "loss": 0.3363,
+      "step": 8508
+    },
+    {
+      "epoch": 0.851,
+      "learning_rate": 1.9054206703115013e-05,
+      "loss": 0.702,
+      "step": 8510
+    },
+    {
+      "epoch": 0.8512,
+      "learning_rate": 1.905124081962002e-05,
+      "loss": 0.52,
+      "step": 8512
+    },
+    {
+      "epoch": 0.8514,
+      "learning_rate": 1.9048270524660203e-05,
+      "loss": 0.4461,
+      "step": 8514
+    },
+    {
+      "epoch": 0.8516,
+      "learning_rate": 1.904529581968324e-05,
+      "loss": 0.9659,
+      "step": 8516
+    },
+    {
+      "epoch": 0.8518,
+      "learning_rate": 1.9042316706138994e-05,
+      "loss": 1.0731,
+      "step": 8518
+    },
+    {
+      "epoch": 0.852,
+      "learning_rate": 1.9039333185479425e-05,
+      "loss": 0.6525,
+      "step": 8520
+    },
+    {
+      "epoch": 0.8522,
+      "learning_rate": 1.9036345259158664e-05,
+      "loss": 1.0049,
+      "step": 8522
+    },
+    {
+      "epoch": 0.8524,
+      "learning_rate": 1.9033352928633017e-05,
+      "loss": 0.9404,
+      "step": 8524
+    },
+    {
+      "epoch": 0.8526,
+      "learning_rate": 1.903035619536087e-05,
+      "loss": 1.0136,
+      "step": 8526
+    },
+    {
+      "epoch": 0.8528,
+      "learning_rate": 1.902735506080283e-05,
+      "loss": 0.2231,
+      "step": 8528
+    },
+    {
+      "epoch": 0.853,
+      "learning_rate": 1.9024349526421603e-05,
+      "loss": 0.8896,
+      "step": 8530
+    },
+    {
+      "epoch": 0.8532,
+      "learning_rate": 1.902133959368203e-05,
+      "loss": 1.0946,
+      "step": 8532
+    },
+    {
+      "epoch": 0.8534,
+      "learning_rate": 1.901832526405114e-05,
+      "loss": 0.5099,
+      "step": 8534
+    },
+    {
+      "epoch": 0.8536,
+      "learning_rate": 1.901530653899808e-05,
+      "loss": 0.4695,
+      "step": 8536
+    },
+    {
+      "epoch": 0.8538,
+      "learning_rate": 1.9012283419994112e-05,
+      "loss": 0.7365,
+      "step": 8538
+    },
+    {
+      "epoch": 0.854,
+      "learning_rate": 1.9009255908512704e-05,
+      "loss": 0.1576,
+      "step": 8540
+    },
+    {
+      "epoch": 0.8542,
+      "learning_rate": 1.9006224006029414e-05,
+      "loss": 0.555,
+      "step": 8542
+    },
+    {
+      "epoch": 0.8544,
+      "learning_rate": 1.9003187714021936e-05,
+      "loss": 0.476,
+      "step": 8544
+    },
+    {
+      "epoch": 0.8546,
+      "learning_rate": 1.9000147033970148e-05,
+      "loss": 0.1571,
+      "step": 8546
+    },
+    {
+      "epoch": 0.8548,
+      "learning_rate": 1.8997101967356038e-05,
+      "loss": 0.6073,
+      "step": 8548
+    },
+    {
+      "epoch": 0.855,
+      "learning_rate": 1.899405251566371e-05,
+      "loss": 0.4764,
+      "step": 8550
+    },
+    {
+      "epoch": 0.8552,
+      "learning_rate": 1.8990998680379458e-05,
+      "loss": 0.4964,
+      "step": 8552
+    },
+    {
+      "epoch": 0.8554,
+      "learning_rate": 1.8987940462991666e-05,
+      "loss": 1.2349,
+      "step": 8554
+    },
+    {
+      "epoch": 0.8556,
+      "learning_rate": 1.898487786499089e-05,
+      "loss": 0.6576,
+      "step": 8556
+    },
+    {
+      "epoch": 0.8558,
+      "learning_rate": 1.8981810887869797e-05,
+      "loss": 0.4509,
+      "step": 8558
+    },
+    {
+      "epoch": 0.856,
+      "learning_rate": 1.897873953312317e-05,
+      "loss": 0.4632,
+      "step": 8560
+    },
+    {
+      "epoch": 0.8562,
+      "learning_rate": 1.8975663802247978e-05,
+      "loss": 0.317,
+      "step": 8562
+    },
+    {
+      "epoch": 0.8564,
+      "learning_rate": 1.8972583696743297e-05,
+      "loss": 1.2229,
+      "step": 8564
+    },
+    {
+      "epoch": 0.8566,
+      "learning_rate": 1.8969499218110302e-05,
+      "loss": 0.8986,
+      "step": 8566
+    },
+    {
+      "epoch": 0.8568,
+      "learning_rate": 1.896641036785236e-05,
+      "loss": 1.3357,
+      "step": 8568
+    },
+    {
+      "epoch": 0.857,
+      "learning_rate": 1.8963317147474943e-05,
+      "loss": 0.3556,
+      "step": 8570
+    },
+    {
+      "epoch": 0.8572,
+      "learning_rate": 1.8960219558485614e-05,
+      "loss": 0.8141,
+      "step": 8572
+    },
+    {
+      "epoch": 0.8574,
+      "learning_rate": 1.8957117602394133e-05,
+      "loss": 0.2506,
+      "step": 8574
+    },
+    {
+      "epoch": 0.8576,
+      "learning_rate": 1.8954011280712338e-05,
+      "loss": 0.3882,
+      "step": 8576
+    },
+    {
+      "epoch": 0.8578,
+      "learning_rate": 1.8950900594954233e-05,
+      "loss": 0.9608,
+      "step": 8578
+    },
+    {
+      "epoch": 0.858,
+      "learning_rate": 1.8947785546635905e-05,
+      "loss": 0.1857,
+      "step": 8580
+    },
+    {
+      "epoch": 0.8582,
+      "learning_rate": 1.8944666137275596e-05,
+      "loss": 0.3584,
+      "step": 8582
+    },
+    {
+      "epoch": 0.8584,
+      "learning_rate": 1.8941542368393687e-05,
+      "loss": 0.6608,
+      "step": 8584
+    },
+    {
+      "epoch": 0.8586,
+      "learning_rate": 1.8938414241512644e-05,
+      "loss": 0.4682,
+      "step": 8586
+    },
+    {
+      "epoch": 0.8588,
+      "learning_rate": 1.8935281758157076e-05,
+      "loss": 0.2902,
+      "step": 8588
+    },
+    {
+      "epoch": 0.859,
+      "learning_rate": 1.8932144919853744e-05,
+      "loss": 0.779,
+      "step": 8590
+    },
+    {
+      "epoch": 0.8592,
+      "learning_rate": 1.8929003728131476e-05,
+      "loss": 0.4891,
+      "step": 8592
+    },
+    {
+      "epoch": 0.8594,
+      "learning_rate": 1.892585818452125e-05,
+      "loss": 0.5317,
+      "step": 8594
+    },
+    {
+      "epoch": 0.8596,
+      "learning_rate": 1.8922708290556204e-05,
+      "loss": 1.0967,
+      "step": 8596
+    },
+    {
+      "epoch": 0.8598,
+      "learning_rate": 1.8919554047771508e-05,
+      "loss": 0.1062,
+      "step": 8598
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8916395457704536e-05,
+      "loss": 0.9414,
+      "step": 8600
+    },
+    {
+      "epoch": 0.8602,
+      "learning_rate": 1.891323252189474e-05,
+      "loss": 1.9252,
+      "step": 8602
+    },
+    {
+      "epoch": 0.8604,
+      "learning_rate": 1.891006524188368e-05,
+      "loss": 0.7091,
+      "step": 8604
+    },
+    {
+      "epoch": 0.8606,
+      "learning_rate": 1.890689361921507e-05,
+      "loss": 0.9156,
+      "step": 8606
+    },
+    {
+      "epoch": 0.8608,
+      "learning_rate": 1.8903717655434714e-05,
+      "loss": 0.3174,
+      "step": 8608
+    },
+    {
+      "epoch": 0.861,
+      "learning_rate": 1.8900537352090523e-05,
+      "loss": 0.6841,
+      "step": 8610
+    },
+    {
+      "epoch": 0.8612,
+      "learning_rate": 1.8897352710732564e-05,
+      "loss": 0.9994,
+      "step": 8612
+    },
+    {
+      "epoch": 0.8614,
+      "learning_rate": 1.8894163732912986e-05,
+      "loss": 1.0837,
+      "step": 8614
+    },
+    {
+      "epoch": 0.8616,
+      "learning_rate": 1.8890970420186035e-05,
+      "loss": 0.9725,
+      "step": 8616
+    },
+    {
+      "epoch": 0.8618,
+      "learning_rate": 1.8887772774108122e-05,
+      "loss": 1.1093,
+      "step": 8618
+    },
+    {
+      "epoch": 0.862,
+      "learning_rate": 1.888457079623773e-05,
+      "loss": 0.1272,
+      "step": 8620
+    },
+    {
+      "epoch": 0.8622,
+      "learning_rate": 1.8881364488135445e-05,
+      "loss": 0.8135,
+      "step": 8622
+    },
+    {
+      "epoch": 0.8624,
+      "learning_rate": 1.8878153851364023e-05,
+      "loss": 0.8435,
+      "step": 8624
+    },
+    {
+      "epoch": 0.8626,
+      "learning_rate": 1.887493888748825e-05,
+      "loss": 0.8139,
+      "step": 8626
+    },
+    {
+      "epoch": 0.8628,
+      "learning_rate": 1.8871719598075083e-05,
+      "loss": 0.7735,
+      "step": 8628
+    },
+    {
+      "epoch": 0.863,
+      "learning_rate": 1.886849598469357e-05,
+      "loss": 1.1336,
+      "step": 8630
+    },
+    {
+      "epoch": 0.8632,
+      "learning_rate": 1.886526804891483e-05,
+      "loss": 0.6089,
+      "step": 8632
+    },
+    {
+      "epoch": 0.8634,
+      "learning_rate": 1.886203579231215e-05,
+      "loss": 0.6253,
+      "step": 8634
+    },
+    {
+      "epoch": 0.8636,
+      "learning_rate": 1.885879921646089e-05,
+      "loss": 0.6816,
+      "step": 8636
+    },
+    {
+      "epoch": 0.8638,
+      "learning_rate": 1.8855558322938492e-05,
+      "loss": 1.0581,
+      "step": 8638
+    },
+    {
+      "epoch": 0.864,
+      "learning_rate": 1.8852313113324556e-05,
+      "loss": 0.4678,
+      "step": 8640
+    },
+    {
+      "epoch": 0.8642,
+      "learning_rate": 1.8849063589200754e-05,
+      "loss": 0.7491,
+      "step": 8642
+    },
+    {
+      "epoch": 0.8644,
+      "learning_rate": 1.8845809752150844e-05,
+      "loss": 0.8111,
+      "step": 8644
+    },
+    {
+      "epoch": 0.8646,
+      "learning_rate": 1.8842551603760725e-05,
+      "loss": 0.5784,
+      "step": 8646
+    },
+    {
+      "epoch": 0.8648,
+      "learning_rate": 1.883928914561837e-05,
+      "loss": 0.5554,
+      "step": 8648
+    },
+    {
+      "epoch": 0.865,
+      "learning_rate": 1.8836022379313884e-05,
+      "loss": 0.2166,
+      "step": 8650
+    },
+    {
+      "epoch": 0.8652,
+      "learning_rate": 1.8832751306439423e-05,
+      "loss": 0.7569,
+      "step": 8652
+    },
+    {
+      "epoch": 0.8654,
+      "learning_rate": 1.8829475928589265e-05,
+      "loss": 0.9112,
+      "step": 8654
+    },
+    {
+      "epoch": 0.8656,
+      "learning_rate": 1.882619624735982e-05,
+      "loss": 0.7468,
+      "step": 8656
+    },
+    {
+      "epoch": 0.8658,
+      "learning_rate": 1.882291226434954e-05,
+      "loss": 0.9445,
+      "step": 8658
+    },
+    {
+      "epoch": 0.866,
+      "learning_rate": 1.8819623981158993e-05,
+      "loss": 0.425,
+      "step": 8660
+    },
+    {
+      "epoch": 0.8662,
+      "learning_rate": 1.8816331399390874e-05,
+      "loss": 0.5927,
+      "step": 8662
+    },
+    {
+      "epoch": 0.8664,
+      "learning_rate": 1.8813034520649927e-05,
+      "loss": 0.8845,
+      "step": 8664
+    },
+    {
+      "epoch": 0.8666,
+      "learning_rate": 1.880973334654301e-05,
+      "loss": 0.6883,
+      "step": 8666
+    },
+    {
+      "epoch": 0.8668,
+      "learning_rate": 1.8806427878679098e-05,
+      "loss": 0.5781,
+      "step": 8668
+    },
+    {
+      "epoch": 0.867,
+      "learning_rate": 1.88031181186692e-05,
+      "loss": 0.112,
+      "step": 8670
+    },
+    {
+      "epoch": 0.8672,
+      "learning_rate": 1.8799804068126484e-05,
+      "loss": 0.8523,
+      "step": 8672
+    },
+    {
+      "epoch": 0.8674,
+      "learning_rate": 1.8796485728666172e-05,
+      "loss": 0.6605,
+      "step": 8674
+    },
+    {
+      "epoch": 0.8676,
+      "learning_rate": 1.8793163101905562e-05,
+      "loss": 0.4703,
+      "step": 8676
+    },
+    {
+      "epoch": 0.8678,
+      "learning_rate": 1.8789836189464092e-05,
+      "loss": 0.5764,
+      "step": 8678
+    },
+    {
+      "epoch": 0.868,
+      "learning_rate": 1.8786504992963234e-05,
+      "loss": 0.5967,
+      "step": 8680
+    },
+    {
+      "epoch": 0.8682,
+      "learning_rate": 1.8783169514026574e-05,
+      "loss": 0.4062,
+      "step": 8682
+    },
+    {
+      "epoch": 0.8684,
+      "learning_rate": 1.8779829754279816e-05,
+      "loss": 0.1906,
+      "step": 8684
+    },
+    {
+      "epoch": 0.8686,
+      "learning_rate": 1.877648571535068e-05,
+      "loss": 0.9464,
+      "step": 8686
+    },
+    {
+      "epoch": 0.8688,
+      "learning_rate": 1.8773137398869014e-05,
+      "loss": 0.5276,
+      "step": 8688
+    },
+    {
+      "epoch": 0.869,
+      "learning_rate": 1.8769784806466775e-05,
+      "loss": 0.4302,
+      "step": 8690
+    },
+    {
+      "epoch": 0.8692,
+      "learning_rate": 1.876642793977795e-05,
+      "loss": 1.1211,
+      "step": 8692
+    },
+    {
+      "epoch": 0.8694,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 1.0627,
+      "step": 8694
+    },
+    {
+      "epoch": 0.8696,
+      "learning_rate": 1.8759701390087036e-05,
+      "loss": 0.6615,
+      "step": 8696
+    },
+    {
+      "epoch": 0.8698,
+      "learning_rate": 1.8756331710363375e-05,
+      "loss": 0.7701,
+      "step": 8698
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8752957762910016e-05,
+      "loss": 1.0227,
+      "step": 8700
+    },
+    {
+      "epoch": 0.8702,
+      "learning_rate": 1.8749579549371387e-05,
+      "loss": 1.1256,
+      "step": 8702
+    },
+    {
+      "epoch": 0.8704,
+      "learning_rate": 1.874619707139396e-05,
+      "loss": 0.8064,
+      "step": 8704
+    },
+    {
+      "epoch": 0.8706,
+      "learning_rate": 1.8742810330626338e-05,
+      "loss": 0.6084,
+      "step": 8706
+    },
+    {
+      "epoch": 0.8708,
+      "learning_rate": 1.873941932871918e-05,
+      "loss": 0.6417,
+      "step": 8708
+    },
+    {
+      "epoch": 0.871,
+      "learning_rate": 1.8736024067325188e-05,
+      "loss": 0.4154,
+      "step": 8710
+    },
+    {
+      "epoch": 0.8712,
+      "learning_rate": 1.8732624548099204e-05,
+      "loss": 0.7318,
+      "step": 8712
+    },
+    {
+      "epoch": 0.8714,
+      "learning_rate": 1.8729220772698106e-05,
+      "loss": 0.8068,
+      "step": 8714
+    },
+    {
+      "epoch": 0.8716,
+      "learning_rate": 1.8725812742780832e-05,
+      "loss": 1.092,
+      "step": 8716
+    },
+    {
+      "epoch": 0.8718,
+      "learning_rate": 1.8722400460008437e-05,
+      "loss": 0.7467,
+      "step": 8718
+    },
+    {
+      "epoch": 0.872,
+      "learning_rate": 1.8718983926044013e-05,
+      "loss": 1.2269,
+      "step": 8720
+    },
+    {
+      "epoch": 0.8722,
+      "learning_rate": 1.8715563142552758e-05,
+      "loss": 0.0771,
+      "step": 8722
+    },
+    {
+      "epoch": 0.8724,
+      "learning_rate": 1.8712138111201898e-05,
+      "loss": 0.0866,
+      "step": 8724
+    },
+    {
+      "epoch": 0.8726,
+      "learning_rate": 1.8708708833660748e-05,
+      "loss": 0.0937,
+      "step": 8726
+    },
+    {
+      "epoch": 0.8728,
+      "learning_rate": 1.8705275311600724e-05,
+      "loss": 0.5967,
+      "step": 8728
+    },
+    {
+      "epoch": 0.873,
+      "learning_rate": 1.870183754669526e-05,
+      "loss": 0.9762,
+      "step": 8730
+    },
+    {
+      "epoch": 0.8732,
+      "learning_rate": 1.8698395540619876e-05,
+      "loss": 1.1437,
+      "step": 8732
+    },
+    {
+      "epoch": 0.8734,
+      "learning_rate": 1.8694949295052198e-05,
+      "loss": 0.4451,
+      "step": 8734
+    },
+    {
+      "epoch": 0.8736,
+      "learning_rate": 1.8691498811671853e-05,
+      "loss": 0.401,
+      "step": 8736
+    },
+    {
+      "epoch": 0.8738,
+      "learning_rate": 1.8688044092160558e-05,
+      "loss": 0.9259,
+      "step": 8738
+    },
+    {
+      "epoch": 0.874,
+      "learning_rate": 1.8684585138202126e-05,
+      "loss": 1.3053,
+      "step": 8740
+    },
+    {
+      "epoch": 0.8742,
+      "learning_rate": 1.868112195148239e-05,
+      "loss": 1.1944,
+      "step": 8742
+    },
+    {
+      "epoch": 0.8744,
+      "learning_rate": 1.867765453368929e-05,
+      "loss": 0.6108,
+      "step": 8744
+    },
+    {
+      "epoch": 0.8746,
+      "learning_rate": 1.867418288651278e-05,
+      "loss": 0.1934,
+      "step": 8746
+    },
+    {
+      "epoch": 0.8748,
+      "learning_rate": 1.8670707011644897e-05,
+      "loss": 0.4944,
+      "step": 8748
+    },
+    {
+      "epoch": 0.875,
+      "learning_rate": 1.866722691077977e-05,
+      "loss": 0.368,
+      "step": 8750
+    },
+    {
+      "epoch": 0.8752,
+      "learning_rate": 1.8663742585613525e-05,
+      "loss": 1.1264,
+      "step": 8752
+    },
+    {
+      "epoch": 0.8754,
+      "learning_rate": 1.8660254037844384e-05,
+      "loss": 0.5572,
+      "step": 8754
+    },
+    {
+      "epoch": 0.8756,
+      "learning_rate": 1.8656761269172648e-05,
+      "loss": 0.6092,
+      "step": 8756
+    },
+    {
+      "epoch": 0.8758,
+      "learning_rate": 1.8653264281300626e-05,
+      "loss": 0.1512,
+      "step": 8758
+    },
+    {
+      "epoch": 0.876,
+      "learning_rate": 1.8649763075932707e-05,
+      "loss": 0.5327,
+      "step": 8760
+    },
+    {
+      "epoch": 0.8762,
+      "learning_rate": 1.8646257654775357e-05,
+      "loss": 0.264,
+      "step": 8762
+    },
+    {
+      "epoch": 0.8764,
+      "learning_rate": 1.8642748019537054e-05,
+      "loss": 0.5633,
+      "step": 8764
+    },
+    {
+      "epoch": 0.8766,
+      "learning_rate": 1.8639234171928355e-05,
+      "loss": 0.806,
+      "step": 8766
+    },
+    {
+      "epoch": 0.8768,
+      "learning_rate": 1.8635716113661883e-05,
+      "loss": 0.5414,
+      "step": 8768
+    },
+    {
+      "epoch": 0.877,
+      "learning_rate": 1.8632193846452267e-05,
+      "loss": 1.1001,
+      "step": 8770
+    },
+    {
+      "epoch": 0.8772,
+      "learning_rate": 1.862866737201625e-05,
+      "loss": 0.8286,
+      "step": 8772
+    },
+    {
+      "epoch": 0.8774,
+      "learning_rate": 1.8625136692072587e-05,
+      "loss": 0.075,
+      "step": 8774
+    },
+    {
+      "epoch": 0.8776,
+      "learning_rate": 1.8621601808342055e-05,
+      "loss": 0.2887,
+      "step": 8776
+    },
+    {
+      "epoch": 0.8778,
+      "learning_rate": 1.861806272254755e-05,
+      "loss": 0.6895,
+      "step": 8778
+    },
+    {
+      "epoch": 0.878,
+      "learning_rate": 1.8614519436413978e-05,
+      "loss": 0.6322,
+      "step": 8780
+    },
+    {
+      "epoch": 0.8782,
+      "learning_rate": 1.8610971951668265e-05,
+      "loss": 1.0418,
+      "step": 8782
+    },
+    {
+      "epoch": 0.8784,
+      "learning_rate": 1.860742027003944e-05,
+      "loss": 0.2534,
+      "step": 8784
+    },
+    {
+      "epoch": 0.8786,
+      "learning_rate": 1.8603864393258547e-05,
+      "loss": 1.0955,
+      "step": 8786
+    },
+    {
+      "epoch": 0.8788,
+      "learning_rate": 1.860030432305865e-05,
+      "loss": 0.3021,
+      "step": 8788
+    },
+    {
+      "epoch": 0.879,
+      "learning_rate": 1.8596740061174912e-05,
+      "loss": 1.1942,
+      "step": 8790
+    },
+    {
+      "epoch": 0.8792,
+      "learning_rate": 1.8593171609344495e-05,
+      "loss": 0.6993,
+      "step": 8792
+    },
+    {
+      "epoch": 0.8794,
+      "learning_rate": 1.8589598969306646e-05,
+      "loss": 0.726,
+      "step": 8794
+    },
+    {
+      "epoch": 0.8796,
+      "learning_rate": 1.8586022142802607e-05,
+      "loss": 0.4703,
+      "step": 8796
+    },
+    {
+      "epoch": 0.8798,
+      "learning_rate": 1.858244113157566e-05,
+      "loss": 0.48,
+      "step": 8798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8578855937371176e-05,
+      "loss": 0.0335,
+      "step": 8800
+    },
+    {
+      "epoch": 0.8802,
+      "learning_rate": 1.8575266561936533e-05,
+      "loss": 0.5559,
+      "step": 8802
+    },
+    {
+      "epoch": 0.8804,
+      "learning_rate": 1.8571673007021127e-05,
+      "loss": 0.3489,
+      "step": 8804
+    },
+    {
+      "epoch": 0.8806,
+      "learning_rate": 1.8568075274376432e-05,
+      "loss": 0.8218,
+      "step": 8806
+    },
+    {
+      "epoch": 0.8808,
+      "learning_rate": 1.8564473365755946e-05,
+      "loss": 0.1821,
+      "step": 8808
+    },
+    {
+      "epoch": 0.881,
+      "learning_rate": 1.8560867282915164e-05,
+      "loss": 0.7198,
+      "step": 8810
+    },
+    {
+      "epoch": 0.8812,
+      "learning_rate": 1.8557257027611677e-05,
+      "loss": 0.5303,
+      "step": 8812
+    },
+    {
+      "epoch": 0.8814,
+      "learning_rate": 1.8553642601605083e-05,
+      "loss": 1.0382,
+      "step": 8814
+    },
+    {
+      "epoch": 0.8816,
+      "learning_rate": 1.855002400665697e-05,
+      "loss": 1.2194,
+      "step": 8816
+    },
+    {
+      "epoch": 0.8818,
+      "learning_rate": 1.8546401244531034e-05,
+      "loss": 0.2434,
+      "step": 8818
+    },
+    {
+      "epoch": 0.882,
+      "learning_rate": 1.854277431699295e-05,
+      "loss": 0.8725,
+      "step": 8820
+    },
+    {
+      "epoch": 0.8822,
+      "learning_rate": 1.8539143225810457e-05,
+      "loss": 0.1422,
+      "step": 8822
+    },
+    {
+      "epoch": 0.8824,
+      "learning_rate": 1.853550797275328e-05,
+      "loss": 0.0128,
+      "step": 8824
+    },
+    {
+      "epoch": 0.8826,
+      "learning_rate": 1.85318685595932e-05,
+      "loss": 1.3037,
+      "step": 8826
+    },
+    {
+      "epoch": 0.8828,
+      "learning_rate": 1.8528224988104047e-05,
+      "loss": 0.4363,
+      "step": 8828
+    },
+    {
+      "epoch": 0.883,
+      "learning_rate": 1.852457726006163e-05,
+      "loss": 0.6935,
+      "step": 8830
+    },
+    {
+      "epoch": 0.8832,
+      "learning_rate": 1.8520925377243805e-05,
+      "loss": 1.1849,
+      "step": 8832
+    },
+    {
+      "epoch": 0.8834,
+      "learning_rate": 1.8517269341430485e-05,
+      "loss": 2.5229,
+      "step": 8834
+    },
+    {
+      "epoch": 0.8836,
+      "learning_rate": 1.8513609154403545e-05,
+      "loss": 0.4796,
+      "step": 8836
+    },
+    {
+      "epoch": 0.8838,
+      "learning_rate": 1.8509944817946917e-05,
+      "loss": 0.6218,
+      "step": 8838
+    },
+    {
+      "epoch": 0.884,
+      "learning_rate": 1.8506276333846585e-05,
+      "loss": 0.9901,
+      "step": 8840
+    },
+    {
+      "epoch": 0.8842,
+      "learning_rate": 1.8502603703890484e-05,
+      "loss": 2.3947,
+      "step": 8842
+    },
+    {
+      "epoch": 0.8844,
+      "learning_rate": 1.8498926929868637e-05,
+      "loss": 1.1616,
+      "step": 8844
+    },
+    {
+      "epoch": 0.8846,
+      "learning_rate": 1.8495246013573064e-05,
+      "loss": 0.502,
+      "step": 8846
+    },
+    {
+      "epoch": 0.8848,
+      "learning_rate": 1.849156095679776e-05,
+      "loss": 0.8989,
+      "step": 8848
+    },
+    {
+      "epoch": 0.885,
+      "learning_rate": 1.8487871761338817e-05,
+      "loss": 0.6784,
+      "step": 8850
+    },
+    {
+      "epoch": 0.8852,
+      "learning_rate": 1.8484178428994298e-05,
+      "loss": 0.5814,
+      "step": 8852
+    },
+    {
+      "epoch": 0.8854,
+      "learning_rate": 1.848048096156426e-05,
+      "loss": 0.1307,
+      "step": 8854
+    },
+    {
+      "epoch": 0.8856,
+      "learning_rate": 1.8476779360850843e-05,
+      "loss": 0.9223,
+      "step": 8856
+    },
+    {
+      "epoch": 0.8858,
+      "learning_rate": 1.847307362865813e-05,
+      "loss": 0.0137,
+      "step": 8858
+    },
+    {
+      "epoch": 0.886,
+      "learning_rate": 1.8469363766792258e-05,
+      "loss": 0.3377,
+      "step": 8860
+    },
+    {
+      "epoch": 0.8862,
+      "learning_rate": 1.8465649777061387e-05,
+      "loss": 0.7499,
+      "step": 8862
+    },
+    {
+      "epoch": 0.8864,
+      "learning_rate": 1.8461931661275642e-05,
+      "loss": 0.4758,
+      "step": 8864
+    },
+    {
+      "epoch": 0.8866,
+      "learning_rate": 1.8458209421247208e-05,
+      "loss": 0.5791,
+      "step": 8866
+    },
+    {
+      "epoch": 0.8868,
+      "learning_rate": 1.8454483058790264e-05,
+      "loss": 1.6919,
+      "step": 8868
+    },
+    {
+      "epoch": 0.887,
+      "learning_rate": 1.8450752575720967e-05,
+      "loss": 0.1542,
+      "step": 8870
+    },
+    {
+      "epoch": 0.8872,
+      "learning_rate": 1.844701797385753e-05,
+      "loss": 1.0856,
+      "step": 8872
+    },
+    {
+      "epoch": 0.8874,
+      "learning_rate": 1.8443279255020163e-05,
+      "loss": 0.9438,
+      "step": 8874
+    },
+    {
+      "epoch": 0.8876,
+      "learning_rate": 1.8439536421031035e-05,
+      "loss": 0.2923,
+      "step": 8876
+    },
+    {
+      "epoch": 0.8878,
+      "learning_rate": 1.843578947371439e-05,
+      "loss": 0.5542,
+      "step": 8878
+    },
+    {
+      "epoch": 0.888,
+      "learning_rate": 1.8432038414896446e-05,
+      "loss": 0.3258,
+      "step": 8880
+    },
+    {
+      "epoch": 0.8882,
+      "learning_rate": 1.842828324640539e-05,
+      "loss": 0.4584,
+      "step": 8882
+    },
+    {
+      "epoch": 0.8884,
+      "learning_rate": 1.842452397007148e-05,
+      "loss": 0.7359,
+      "step": 8884
+    },
+    {
+      "epoch": 0.8886,
+      "learning_rate": 1.8420760587726935e-05,
+      "loss": 0.2066,
+      "step": 8886
+    },
+    {
+      "epoch": 0.8888,
+      "learning_rate": 1.841699310120596e-05,
+      "loss": 0.7038,
+      "step": 8888
+    },
+    {
+      "epoch": 0.889,
+      "learning_rate": 1.8413221512344808e-05,
+      "loss": 0.4984,
+      "step": 8890
+    },
+    {
+      "epoch": 0.8892,
+      "learning_rate": 1.8409445822981684e-05,
+      "loss": 1.3347,
+      "step": 8892
+    },
+    {
+      "epoch": 0.8894,
+      "learning_rate": 1.8405666034956846e-05,
+      "loss": 0.5485,
+      "step": 8894
+    },
+    {
+      "epoch": 0.8896,
+      "learning_rate": 1.8401882150112488e-05,
+      "loss": 1.1981,
+      "step": 8896
+    },
+    {
+      "epoch": 0.8898,
+      "learning_rate": 1.8398094170292826e-05,
+      "loss": 0.4711,
+      "step": 8898
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8394302097344106e-05,
+      "loss": 0.6369,
+      "step": 8900
+    },
+    {
+      "epoch": 0.8902,
+      "learning_rate": 1.8390505933114507e-05,
+      "loss": 1.0665,
+      "step": 8902
+    },
+    {
+      "epoch": 0.8904,
+      "learning_rate": 1.8386705679454236e-05,
+      "loss": 0.4518,
+      "step": 8904
+    },
+    {
+      "epoch": 0.8906,
+      "learning_rate": 1.838290133821552e-05,
+      "loss": 1.1212,
+      "step": 8906
+    },
+    {
+      "epoch": 0.8908,
+      "learning_rate": 1.837909291125252e-05,
+      "loss": 0.9502,
+      "step": 8908
+    },
+    {
+      "epoch": 0.891,
+      "learning_rate": 1.8375280400421414e-05,
+      "loss": 0.811,
+      "step": 8910
+    },
+    {
+      "epoch": 0.8912,
+      "learning_rate": 1.8371463807580406e-05,
+      "loss": 1.106,
+      "step": 8912
+    },
+    {
+      "epoch": 0.8914,
+      "learning_rate": 1.8367643134589613e-05,
+      "loss": 0.7423,
+      "step": 8914
+    },
+    {
+      "epoch": 0.8916,
+      "learning_rate": 1.8363818383311233e-05,
+      "loss": 1.2555,
+      "step": 8916
+    },
+    {
+      "epoch": 0.8918,
+      "learning_rate": 1.8359989555609365e-05,
+      "loss": 0.9886,
+      "step": 8918
+    },
+    {
+      "epoch": 0.892,
+      "learning_rate": 1.835615665335013e-05,
+      "loss": 0.4981,
+      "step": 8920
+    },
+    {
+      "epoch": 0.8922,
+      "learning_rate": 1.835231967840168e-05,
+      "loss": 0.8706,
+      "step": 8922
+    },
+    {
+      "epoch": 0.8924,
+      "learning_rate": 1.8348478632634074e-05,
+      "loss": 0.683,
+      "step": 8924
+    },
+    {
+      "epoch": 0.8926,
+      "learning_rate": 1.834463351791939e-05,
+      "loss": 0.0082,
+      "step": 8926
+    },
+    {
+      "epoch": 0.8928,
+      "learning_rate": 1.8340784336131722e-05,
+      "loss": 1.0216,
+      "step": 8928
+    },
+    {
+      "epoch": 0.893,
+      "learning_rate": 1.8336931089147082e-05,
+      "loss": 1.0474,
+      "step": 8930
+    },
+    {
+      "epoch": 0.8932,
+      "learning_rate": 1.8333073778843497e-05,
+      "loss": 1.2153,
+      "step": 8932
+    },
+    {
+      "epoch": 0.8934,
+      "learning_rate": 1.8329212407101006e-05,
+      "loss": 0.6688,
+      "step": 8934
+    },
+    {
+      "epoch": 0.8936,
+      "learning_rate": 1.832534697580155e-05,
+      "loss": 0.7191,
+      "step": 8936
+    },
+    {
+      "epoch": 0.8938,
+      "learning_rate": 1.8321477486829128e-05,
+      "loss": 0.4146,
+      "step": 8938
+    },
+    {
+      "epoch": 0.894,
+      "learning_rate": 1.8317603942069672e-05,
+      "loss": 0.9323,
+      "step": 8940
+    },
+    {
+      "epoch": 0.8942,
+      "learning_rate": 1.8313726343411085e-05,
+      "loss": 0.5111,
+      "step": 8942
+    },
+    {
+      "epoch": 0.8944,
+      "learning_rate": 1.8309844692743283e-05,
+      "loss": 0.0459,
+      "step": 8944
+    },
+    {
+      "epoch": 0.8946,
+      "learning_rate": 1.8305958991958135e-05,
+      "loss": 0.9198,
+      "step": 8946
+    },
+    {
+      "epoch": 0.8948,
+      "learning_rate": 1.830206924294946e-05,
+      "loss": 0.0431,
+      "step": 8948
+    },
+    {
+      "epoch": 0.895,
+      "learning_rate": 1.82981754476131e-05,
+      "loss": 1.5649,
+      "step": 8950
+    },
+    {
+      "epoch": 0.8952,
+      "learning_rate": 1.829427760784684e-05,
+      "loss": 0.5642,
+      "step": 8952
+    },
+    {
+      "epoch": 0.8954,
+      "learning_rate": 1.8290375725550417e-05,
+      "loss": 1.3694,
+      "step": 8954
+    },
+    {
+      "epoch": 0.8956,
+      "learning_rate": 1.8286469802625592e-05,
+      "loss": 0.1339,
+      "step": 8956
+    },
+    {
+      "epoch": 0.8958,
+      "learning_rate": 1.8282559840976053e-05,
+      "loss": 1.0618,
+      "step": 8958
+    },
+    {
+      "epoch": 0.896,
+      "learning_rate": 1.827864584250745e-05,
+      "loss": 0.8714,
+      "step": 8960
+    },
+    {
+      "epoch": 0.8962,
+      "learning_rate": 1.827472780912744e-05,
+      "loss": 0.2321,
+      "step": 8962
+    },
+    {
+      "epoch": 0.8964,
+      "learning_rate": 1.8270805742745612e-05,
+      "loss": 0.873,
+      "step": 8964
+    },
+    {
+      "epoch": 0.8966,
+      "learning_rate": 1.8266879645273557e-05,
+      "loss": 0.6791,
+      "step": 8966
+    },
+    {
+      "epoch": 0.8968,
+      "learning_rate": 1.826294951862478e-05,
+      "loss": 0.6529,
+      "step": 8968
+    },
+    {
+      "epoch": 0.897,
+      "learning_rate": 1.825901536471478e-05,
+      "loss": 0.5059,
+      "step": 8970
+    },
+    {
+      "epoch": 0.8972,
+      "learning_rate": 1.825507718546104e-05,
+      "loss": 0.8983,
+      "step": 8972
+    },
+    {
+      "epoch": 0.8974,
+      "learning_rate": 1.8251134982782966e-05,
+      "loss": 0.7308,
+      "step": 8974
+    },
+    {
+      "epoch": 0.8976,
+      "learning_rate": 1.8247188758601916e-05,
+      "loss": 0.195,
+      "step": 8976
+    },
+    {
+      "epoch": 0.8978,
+      "learning_rate": 1.824323851484126e-05,
+      "loss": 0.1607,
+      "step": 8978
+    },
+    {
+      "epoch": 0.898,
+      "learning_rate": 1.8239284253426308e-05,
+      "loss": 1.2183,
+      "step": 8980
+    },
+    {
+      "epoch": 0.8982,
+      "learning_rate": 1.823532597628428e-05,
+      "loss": 1.1401,
+      "step": 8982
+    },
+    {
+      "epoch": 0.8984,
+      "learning_rate": 1.8231363685344426e-05,
+      "loss": 1.7624,
+      "step": 8984
+    },
+    {
+      "epoch": 0.8986,
+      "learning_rate": 1.8227397382537893e-05,
+      "loss": 0.6289,
+      "step": 8986
+    },
+    {
+      "epoch": 0.8988,
+      "learning_rate": 1.822342706979785e-05,
+      "loss": 1.041,
+      "step": 8988
+    },
+    {
+      "epoch": 0.899,
+      "learning_rate": 1.8219452749059336e-05,
+      "loss": 0.6068,
+      "step": 8990
+    },
+    {
+      "epoch": 0.8992,
+      "learning_rate": 1.82154744222594e-05,
+      "loss": 0.3657,
+      "step": 8992
+    },
+    {
+      "epoch": 0.8994,
+      "learning_rate": 1.8211492091337048e-05,
+      "loss": 0.9685,
+      "step": 8994
+    },
+    {
+      "epoch": 0.8996,
+      "learning_rate": 1.8207505758233196e-05,
+      "loss": 0.2576,
+      "step": 8996
+    },
+    {
+      "epoch": 0.8998,
+      "learning_rate": 1.8203515424890734e-05,
+      "loss": 1.0307,
+      "step": 8998
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.819952109325453e-05,
+      "loss": 0.5051,
+      "step": 9000
+    },
+    {
+      "epoch": 0.9002,
+      "learning_rate": 1.8195522765271346e-05,
+      "loss": 0.2076,
+      "step": 9002
+    },
+    {
+      "epoch": 0.9004,
+      "learning_rate": 1.8191520442889914e-05,
+      "loss": 1.0355,
+      "step": 9004
+    },
+    {
+      "epoch": 0.9006,
+      "learning_rate": 1.8187514128060956e-05,
+      "loss": 0.8128,
+      "step": 9006
+    },
+    {
+      "epoch": 0.9008,
+      "learning_rate": 1.818350382273706e-05,
+      "loss": 0.4494,
+      "step": 9008
+    },
+    {
+      "epoch": 0.901,
+      "learning_rate": 1.8179489528872804e-05,
+      "loss": 0.5253,
+      "step": 9010
+    },
+    {
+      "epoch": 0.9012,
+      "learning_rate": 1.817547124842474e-05,
+      "loss": 0.9839,
+      "step": 9012
+    },
+    {
+      "epoch": 0.9014,
+      "learning_rate": 1.8171448983351284e-05,
+      "loss": 0.7219,
+      "step": 9014
+    },
+    {
+      "epoch": 0.9016,
+      "learning_rate": 1.8167422735612873e-05,
+      "loss": 0.6332,
+      "step": 9016
+    },
+    {
+      "epoch": 0.9018,
+      "learning_rate": 1.816339250717185e-05,
+      "loss": 0.9006,
+      "step": 9018
+    },
+    {
+      "epoch": 0.902,
+      "learning_rate": 1.815935829999247e-05,
+      "loss": 0.974,
+      "step": 9020
+    },
+    {
+      "epoch": 0.9022,
+      "learning_rate": 1.8155320116040983e-05,
+      "loss": 0.5001,
+      "step": 9022
+    },
+    {
+      "epoch": 0.9024,
+      "learning_rate": 1.8151277957285555e-05,
+      "loss": 0.1166,
+      "step": 9024
+    },
+    {
+      "epoch": 0.9026,
+      "learning_rate": 1.814723182569625e-05,
+      "loss": 0.7931,
+      "step": 9026
+    },
+    {
+      "epoch": 0.9028,
+      "learning_rate": 1.814318172324514e-05,
+      "loss": 0.5737,
+      "step": 9028
+    },
+    {
+      "epoch": 0.903,
+      "learning_rate": 1.8139127651906193e-05,
+      "loss": 0.8399,
+      "step": 9030
+    },
+    {
+      "epoch": 0.9032,
+      "learning_rate": 1.813506961365528e-05,
+      "loss": 0.5598,
+      "step": 9032
+    },
+    {
+      "epoch": 0.9034,
+      "learning_rate": 1.813100761047029e-05,
+      "loss": 0.641,
+      "step": 9034
+    },
+    {
+      "epoch": 0.9036,
+      "learning_rate": 1.8126941644330942e-05,
+      "loss": 0.9707,
+      "step": 9036
+    },
+    {
+      "epoch": 0.9038,
+      "learning_rate": 1.8122871717218974e-05,
+      "loss": 0.2418,
+      "step": 9038
+    },
+    {
+      "epoch": 0.904,
+      "learning_rate": 1.811879783111802e-05,
+      "loss": 0.4902,
+      "step": 9040
+    },
+    {
+      "epoch": 0.9042,
+      "learning_rate": 1.8114719988013612e-05,
+      "loss": 0.7127,
+      "step": 9042
+    },
+    {
+      "epoch": 0.9044,
+      "learning_rate": 1.811063818989327e-05,
+      "loss": 1.3044,
+      "step": 9044
+    },
+    {
+      "epoch": 0.9046,
+      "learning_rate": 1.8106552438746413e-05,
+      "loss": 1.7619,
+      "step": 9046
+    },
+    {
+      "epoch": 0.9048,
+      "learning_rate": 1.8102462736564358e-05,
+      "loss": 1.1535,
+      "step": 9048
+    },
+    {
+      "epoch": 0.905,
+      "learning_rate": 1.8098369085340404e-05,
+      "loss": 1.0134,
+      "step": 9050
+    },
+    {
+      "epoch": 0.9052,
+      "learning_rate": 1.8094271487069747e-05,
+      "loss": 0.0415,
+      "step": 9052
+    },
+    {
+      "epoch": 0.9054,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 1.2649,
+      "step": 9054
+    },
+    {
+      "epoch": 0.9056,
+      "learning_rate": 1.808606445737867e-05,
+      "loss": 1.1953,
+      "step": 9056
+    },
+    {
+      "epoch": 0.9058,
+      "learning_rate": 1.8081955029958265e-05,
+      "loss": 0.9194,
+      "step": 9058
+    },
+    {
+      "epoch": 0.906,
+      "learning_rate": 1.807784166349118e-05,
+      "loss": 0.7285,
+      "step": 9060
+    },
+    {
+      "epoch": 0.9062,
+      "learning_rate": 1.807372435998219e-05,
+      "loss": 0.6855,
+      "step": 9062
+    },
+    {
+      "epoch": 0.9064,
+      "learning_rate": 1.8069603121438015e-05,
+      "loss": 1.1283,
+      "step": 9064
+    },
+    {
+      "epoch": 0.9066,
+      "learning_rate": 1.806547794986733e-05,
+      "loss": 0.7766,
+      "step": 9066
+    },
+    {
+      "epoch": 0.9068,
+      "learning_rate": 1.8061348847280664e-05,
+      "loss": 0.5401,
+      "step": 9068
+    },
+    {
+      "epoch": 0.907,
+      "learning_rate": 1.8057215815690487e-05,
+      "loss": 0.9291,
+      "step": 9070
+    },
+    {
+      "epoch": 0.9072,
+      "learning_rate": 1.8053078857111225e-05,
+      "loss": 0.7811,
+      "step": 9072
+    },
+    {
+      "epoch": 0.9074,
+      "learning_rate": 1.8048937973559148e-05,
+      "loss": 0.5791,
+      "step": 9074
+    },
+    {
+      "epoch": 0.9076,
+      "learning_rate": 1.8044793167052473e-05,
+      "loss": 0.8756,
+      "step": 9076
+    },
+    {
+      "epoch": 0.9078,
+      "learning_rate": 1.8040644439611355e-05,
+      "loss": 0.4322,
+      "step": 9078
+    },
+    {
+      "epoch": 0.908,
+      "learning_rate": 1.8036491793257805e-05,
+      "loss": 0.5223,
+      "step": 9080
+    },
+    {
+      "epoch": 0.9082,
+      "learning_rate": 1.8032335230015777e-05,
+      "loss": 0.1952,
+      "step": 9082
+    },
+    {
+      "epoch": 0.9084,
+      "learning_rate": 1.8028174751911154e-05,
+      "loss": 0.3792,
+      "step": 9084
+    },
+    {
+      "epoch": 0.9086,
+      "learning_rate": 1.8024010360971665e-05,
+      "loss": 0.9309,
+      "step": 9086
+    },
+    {
+      "epoch": 0.9088,
+      "learning_rate": 1.801984205922701e-05,
+      "loss": 0.3482,
+      "step": 9088
+    },
+    {
+      "epoch": 0.909,
+      "learning_rate": 1.8015669848708774e-05,
+      "loss": 1.7357,
+      "step": 9090
+    },
+    {
+      "epoch": 0.9092,
+      "learning_rate": 1.801149373145041e-05,
+      "loss": 0.8093,
+      "step": 9092
+    },
+    {
+      "epoch": 0.9094,
+      "learning_rate": 1.8007313709487345e-05,
+      "loss": 0.5086,
+      "step": 9094
+    },
+    {
+      "epoch": 0.9096,
+      "learning_rate": 1.8003129784856843e-05,
+      "loss": 1.0294,
+      "step": 9096
+    },
+    {
+      "epoch": 0.9098,
+      "learning_rate": 1.7998941959598093e-05,
+      "loss": 0.4912,
+      "step": 9098
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.799475023575223e-05,
+      "loss": 1.1283,
+      "step": 9100
+    },
+    {
+      "epoch": 0.9102,
+      "learning_rate": 1.7990554615362207e-05,
+      "loss": 1.1859,
+      "step": 9102
+    },
+    {
+      "epoch": 0.9104,
+      "learning_rate": 1.798635510047293e-05,
+      "loss": 0.603,
+      "step": 9104
+    },
+    {
+      "epoch": 0.9106,
+      "learning_rate": 1.7982151693131213e-05,
+      "loss": 0.4123,
+      "step": 9106
+    },
+    {
+      "epoch": 0.9108,
+      "learning_rate": 1.7977944395385713e-05,
+      "loss": 0.3238,
+      "step": 9108
+    },
+    {
+      "epoch": 0.911,
+      "learning_rate": 1.7973733209287036e-05,
+      "loss": 1.0616,
+      "step": 9110
+    },
+    {
+      "epoch": 0.9112,
+      "learning_rate": 1.7969518136887677e-05,
+      "loss": 0.6051,
+      "step": 9112
+    },
+    {
+      "epoch": 0.9114,
+      "learning_rate": 1.7965299180241963e-05,
+      "loss": 0.5054,
+      "step": 9114
+    },
+    {
+      "epoch": 0.9116,
+      "learning_rate": 1.796107634140621e-05,
+      "loss": 1.1384,
+      "step": 9116
+    },
+    {
+      "epoch": 0.9118,
+      "learning_rate": 1.7956849622438568e-05,
+      "loss": 0.9036,
+      "step": 9118
+    },
+    {
+      "epoch": 0.912,
+      "learning_rate": 1.795261902539906e-05,
+      "loss": 0.8059,
+      "step": 9120
+    },
+    {
+      "epoch": 0.9122,
+      "learning_rate": 1.794838455234966e-05,
+      "loss": 0.8486,
+      "step": 9122
+    },
+    {
+      "epoch": 0.9124,
+      "learning_rate": 1.7944146205354192e-05,
+      "loss": 0.9952,
+      "step": 9124
+    },
+    {
+      "epoch": 0.9126,
+      "learning_rate": 1.7939903986478357e-05,
+      "loss": 0.6163,
+      "step": 9126
+    },
+    {
+      "epoch": 0.9128,
+      "learning_rate": 1.793565789778978e-05,
+      "loss": 0.8766,
+      "step": 9128
+    },
+    {
+      "epoch": 0.913,
+      "learning_rate": 1.7931407941357938e-05,
+      "loss": 1.004,
+      "step": 9130
+    },
+    {
+      "epoch": 0.9132,
+      "learning_rate": 1.792715411925424e-05,
+      "loss": 0.0515,
+      "step": 9132
+    },
+    {
+      "epoch": 0.9134,
+      "learning_rate": 1.7922896433551913e-05,
+      "loss": 0.5294,
+      "step": 9134
+    },
+    {
+      "epoch": 0.9136,
+      "learning_rate": 1.79186348863261e-05,
+      "loss": 1.0422,
+      "step": 9136
+    },
+    {
+      "epoch": 0.9138,
+      "learning_rate": 1.7914369479653864e-05,
+      "loss": 1.0074,
+      "step": 9138
+    },
+    {
+      "epoch": 0.914,
+      "learning_rate": 1.7910100215614076e-05,
+      "loss": 0.2672,
+      "step": 9140
+    },
+    {
+      "epoch": 0.9142,
+      "learning_rate": 1.7905827096287525e-05,
+      "loss": 0.3199,
+      "step": 9142
+    },
+    {
+      "epoch": 0.9144,
+      "learning_rate": 1.790155012375691e-05,
+      "loss": 0.9572,
+      "step": 9144
+    },
+    {
+      "epoch": 0.9146,
+      "learning_rate": 1.7897269300106752e-05,
+      "loss": 0.7235,
+      "step": 9146
+    },
+    {
+      "epoch": 0.9148,
+      "learning_rate": 1.7892984627423446e-05,
+      "loss": 0.6838,
+      "step": 9148
+    },
+    {
+      "epoch": 0.915,
+      "learning_rate": 1.7888696107795347e-05,
+      "loss": 0.5168,
+      "step": 9150
+    },
+    {
+      "epoch": 0.9152,
+      "learning_rate": 1.78844037433126e-05,
+      "loss": 1.186,
+      "step": 9152
+    },
+    {
+      "epoch": 0.9154,
+      "learning_rate": 1.7880107536067228e-05,
+      "loss": 0.6151,
+      "step": 9154
+    },
+    {
+      "epoch": 0.9156,
+      "learning_rate": 1.787580748815318e-05,
+      "loss": 0.5475,
+      "step": 9156
+    },
+    {
+      "epoch": 0.9158,
+      "learning_rate": 1.787150360166623e-05,
+      "loss": 0.0876,
+      "step": 9158
+    },
+    {
+      "epoch": 0.916,
+      "learning_rate": 1.786719587870407e-05,
+      "loss": 0.1179,
+      "step": 9160
+    },
+    {
+      "epoch": 0.9162,
+      "learning_rate": 1.78628843213662e-05,
+      "loss": 0.406,
+      "step": 9162
+    },
+    {
+      "epoch": 0.9164,
+      "learning_rate": 1.7858568931754017e-05,
+      "loss": 0.9038,
+      "step": 9164
+    },
+    {
+      "epoch": 0.9166,
+      "learning_rate": 1.7854249711970826e-05,
+      "loss": 0.5451,
+      "step": 9166
+    },
+    {
+      "epoch": 0.9168,
+      "learning_rate": 1.784992666412173e-05,
+      "loss": 0.6724,
+      "step": 9168
+    },
+    {
+      "epoch": 0.917,
+      "learning_rate": 1.7845599790313732e-05,
+      "loss": 1.3125,
+      "step": 9170
+    },
+    {
+      "epoch": 0.9172,
+      "learning_rate": 1.7841269092655724e-05,
+      "loss": 1.206,
+      "step": 9172
+    },
+    {
+      "epoch": 0.9174,
+      "learning_rate": 1.783693457325841e-05,
+      "loss": 0.383,
+      "step": 9174
+    },
+    {
+      "epoch": 0.9176,
+      "learning_rate": 1.7832596234234373e-05,
+      "loss": 0.8692,
+      "step": 9176
+    },
+    {
+      "epoch": 0.9178,
+      "learning_rate": 1.782825407769811e-05,
+      "loss": 0.46,
+      "step": 9178
+    },
+    {
+      "epoch": 0.918,
+      "learning_rate": 1.7823908105765883e-05,
+      "loss": 0.3941,
+      "step": 9180
+    },
+    {
+      "epoch": 0.9182,
+      "learning_rate": 1.7819558320555902e-05,
+      "loss": 0.035,
+      "step": 9182
+    },
+    {
+      "epoch": 0.9184,
+      "learning_rate": 1.78152047241882e-05,
+      "loss": 0.4811,
+      "step": 9184
+    },
+    {
+      "epoch": 0.9186,
+      "learning_rate": 1.7810847318784632e-05,
+      "loss": 1.2301,
+      "step": 9186
+    },
+    {
+      "epoch": 0.9188,
+      "learning_rate": 1.7806486106468983e-05,
+      "loss": 0.7766,
+      "step": 9188
+    },
+    {
+      "epoch": 0.919,
+      "learning_rate": 1.780212108936685e-05,
+      "loss": 0.5179,
+      "step": 9190
+    },
+    {
+      "epoch": 0.9192,
+      "learning_rate": 1.7797752269605654e-05,
+      "loss": 1.2182,
+      "step": 9192
+    },
+    {
+      "epoch": 0.9194,
+      "learning_rate": 1.7793379649314743e-05,
+      "loss": 1.2538,
+      "step": 9194
+    },
+    {
+      "epoch": 0.9196,
+      "learning_rate": 1.7789003230625276e-05,
+      "loss": 0.5815,
+      "step": 9196
+    },
+    {
+      "epoch": 0.9198,
+      "learning_rate": 1.7784623015670237e-05,
+      "loss": 1.0059,
+      "step": 9198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7780239006584515e-05,
+      "loss": 0.9377,
+      "step": 9200
+    },
+    {
+      "epoch": 0.9202,
+      "learning_rate": 1.777585120550481e-05,
+      "loss": 0.9662,
+      "step": 9202
+    },
+    {
+      "epoch": 0.9204,
+      "learning_rate": 1.777145961456971e-05,
+      "loss": 0.9229,
+      "step": 9204
+    },
+    {
+      "epoch": 0.9206,
+      "learning_rate": 1.7767064235919594e-05,
+      "loss": 0.5915,
+      "step": 9206
+    },
+    {
+      "epoch": 0.9208,
+      "learning_rate": 1.776266507169671e-05,
+      "loss": 0.6373,
+      "step": 9208
+    },
+    {
+      "epoch": 0.921,
+      "learning_rate": 1.77582621240452e-05,
+      "loss": 0.8079,
+      "step": 9210
+    },
+    {
+      "epoch": 0.9212,
+      "learning_rate": 1.7753855395110973e-05,
+      "loss": 0.9754,
+      "step": 9212
+    },
+    {
+      "epoch": 0.9214,
+      "learning_rate": 1.77494448870418e-05,
+      "loss": 0.7114,
+      "step": 9214
+    },
+    {
+      "epoch": 0.9216,
+      "learning_rate": 1.7745030601987345e-05,
+      "loss": 0.3453,
+      "step": 9216
+    },
+    {
+      "epoch": 0.9218,
+      "learning_rate": 1.774061254209907e-05,
+      "loss": 0.9667,
+      "step": 9218
+    },
+    {
+      "epoch": 0.922,
+      "learning_rate": 1.7736190709530254e-05,
+      "loss": 0.8518,
+      "step": 9220
+    },
+    {
+      "epoch": 0.9222,
+      "learning_rate": 1.773176510643608e-05,
+      "loss": 1.1908,
+      "step": 9222
+    },
+    {
+      "epoch": 0.9224,
+      "learning_rate": 1.772733573497353e-05,
+      "loss": 0.6887,
+      "step": 9224
+    },
+    {
+      "epoch": 0.9226,
+      "learning_rate": 1.7722902597301388e-05,
+      "loss": 1.1836,
+      "step": 9226
+    },
+    {
+      "epoch": 0.9228,
+      "learning_rate": 1.7718465695580357e-05,
+      "loss": 0.5297,
+      "step": 9228
+    },
+    {
+      "epoch": 0.923,
+      "learning_rate": 1.7714025031972894e-05,
+      "loss": 0.5086,
+      "step": 9230
+    },
+    {
+      "epoch": 0.9232,
+      "learning_rate": 1.7709580608643367e-05,
+      "loss": 0.7817,
+      "step": 9232
+    },
+    {
+      "epoch": 0.9234,
+      "learning_rate": 1.77051324277579e-05,
+      "loss": 0.3666,
+      "step": 9234
+    },
+    {
+      "epoch": 0.9236,
+      "learning_rate": 1.7700680491484474e-05,
+      "loss": 1.1486,
+      "step": 9236
+    },
+    {
+      "epoch": 0.9238,
+      "learning_rate": 1.769622480199295e-05,
+      "loss": 0.3395,
+      "step": 9238
+    },
+    {
+      "epoch": 0.924,
+      "learning_rate": 1.7691765361454945e-05,
+      "loss": 0.5748,
+      "step": 9240
+    },
+    {
+      "epoch": 0.9242,
+      "learning_rate": 1.7687302172043926e-05,
+      "loss": 0.605,
+      "step": 9242
+    },
+    {
+      "epoch": 0.9244,
+      "learning_rate": 1.768283523593524e-05,
+      "loss": 0.4469,
+      "step": 9244
+    },
+    {
+      "epoch": 0.9246,
+      "learning_rate": 1.7678364555305982e-05,
+      "loss": 0.7423,
+      "step": 9246
+    },
+    {
+      "epoch": 0.9248,
+      "learning_rate": 1.7673890132335105e-05,
+      "loss": 1.095,
+      "step": 9248
+    },
+    {
+      "epoch": 0.925,
+      "learning_rate": 1.7669411969203424e-05,
+      "loss": 1.0999,
+      "step": 9250
+    },
+    {
+      "epoch": 0.9252,
+      "learning_rate": 1.7664930068093497e-05,
+      "loss": 0.3631,
+      "step": 9252
+    },
+    {
+      "epoch": 0.9254,
+      "learning_rate": 1.7660444431189777e-05,
+      "loss": 0.6018,
+      "step": 9254
+    },
+    {
+      "epoch": 0.9256,
+      "learning_rate": 1.7655955060678515e-05,
+      "loss": 0.7606,
+      "step": 9256
+    },
+    {
+      "epoch": 0.9258,
+      "learning_rate": 1.765146195874774e-05,
+      "loss": 0.4011,
+      "step": 9258
+    },
+    {
+      "epoch": 0.926,
+      "learning_rate": 1.7646965127587367e-05,
+      "loss": 0.753,
+      "step": 9260
+    },
+    {
+      "epoch": 0.9262,
+      "learning_rate": 1.76424645693891e-05,
+      "loss": 1.4608,
+      "step": 9262
+    },
+    {
+      "epoch": 0.9264,
+      "learning_rate": 1.763796028634642e-05,
+      "loss": 0.5028,
+      "step": 9264
+    },
+    {
+      "epoch": 0.9266,
+      "learning_rate": 1.7633452280654696e-05,
+      "loss": 1.0151,
+      "step": 9266
+    },
+    {
+      "epoch": 0.9268,
+      "learning_rate": 1.7628940554511074e-05,
+      "loss": 0.9431,
+      "step": 9268
+    },
+    {
+      "epoch": 0.927,
+      "learning_rate": 1.762442511011448e-05,
+      "loss": 0.8361,
+      "step": 9270
+    },
+    {
+      "epoch": 0.9272,
+      "learning_rate": 1.761990594966573e-05,
+      "loss": 1.0342,
+      "step": 9272
+    },
+    {
+      "epoch": 0.9274,
+      "learning_rate": 1.761538307536738e-05,
+      "loss": 0.4109,
+      "step": 9274
+    },
+    {
+      "epoch": 0.9276,
+      "learning_rate": 1.7610856489423817e-05,
+      "loss": 1.085,
+      "step": 9276
+    },
+    {
+      "epoch": 0.9278,
+      "learning_rate": 1.7606326194041285e-05,
+      "loss": 0.7826,
+      "step": 9278
+    },
+    {
+      "epoch": 0.928,
+      "learning_rate": 1.7601792191427745e-05,
+      "loss": 2.4953,
+      "step": 9280
+    },
+    {
+      "epoch": 0.9282,
+      "learning_rate": 1.759725448379305e-05,
+      "loss": 0.6555,
+      "step": 9282
+    },
+    {
+      "epoch": 0.9284,
+      "learning_rate": 1.7592713073348822e-05,
+      "loss": 0.4315,
+      "step": 9284
+    },
+    {
+      "epoch": 0.9286,
+      "learning_rate": 1.7588167962308458e-05,
+      "loss": 0.8228,
+      "step": 9286
+    },
+    {
+      "epoch": 0.9288,
+      "learning_rate": 1.7583619152887222e-05,
+      "loss": 0.2139,
+      "step": 9288
+    },
+    {
+      "epoch": 0.929,
+      "learning_rate": 1.7579066647302147e-05,
+      "loss": 0.4473,
+      "step": 9290
+    },
+    {
+      "epoch": 0.9292,
+      "learning_rate": 1.757451044777204e-05,
+      "loss": 0.5386,
+      "step": 9292
+    },
+    {
+      "epoch": 0.9294,
+      "learning_rate": 1.756995055651757e-05,
+      "loss": 0.1139,
+      "step": 9294
+    },
+    {
+      "epoch": 0.9296,
+      "learning_rate": 1.7565386975761167e-05,
+      "loss": 0.597,
+      "step": 9296
+    },
+    {
+      "epoch": 0.9298,
+      "learning_rate": 1.7560819707727037e-05,
+      "loss": 0.2974,
+      "step": 9298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.755624875464124e-05,
+      "loss": 1.3389,
+      "step": 9300
+    },
+    {
+      "epoch": 0.9302,
+      "learning_rate": 1.7551674118731585e-05,
+      "loss": 0.6481,
+      "step": 9302
+    },
+    {
+      "epoch": 0.9304,
+      "learning_rate": 1.7547095802227726e-05,
+      "loss": 0.4967,
+      "step": 9304
+    },
+    {
+      "epoch": 0.9306,
+      "learning_rate": 1.7542513807361044e-05,
+      "loss": 1.2262,
+      "step": 9306
+    },
+    {
+      "epoch": 0.9308,
+      "learning_rate": 1.753792813636475e-05,
+      "loss": 0.5218,
+      "step": 9308
+    },
+    {
+      "epoch": 0.931,
+      "learning_rate": 1.7533338791473875e-05,
+      "loss": 0.9287,
+      "step": 9310
+    },
+    {
+      "epoch": 0.9312,
+      "learning_rate": 1.752874577492518e-05,
+      "loss": 0.1945,
+      "step": 9312
+    },
+    {
+      "epoch": 0.9314,
+      "learning_rate": 1.7524149088957238e-05,
+      "loss": 0.5456,
+      "step": 9314
+    },
+    {
+      "epoch": 0.9316,
+      "learning_rate": 1.7519548735810462e-05,
+      "loss": 0.5663,
+      "step": 9316
+    },
+    {
+      "epoch": 0.9318,
+      "learning_rate": 1.751494471772697e-05,
+      "loss": 0.5,
+      "step": 9318
+    },
+    {
+      "epoch": 0.932,
+      "learning_rate": 1.75103370369507e-05,
+      "loss": 0.5105,
+      "step": 9320
+    },
+    {
+      "epoch": 0.9322,
+      "learning_rate": 1.750572569572742e-05,
+      "loss": 0.8681,
+      "step": 9322
+    },
+    {
+      "epoch": 0.9324,
+      "learning_rate": 1.7501110696304595e-05,
+      "loss": 0.8127,
+      "step": 9324
+    },
+    {
+      "epoch": 0.9326,
+      "learning_rate": 1.7496492040931548e-05,
+      "loss": 0.5462,
+      "step": 9326
+    },
+    {
+      "epoch": 0.9328,
+      "learning_rate": 1.7491869731859363e-05,
+      "loss": 0.9507,
+      "step": 9328
+    },
+    {
+      "epoch": 0.933,
+      "learning_rate": 1.7487243771340862e-05,
+      "loss": 0.9605,
+      "step": 9330
+    },
+    {
+      "epoch": 0.9332,
+      "learning_rate": 1.748261416163072e-05,
+      "loss": 1.0732,
+      "step": 9332
+    },
+    {
+      "epoch": 0.9334,
+      "learning_rate": 1.747798090498533e-05,
+      "loss": 1.0183,
+      "step": 9334
+    },
+    {
+      "epoch": 0.9336,
+      "learning_rate": 1.7473344003662873e-05,
+      "loss": 0.4833,
+      "step": 9336
+    },
+    {
+      "epoch": 0.9338,
+      "learning_rate": 1.7468703459923365e-05,
+      "loss": 0.4708,
+      "step": 9338
+    },
+    {
+      "epoch": 0.934,
+      "learning_rate": 1.7464059276028507e-05,
+      "loss": 0.7059,
+      "step": 9340
+    },
+    {
+      "epoch": 0.9342,
+      "learning_rate": 1.745941145424182e-05,
+      "loss": 0.9299,
+      "step": 9342
+    },
+    {
+      "epoch": 0.9344,
+      "learning_rate": 1.7454759996828632e-05,
+      "loss": 2.1898,
+      "step": 9344
+    },
+    {
+      "epoch": 0.9346,
+      "learning_rate": 1.7450104906055973e-05,
+      "loss": 0.0994,
+      "step": 9346
+    },
+    {
+      "epoch": 0.9348,
+      "learning_rate": 1.7445446184192674e-05,
+      "loss": 0.0172,
+      "step": 9348
+    },
+    {
+      "epoch": 0.935,
+      "learning_rate": 1.744078383350938e-05,
+      "loss": 0.6793,
+      "step": 9350
+    },
+    {
+      "epoch": 0.9352,
+      "learning_rate": 1.7436117856278407e-05,
+      "loss": 0.3646,
+      "step": 9352
+    },
+    {
+      "epoch": 0.9354,
+      "learning_rate": 1.7431448254773943e-05,
+      "loss": 0.9564,
+      "step": 9354
+    },
+    {
+      "epoch": 0.9356,
+      "learning_rate": 1.742677503127189e-05,
+      "loss": 0.7588,
+      "step": 9356
+    },
+    {
+      "epoch": 0.9358,
+      "learning_rate": 1.7422098188049885e-05,
+      "loss": 0.818,
+      "step": 9358
+    },
+    {
+      "epoch": 0.936,
+      "learning_rate": 1.7417417727387392e-05,
+      "loss": 0.5107,
+      "step": 9360
+    },
+    {
+      "epoch": 0.9362,
+      "learning_rate": 1.7412733651565624e-05,
+      "loss": 0.8383,
+      "step": 9362
+    },
+    {
+      "epoch": 0.9364,
+      "learning_rate": 1.74080459628675e-05,
+      "loss": 0.6141,
+      "step": 9364
+    },
+    {
+      "epoch": 0.9366,
+      "learning_rate": 1.7403354663577782e-05,
+      "loss": 1.0191,
+      "step": 9366
+    },
+    {
+      "epoch": 0.9368,
+      "learning_rate": 1.739865975598295e-05,
+      "loss": 0.4997,
+      "step": 9368
+    },
+    {
+      "epoch": 0.937,
+      "learning_rate": 1.739396124237121e-05,
+      "loss": 0.1162,
+      "step": 9370
+    },
+    {
+      "epoch": 0.9372,
+      "learning_rate": 1.738925912503259e-05,
+      "loss": 1.0333,
+      "step": 9372
+    },
+    {
+      "epoch": 0.9374,
+      "learning_rate": 1.738455340625883e-05,
+      "loss": 1.0754,
+      "step": 9374
+    },
+    {
+      "epoch": 0.9376,
+      "learning_rate": 1.7379844088343473e-05,
+      "loss": 0.1496,
+      "step": 9376
+    },
+    {
+      "epoch": 0.9378,
+      "learning_rate": 1.7375131173581744e-05,
+      "loss": 0.497,
+      "step": 9378
+    },
+    {
+      "epoch": 0.938,
+      "learning_rate": 1.7370414664270665e-05,
+      "loss": 0.5407,
+      "step": 9380
+    },
+    {
+      "epoch": 0.9382,
+      "learning_rate": 1.7365694562709038e-05,
+      "loss": 0.5016,
+      "step": 9382
+    },
+    {
+      "epoch": 0.9384,
+      "learning_rate": 1.736097087119736e-05,
+      "loss": 0.4861,
+      "step": 9384
+    },
+    {
+      "epoch": 0.9386,
+      "learning_rate": 1.7356243592037865e-05,
+      "loss": 0.5634,
+      "step": 9386
+    },
+    {
+      "epoch": 0.9388,
+      "learning_rate": 1.7351512727534648e-05,
+      "loss": 0.1949,
+      "step": 9388
+    },
+    {
+      "epoch": 0.939,
+      "learning_rate": 1.7346778279993433e-05,
+      "loss": 0.5465,
+      "step": 9390
+    },
+    {
+      "epoch": 0.9392,
+      "learning_rate": 1.734204025172171e-05,
+      "loss": 0.2497,
+      "step": 9392
+    },
+    {
+      "epoch": 0.9394,
+      "learning_rate": 1.733729864502877e-05,
+      "loss": 0.5805,
+      "step": 9394
+    },
+    {
+      "epoch": 0.9396,
+      "learning_rate": 1.7332553462225593e-05,
+      "loss": 0.6883,
+      "step": 9396
+    },
+    {
+      "epoch": 0.9398,
+      "learning_rate": 1.7327804705624962e-05,
+      "loss": 0.2467,
+      "step": 9398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7323052377541327e-05,
+      "loss": 0.4512,
+      "step": 9400
+    },
+    {
+      "epoch": 0.9402,
+      "learning_rate": 1.731829648029091e-05,
+      "loss": 0.656,
+      "step": 9402
+    },
+    {
+      "epoch": 0.9404,
+      "learning_rate": 1.7313537016191713e-05,
+      "loss": 0.238,
+      "step": 9404
+    },
+    {
+      "epoch": 0.9406,
+      "learning_rate": 1.730877398756341e-05,
+      "loss": 0.4658,
+      "step": 9406
+    },
+    {
+      "epoch": 0.9408,
+      "learning_rate": 1.7304007396727442e-05,
+      "loss": 0.8381,
+      "step": 9408
+    },
+    {
+      "epoch": 0.941,
+      "learning_rate": 1.7299237246007025e-05,
+      "loss": 0.8366,
+      "step": 9410
+    },
+    {
+      "epoch": 0.9412,
+      "learning_rate": 1.7294463537727032e-05,
+      "loss": 0.9903,
+      "step": 9412
+    },
+    {
+      "epoch": 0.9414,
+      "learning_rate": 1.7289686274214113e-05,
+      "loss": 0.7337,
+      "step": 9414
+    },
+    {
+      "epoch": 0.9416,
+      "learning_rate": 1.7284905457796685e-05,
+      "loss": 0.6412,
+      "step": 9416
+    },
+    {
+      "epoch": 0.9418,
+      "learning_rate": 1.7280121090804824e-05,
+      "loss": 0.1746,
+      "step": 9418
+    },
+    {
+      "epoch": 0.942,
+      "learning_rate": 1.7275333175570364e-05,
+      "loss": 0.9721,
+      "step": 9420
+    },
+    {
+      "epoch": 0.9422,
+      "learning_rate": 1.727054171442693e-05,
+      "loss": 0.89,
+      "step": 9422
+    },
+    {
+      "epoch": 0.9424,
+      "learning_rate": 1.726574670970976e-05,
+      "loss": 0.0173,
+      "step": 9424
+    },
+    {
+      "epoch": 0.9426,
+      "learning_rate": 1.7260948163755918e-05,
+      "loss": 0.5585,
+      "step": 9426
+    },
+    {
+      "epoch": 0.9428,
+      "learning_rate": 1.7256146078904163e-05,
+      "loss": 0.6739,
+      "step": 9428
+    },
+    {
+      "epoch": 0.943,
+      "learning_rate": 1.7251340457494934e-05,
+      "loss": 1.1283,
+      "step": 9430
+    },
+    {
+      "epoch": 0.9432,
+      "learning_rate": 1.7246531301870467e-05,
+      "loss": 0.8942,
+      "step": 9432
+    },
+    {
+      "epoch": 0.9434,
+      "learning_rate": 1.7241718614374688e-05,
+      "loss": 1.1614,
+      "step": 9434
+    },
+    {
+      "epoch": 0.9436,
+      "learning_rate": 1.7236902397353204e-05,
+      "loss": 1.9947,
+      "step": 9436
+    },
+    {
+      "epoch": 0.9438,
+      "learning_rate": 1.7232082653153422e-05,
+      "loss": 0.6927,
+      "step": 9438
+    },
+    {
+      "epoch": 0.944,
+      "learning_rate": 1.7227259384124422e-05,
+      "loss": 0.6241,
+      "step": 9440
+    },
+    {
+      "epoch": 0.9442,
+      "learning_rate": 1.722243259261697e-05,
+      "loss": 0.2967,
+      "step": 9442
+    },
+    {
+      "epoch": 0.9444,
+      "learning_rate": 1.7217602280983636e-05,
+      "loss": 0.8284,
+      "step": 9444
+    },
+    {
+      "epoch": 0.9446,
+      "learning_rate": 1.7212768451578595e-05,
+      "loss": 0.6953,
+      "step": 9446
+    },
+    {
+      "epoch": 0.9448,
+      "learning_rate": 1.7207931106757867e-05,
+      "loss": 1.0054,
+      "step": 9448
+    },
+    {
+      "epoch": 0.945,
+      "learning_rate": 1.7203090248879084e-05,
+      "loss": 0.4771,
+      "step": 9450
+    },
+    {
+      "epoch": 0.9452,
+      "learning_rate": 1.7198245880301593e-05,
+      "loss": 0.5013,
+      "step": 9452
+    },
+    {
+      "epoch": 0.9454,
+      "learning_rate": 1.7193398003386517e-05,
+      "loss": 0.7259,
+      "step": 9454
+    },
+    {
+      "epoch": 0.9456,
+      "learning_rate": 1.7188546620496648e-05,
+      "loss": 0.7632,
+      "step": 9456
+    },
+    {
+      "epoch": 0.9458,
+      "learning_rate": 1.7183691733996463e-05,
+      "loss": 0.5939,
+      "step": 9458
+    },
+    {
+      "epoch": 0.946,
+      "learning_rate": 1.717883334625221e-05,
+      "loss": 1.39,
+      "step": 9460
+    },
+    {
+      "epoch": 0.9462,
+      "learning_rate": 1.7173971459631803e-05,
+      "loss": 0.7171,
+      "step": 9462
+    },
+    {
+      "epoch": 0.9464,
+      "learning_rate": 1.7169106076504833e-05,
+      "loss": 0.1009,
+      "step": 9464
+    },
+    {
+      "epoch": 0.9466,
+      "learning_rate": 1.7164237199242663e-05,
+      "loss": 0.873,
+      "step": 9466
+    },
+    {
+      "epoch": 0.9468,
+      "learning_rate": 1.7159364830218305e-05,
+      "loss": 0.2885,
+      "step": 9468
+    },
+    {
+      "epoch": 0.947,
+      "learning_rate": 1.7154488971806525e-05,
+      "loss": 0.551,
+      "step": 9470
+    },
+    {
+      "epoch": 0.9472,
+      "learning_rate": 1.7149609626383725e-05,
+      "loss": 1.1057,
+      "step": 9472
+    },
+    {
+      "epoch": 0.9474,
+      "learning_rate": 1.7144726796328027e-05,
+      "loss": 0.1065,
+      "step": 9474
+    },
+    {
+      "epoch": 0.9476,
+      "learning_rate": 1.7139840484019313e-05,
+      "loss": 0.5936,
+      "step": 9476
+    },
+    {
+      "epoch": 0.9478,
+      "learning_rate": 1.713495069183907e-05,
+      "loss": 1.0046,
+      "step": 9478
+    },
+    {
+      "epoch": 0.948,
+      "learning_rate": 1.7130057422170524e-05,
+      "loss": 0.6075,
+      "step": 9480
+    },
+    {
+      "epoch": 0.9482,
+      "learning_rate": 1.7125160677398632e-05,
+      "loss": 0.6475,
+      "step": 9482
+    },
+    {
+      "epoch": 0.9484,
+      "learning_rate": 1.7120260459909973e-05,
+      "loss": 1.0295,
+      "step": 9484
+    },
+    {
+      "epoch": 0.9486,
+      "learning_rate": 1.7115356772092847e-05,
+      "loss": 0.8436,
+      "step": 9486
+    },
+    {
+      "epoch": 0.9488,
+      "learning_rate": 1.71104496163373e-05,
+      "loss": 0.9035,
+      "step": 9488
+    },
+    {
+      "epoch": 0.949,
+      "learning_rate": 1.710553899503497e-05,
+      "loss": 0.1236,
+      "step": 9490
+    },
+    {
+      "epoch": 0.9492,
+      "learning_rate": 1.7100624910579245e-05,
+      "loss": 0.9367,
+      "step": 9492
+    },
+    {
+      "epoch": 0.9494,
+      "learning_rate": 1.709570736536522e-05,
+      "loss": 0.7842,
+      "step": 9494
+    },
+    {
+      "epoch": 0.9496,
+      "learning_rate": 1.70907863617896e-05,
+      "loss": 0.3806,
+      "step": 9496
+    },
+    {
+      "epoch": 0.9498,
+      "learning_rate": 1.708586190225086e-05,
+      "loss": 0.5482,
+      "step": 9498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.708093398914912e-05,
+      "loss": 1.0731,
+      "step": 9500
+    },
+    {
+      "epoch": 0.9502,
+      "learning_rate": 1.7076002624886152e-05,
+      "loss": 0.5978,
+      "step": 9502
+    },
+    {
+      "epoch": 0.9504,
+      "learning_rate": 1.7071067811865474e-05,
+      "loss": 0.511,
+      "step": 9504
+    },
+    {
+      "epoch": 0.9506,
+      "learning_rate": 1.7066129552492258e-05,
+      "loss": 0.5069,
+      "step": 9506
+    },
+    {
+      "epoch": 0.9508,
+      "learning_rate": 1.7061187849173314e-05,
+      "loss": 0.4626,
+      "step": 9508
+    },
+    {
+      "epoch": 0.951,
+      "learning_rate": 1.705624270431722e-05,
+      "loss": 0.564,
+      "step": 9510
+    },
+    {
+      "epoch": 0.9512,
+      "learning_rate": 1.7051294120334136e-05,
+      "loss": 1.5777,
+      "step": 9512
+    },
+    {
+      "epoch": 0.9514,
+      "learning_rate": 1.7046342099635945e-05,
+      "loss": 0.427,
+      "step": 9514
+    },
+    {
+      "epoch": 0.9516,
+      "learning_rate": 1.704138664463624e-05,
+      "loss": 1.2082,
+      "step": 9516
+    },
+    {
+      "epoch": 0.9518,
+      "learning_rate": 1.70364277577502e-05,
+      "loss": 0.199,
+      "step": 9518
+    },
+    {
+      "epoch": 0.952,
+      "learning_rate": 1.7031465441394766e-05,
+      "loss": 1.2122,
+      "step": 9520
+    },
+    {
+      "epoch": 0.9522,
+      "learning_rate": 1.702649969798851e-05,
+      "loss": 0.7488,
+      "step": 9522
+    },
+    {
+      "epoch": 0.9524,
+      "learning_rate": 1.7021530529951627e-05,
+      "loss": 0.8994,
+      "step": 9524
+    },
+    {
+      "epoch": 0.9526,
+      "learning_rate": 1.7016557939706078e-05,
+      "loss": 0.7238,
+      "step": 9526
+    },
+    {
+      "epoch": 0.9528,
+      "learning_rate": 1.7011581929675438e-05,
+      "loss": 0.9032,
+      "step": 9528
+    },
+    {
+      "epoch": 0.953,
+      "learning_rate": 1.700660250228492e-05,
+      "loss": 0.4305,
+      "step": 9530
+    },
+    {
+      "epoch": 0.9532,
+      "learning_rate": 1.7001619659961474e-05,
+      "loss": 1.0266,
+      "step": 9532
+    },
+    {
+      "epoch": 0.9534,
+      "learning_rate": 1.6996633405133673e-05,
+      "loss": 0.5272,
+      "step": 9534
+    },
+    {
+      "epoch": 0.9536,
+      "learning_rate": 1.6991643740231718e-05,
+      "loss": 0.5308,
+      "step": 9536
+    },
+    {
+      "epoch": 0.9538,
+      "learning_rate": 1.6986650667687556e-05,
+      "loss": 1.1428,
+      "step": 9538
+    },
+    {
+      "epoch": 0.954,
+      "learning_rate": 1.6981654189934746e-05,
+      "loss": 0.8819,
+      "step": 9540
+    },
+    {
+      "epoch": 0.9542,
+      "learning_rate": 1.6976654309408468e-05,
+      "loss": 1.1208,
+      "step": 9542
+    },
+    {
+      "epoch": 0.9544,
+      "learning_rate": 1.697165102854565e-05,
+      "loss": 0.9961,
+      "step": 9544
+    },
+    {
+      "epoch": 0.9546,
+      "learning_rate": 1.69666443497848e-05,
+      "loss": 0.6836,
+      "step": 9546
+    },
+    {
+      "epoch": 0.9548,
+      "learning_rate": 1.696163427556615e-05,
+      "loss": 0.5221,
+      "step": 9548
+    },
+    {
+      "epoch": 0.955,
+      "learning_rate": 1.6956620808331515e-05,
+      "loss": 0.4546,
+      "step": 9550
+    },
+    {
+      "epoch": 0.9552,
+      "learning_rate": 1.6951603950524392e-05,
+      "loss": 0.6162,
+      "step": 9552
+    },
+    {
+      "epoch": 0.9554,
+      "learning_rate": 1.694658370458998e-05,
+      "loss": 0.172,
+      "step": 9554
+    },
+    {
+      "epoch": 0.9556,
+      "learning_rate": 1.6941560072975044e-05,
+      "loss": 0.5882,
+      "step": 9556
+    },
+    {
+      "epoch": 0.9558,
+      "learning_rate": 1.6936533058128042e-05,
+      "loss": 0.6694,
+      "step": 9558
+    },
+    {
+      "epoch": 0.956,
+      "learning_rate": 1.6931502662499123e-05,
+      "loss": 0.9619,
+      "step": 9560
+    },
+    {
+      "epoch": 0.9562,
+      "learning_rate": 1.692646888854001e-05,
+      "loss": 1.2658,
+      "step": 9562
+    },
+    {
+      "epoch": 0.9564,
+      "learning_rate": 1.6921431738704062e-05,
+      "loss": 0.8044,
+      "step": 9564
+    },
+    {
+      "epoch": 0.9566,
+      "learning_rate": 1.691639121544641e-05,
+      "loss": 0.7388,
+      "step": 9566
+    },
+    {
+      "epoch": 0.9568,
+      "learning_rate": 1.6911347321223673e-05,
+      "loss": 0.3058,
+      "step": 9568
+    },
+    {
+      "epoch": 0.957,
+      "learning_rate": 1.690630005849424e-05,
+      "loss": 0.3356,
+      "step": 9570
+    },
+    {
+      "epoch": 0.9572,
+      "learning_rate": 1.690124942971804e-05,
+      "loss": 1.0672,
+      "step": 9572
+    },
+    {
+      "epoch": 0.9574,
+      "learning_rate": 1.6896195437356696e-05,
+      "loss": 0.4001,
+      "step": 9574
+    },
+    {
+      "epoch": 0.9576,
+      "learning_rate": 1.6891138083873496e-05,
+      "loss": 1.0472,
+      "step": 9576
+    },
+    {
+      "epoch": 0.9578,
+      "learning_rate": 1.6886077371733295e-05,
+      "loss": 0.6416,
+      "step": 9578
+    },
+    {
+      "epoch": 0.958,
+      "learning_rate": 1.6881013303402623e-05,
+      "loss": 0.6109,
+      "step": 9580
+    },
+    {
+      "epoch": 0.9582,
+      "learning_rate": 1.6875945881349686e-05,
+      "loss": 0.331,
+      "step": 9582
+    },
+    {
+      "epoch": 0.9584,
+      "learning_rate": 1.6870875108044243e-05,
+      "loss": 0.9889,
+      "step": 9584
+    },
+    {
+      "epoch": 0.9586,
+      "learning_rate": 1.6865800985957725e-05,
+      "loss": 0.3706,
+      "step": 9586
+    },
+    {
+      "epoch": 0.9588,
+      "learning_rate": 1.686072351756324e-05,
+      "loss": 1.7061,
+      "step": 9588
+    },
+    {
+      "epoch": 0.959,
+      "learning_rate": 1.6855642705335435e-05,
+      "loss": 0.4174,
+      "step": 9590
+    },
+    {
+      "epoch": 0.9592,
+      "learning_rate": 1.6850558551750668e-05,
+      "loss": 0.7081,
+      "step": 9592
+    },
+    {
+      "epoch": 0.9594,
+      "learning_rate": 1.68454710592869e-05,
+      "loss": 0.1633,
+      "step": 9594
+    },
+    {
+      "epoch": 0.9596,
+      "learning_rate": 1.684038023042367e-05,
+      "loss": 1.2731,
+      "step": 9596
+    },
+    {
+      "epoch": 0.9598,
+      "learning_rate": 1.6835286067642228e-05,
+      "loss": 0.9249,
+      "step": 9598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6830188573425402e-05,
+      "loss": 0.4056,
+      "step": 9600
+    },
+    {
+      "epoch": 0.9602,
+      "learning_rate": 1.6825087750257617e-05,
+      "loss": 0.4466,
+      "step": 9602
+    },
+    {
+      "epoch": 0.9604,
+      "learning_rate": 1.6819983600624986e-05,
+      "loss": 0.6317,
+      "step": 9604
+    },
+    {
+      "epoch": 0.9606,
+      "learning_rate": 1.681487612701521e-05,
+      "loss": 0.6156,
+      "step": 9606
+    },
+    {
+      "epoch": 0.9608,
+      "learning_rate": 1.6809765331917576e-05,
+      "loss": 0.6018,
+      "step": 9608
+    },
+    {
+      "epoch": 0.961,
+      "learning_rate": 1.6804651217823055e-05,
+      "loss": 1.0021,
+      "step": 9610
+    },
+    {
+      "epoch": 0.9612,
+      "learning_rate": 1.679953378722421e-05,
+      "loss": 0.7946,
+      "step": 9612
+    },
+    {
+      "epoch": 0.9614,
+      "learning_rate": 1.6794413042615168e-05,
+      "loss": 0.5325,
+      "step": 9614
+    },
+    {
+      "epoch": 0.9616,
+      "learning_rate": 1.6789288986491764e-05,
+      "loss": 1.2557,
+      "step": 9616
+    },
+    {
+      "epoch": 0.9618,
+      "learning_rate": 1.6784161621351374e-05,
+      "loss": 0.0071,
+      "step": 9618
+    },
+    {
+      "epoch": 0.962,
+      "learning_rate": 1.6779030949693048e-05,
+      "loss": 0.9299,
+      "step": 9620
+    },
+    {
+      "epoch": 0.9622,
+      "learning_rate": 1.677389697401739e-05,
+      "loss": 0.4899,
+      "step": 9622
+    },
+    {
+      "epoch": 0.9624,
+      "learning_rate": 1.67687596968266e-05,
+      "loss": 0.6604,
+      "step": 9624
+    },
+    {
+      "epoch": 0.9626,
+      "learning_rate": 1.67636191206246e-05,
+      "loss": 0.4843,
+      "step": 9626
+    },
+    {
+      "epoch": 0.9628,
+      "learning_rate": 1.6758475247916803e-05,
+      "loss": 0.5838,
+      "step": 9628
+    },
+    {
+      "epoch": 0.963,
+      "learning_rate": 1.675332808121025e-05,
+      "loss": 0.7187,
+      "step": 9630
+    },
+    {
+      "epoch": 0.9632,
+      "learning_rate": 1.6748177623013645e-05,
+      "loss": 0.6128,
+      "step": 9632
+    },
+    {
+      "epoch": 0.9634,
+      "learning_rate": 1.6743023875837253e-05,
+      "loss": 0.7841,
+      "step": 9634
+    },
+    {
+      "epoch": 0.9636,
+      "learning_rate": 1.6737866842192914e-05,
+      "loss": 0.5655,
+      "step": 9636
+    },
+    {
+      "epoch": 0.9638,
+      "learning_rate": 1.6732706524594145e-05,
+      "loss": 0.441,
+      "step": 9638
+    },
+    {
+      "epoch": 0.964,
+      "learning_rate": 1.672754292555599e-05,
+      "loss": 0.4403,
+      "step": 9640
+    },
+    {
+      "epoch": 0.9642,
+      "learning_rate": 1.672237604759517e-05,
+      "loss": 0.6228,
+      "step": 9642
+    },
+    {
+      "epoch": 0.9644,
+      "learning_rate": 1.671720589322991e-05,
+      "loss": 0.5198,
+      "step": 9644
+    },
+    {
+      "epoch": 0.9646,
+      "learning_rate": 1.671203246498009e-05,
+      "loss": 0.5799,
+      "step": 9646
+    },
+    {
+      "epoch": 0.9648,
+      "learning_rate": 1.670685576536721e-05,
+      "loss": 0.0971,
+      "step": 9648
+    },
+    {
+      "epoch": 0.965,
+      "learning_rate": 1.670167579691429e-05,
+      "loss": 0.463,
+      "step": 9650
+    },
+    {
+      "epoch": 0.9652,
+      "learning_rate": 1.6696492562145992e-05,
+      "loss": 0.1304,
+      "step": 9652
+    },
+    {
+      "epoch": 0.9654,
+      "learning_rate": 1.6691306063588593e-05,
+      "loss": 0.0011,
+      "step": 9654
+    },
+    {
+      "epoch": 0.9656,
+      "learning_rate": 1.668611630376989e-05,
+      "loss": 0.3997,
+      "step": 9656
+    },
+    {
+      "epoch": 0.9658,
+      "learning_rate": 1.668092328521931e-05,
+      "loss": 0.3031,
+      "step": 9658
+    },
+    {
+      "epoch": 0.966,
+      "learning_rate": 1.6675727010467915e-05,
+      "loss": 1.166,
+      "step": 9660
+    },
+    {
+      "epoch": 0.9662,
+      "learning_rate": 1.6670527482048242e-05,
+      "loss": 0.4139,
+      "step": 9662
+    },
+    {
+      "epoch": 0.9664,
+      "learning_rate": 1.6665324702494524e-05,
+      "loss": 1.2796,
+      "step": 9664
+    },
+    {
+      "epoch": 0.9666,
+      "learning_rate": 1.6660118674342525e-05,
+      "loss": 0.6968,
+      "step": 9666
+    },
+    {
+      "epoch": 0.9668,
+      "learning_rate": 1.6654909400129572e-05,
+      "loss": 0.9846,
+      "step": 9668
+    },
+    {
+      "epoch": 0.967,
+      "learning_rate": 1.6649696882394635e-05,
+      "loss": 0.9478,
+      "step": 9670
+    },
+    {
+      "epoch": 0.9672,
+      "learning_rate": 1.664448112367823e-05,
+      "loss": 0.6493,
+      "step": 9672
+    },
+    {
+      "epoch": 0.9674,
+      "learning_rate": 1.6639262126522414e-05,
+      "loss": 1.1936,
+      "step": 9674
+    },
+    {
+      "epoch": 0.9676,
+      "learning_rate": 1.663403989347091e-05,
+      "loss": 0.6998,
+      "step": 9676
+    },
+    {
+      "epoch": 0.9678,
+      "learning_rate": 1.6628814427068968e-05,
+      "loss": 0.2143,
+      "step": 9678
+    },
+    {
+      "epoch": 0.968,
+      "learning_rate": 1.662358572986337e-05,
+      "loss": 1.17,
+      "step": 9680
+    },
+    {
+      "epoch": 0.9682,
+      "learning_rate": 1.661835380440258e-05,
+      "loss": 1.8924,
+      "step": 9682
+    },
+    {
+      "epoch": 0.9684,
+      "learning_rate": 1.6613118653236533e-05,
+      "loss": 0.6018,
+      "step": 9684
+    },
+    {
+      "epoch": 0.9686,
+      "learning_rate": 1.6607880278916778e-05,
+      "loss": 0.1148,
+      "step": 9686
+    },
+    {
+      "epoch": 0.9688,
+      "learning_rate": 1.6602638683996476e-05,
+      "loss": 0.4106,
+      "step": 9688
+    },
+    {
+      "epoch": 0.969,
+      "learning_rate": 1.6597393871030264e-05,
+      "loss": 0.587,
+      "step": 9690
+    },
+    {
+      "epoch": 0.9692,
+      "learning_rate": 1.6592145842574437e-05,
+      "loss": 0.8882,
+      "step": 9692
+    },
+    {
+      "epoch": 0.9694,
+      "learning_rate": 1.6586894601186824e-05,
+      "loss": 0.8633,
+      "step": 9694
+    },
+    {
+      "epoch": 0.9696,
+      "learning_rate": 1.658164014942677e-05,
+      "loss": 0.5655,
+      "step": 9696
+    },
+    {
+      "epoch": 0.9698,
+      "learning_rate": 1.6576382489855278e-05,
+      "loss": 1.2178,
+      "step": 9698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6571121625034863e-05,
+      "loss": 0.7497,
+      "step": 9700
+    },
+    {
+      "epoch": 0.9702,
+      "learning_rate": 1.656585755752957e-05,
+      "loss": 0.7619,
+      "step": 9702
+    },
+    {
+      "epoch": 0.9704,
+      "learning_rate": 1.6560590289905078e-05,
+      "loss": 0.9214,
+      "step": 9704
+    },
+    {
+      "epoch": 0.9706,
+      "learning_rate": 1.655531982472859e-05,
+      "loss": 1.0947,
+      "step": 9706
+    },
+    {
+      "epoch": 0.9708,
+      "learning_rate": 1.655004616456883e-05,
+      "loss": 0.3188,
+      "step": 9708
+    },
+    {
+      "epoch": 0.971,
+      "learning_rate": 1.6544769311996153e-05,
+      "loss": 1.215,
+      "step": 9710
+    },
+    {
+      "epoch": 0.9712,
+      "learning_rate": 1.6539489269582408e-05,
+      "loss": 0.9216,
+      "step": 9712
+    },
+    {
+      "epoch": 0.9714,
+      "learning_rate": 1.653420603990106e-05,
+      "loss": 0.4448,
+      "step": 9714
+    },
+    {
+      "epoch": 0.9716,
+      "learning_rate": 1.6528919625527054e-05,
+      "loss": 0.8367,
+      "step": 9716
+    },
+    {
+      "epoch": 0.9718,
+      "learning_rate": 1.6523630029036924e-05,
+      "loss": 1.8432,
+      "step": 9718
+    },
+    {
+      "epoch": 0.972,
+      "learning_rate": 1.6518337253008794e-05,
+      "loss": 0.1677,
+      "step": 9720
+    },
+    {
+      "epoch": 0.9722,
+      "learning_rate": 1.651304130002226e-05,
+      "loss": 1.003,
+      "step": 9722
+    },
+    {
+      "epoch": 0.9724,
+      "learning_rate": 1.6507742172658504e-05,
+      "loss": 0.681,
+      "step": 9724
+    },
+    {
+      "epoch": 0.9726,
+      "learning_rate": 1.6502439873500294e-05,
+      "loss": 0.7792,
+      "step": 9726
+    },
+    {
+      "epoch": 0.9728,
+      "learning_rate": 1.6497134405131872e-05,
+      "loss": 0.5827,
+      "step": 9728
+    },
+    {
+      "epoch": 0.973,
+      "learning_rate": 1.6491825770139058e-05,
+      "loss": 1.1028,
+      "step": 9730
+    },
+    {
+      "epoch": 0.9732,
+      "learning_rate": 1.6486513971109248e-05,
+      "loss": 0.5318,
+      "step": 9732
+    },
+    {
+      "epoch": 0.9734,
+      "learning_rate": 1.6481199010631305e-05,
+      "loss": 0.6158,
+      "step": 9734
+    },
+    {
+      "epoch": 0.9736,
+      "learning_rate": 1.647588089129571e-05,
+      "loss": 0.3893,
+      "step": 9736
+    },
+    {
+      "epoch": 0.9738,
+      "learning_rate": 1.6470559615694455e-05,
+      "loss": 1.0409,
+      "step": 9738
+    },
+    {
+      "epoch": 0.974,
+      "learning_rate": 1.6465235186421017e-05,
+      "loss": 0.8614,
+      "step": 9740
+    },
+    {
+      "epoch": 0.9742,
+      "learning_rate": 1.645990760607052e-05,
+      "loss": 0.5936,
+      "step": 9742
+    },
+    {
+      "epoch": 0.9744,
+      "learning_rate": 1.6454576877239515e-05,
+      "loss": 1.0551,
+      "step": 9744
+    },
+    {
+      "epoch": 0.9746,
+      "learning_rate": 1.644924300252614e-05,
+      "loss": 0.4149,
+      "step": 9746
+    },
+    {
+      "epoch": 0.9748,
+      "learning_rate": 1.64439059845301e-05,
+      "loss": 0.9489,
+      "step": 9748
+    },
+    {
+      "epoch": 0.975,
+      "learning_rate": 1.643856582585255e-05,
+      "loss": 0.5818,
+      "step": 9750
+    },
+    {
+      "epoch": 0.9752,
+      "learning_rate": 1.6433222529096217e-05,
+      "loss": 0.9421,
+      "step": 9752
+    },
+    {
+      "epoch": 0.9754,
+      "learning_rate": 1.6427876096865407e-05,
+      "loss": 0.1144,
+      "step": 9754
+    },
+    {
+      "epoch": 0.9756,
+      "learning_rate": 1.6422526531765856e-05,
+      "loss": 0.4318,
+      "step": 9756
+    },
+    {
+      "epoch": 0.9758,
+      "learning_rate": 1.641717383640488e-05,
+      "loss": 0.2691,
+      "step": 9758
+    },
+    {
+      "epoch": 0.976,
+      "learning_rate": 1.6411818013391367e-05,
+      "loss": 0.9003,
+      "step": 9760
+    },
+    {
+      "epoch": 0.9762,
+      "learning_rate": 1.6406459065335616e-05,
+      "loss": 1.0966,
+      "step": 9762
+    },
+    {
+      "epoch": 0.9764,
+      "learning_rate": 1.6401096994849558e-05,
+      "loss": 0.7908,
+      "step": 9764
+    },
+    {
+      "epoch": 0.9766,
+      "learning_rate": 1.6395731804546596e-05,
+      "loss": 0.7986,
+      "step": 9766
+    },
+    {
+      "epoch": 0.9768,
+      "learning_rate": 1.639036349704162e-05,
+      "loss": 0.4683,
+      "step": 9768
+    },
+    {
+      "epoch": 0.977,
+      "learning_rate": 1.6384992074951128e-05,
+      "loss": 0.4501,
+      "step": 9770
+    },
+    {
+      "epoch": 0.9772,
+      "learning_rate": 1.6379617540893073e-05,
+      "loss": 0.2391,
+      "step": 9772
+    },
+    {
+      "epoch": 0.9774,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 0.6412,
+      "step": 9774
+    },
+    {
+      "epoch": 0.9776,
+      "learning_rate": 1.6368859147353653e-05,
+      "loss": 0.7396,
+      "step": 9776
+    },
+    {
+      "epoch": 0.9778,
+      "learning_rate": 1.6363475293115838e-05,
+      "loss": 0.6677,
+      "step": 9778
+    },
+    {
+      "epoch": 0.978,
+      "learning_rate": 1.6358088337397444e-05,
+      "loss": 1.004,
+      "step": 9780
+    },
+    {
+      "epoch": 0.9782,
+      "learning_rate": 1.6352698282824045e-05,
+      "loss": 0.107,
+      "step": 9782
+    },
+    {
+      "epoch": 0.9784,
+      "learning_rate": 1.6347305132022667e-05,
+      "loss": 0.2589,
+      "step": 9784
+    },
+    {
+      "epoch": 0.9786,
+      "learning_rate": 1.63419088876219e-05,
+      "loss": 0.9591,
+      "step": 9786
+    },
+    {
+      "epoch": 0.9788,
+      "learning_rate": 1.6336509552251773e-05,
+      "loss": 1.1466,
+      "step": 9788
+    },
+    {
+      "epoch": 0.979,
+      "learning_rate": 1.633110712854385e-05,
+      "loss": 1.0678,
+      "step": 9790
+    },
+    {
+      "epoch": 0.9792,
+      "learning_rate": 1.632570161913125e-05,
+      "loss": 0.6923,
+      "step": 9792
+    },
+    {
+      "epoch": 0.9794,
+      "learning_rate": 1.6320293026648515e-05,
+      "loss": 0.8819,
+      "step": 9794
+    },
+    {
+      "epoch": 0.9796,
+      "learning_rate": 1.6314881353731723e-05,
+      "loss": 0.4417,
+      "step": 9796
+    },
+    {
+      "epoch": 0.9798,
+      "learning_rate": 1.6309466603018504e-05,
+      "loss": 0.9312,
+      "step": 9798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6304048777147908e-05,
+      "loss": 0.8708,
+      "step": 9800
+    },
+    {
+      "epoch": 0.9802,
+      "learning_rate": 1.6298627878760495e-05,
+      "loss": 0.081,
+      "step": 9802
+    },
+    {
+      "epoch": 0.9804,
+      "learning_rate": 1.629320391049838e-05,
+      "loss": 0.2822,
+      "step": 9804
+    },
+    {
+      "epoch": 0.9806,
+      "learning_rate": 1.6287776875005148e-05,
+      "loss": 0.6508,
+      "step": 9806
+    },
+    {
+      "epoch": 0.9808,
+      "learning_rate": 1.6282346774925823e-05,
+      "loss": 1.0203,
+      "step": 9808
+    },
+    {
+      "epoch": 0.981,
+      "learning_rate": 1.6276913612907015e-05,
+      "loss": 0.6893,
+      "step": 9810
+    },
+    {
+      "epoch": 0.9812,
+      "learning_rate": 1.6271477391596747e-05,
+      "loss": 0.851,
+      "step": 9812
+    },
+    {
+      "epoch": 0.9814,
+      "learning_rate": 1.6266038113644612e-05,
+      "loss": 0.3177,
+      "step": 9814
+    },
+    {
+      "epoch": 0.9816,
+      "learning_rate": 1.6260595781701612e-05,
+      "loss": 0.7341,
+      "step": 9816
+    },
+    {
+      "epoch": 0.9818,
+      "learning_rate": 1.6255150398420266e-05,
+      "loss": 0.6488,
+      "step": 9818
+    },
+    {
+      "epoch": 0.982,
+      "learning_rate": 1.6249701966454632e-05,
+      "loss": 0.8733,
+      "step": 9820
+    },
+    {
+      "epoch": 0.9822,
+      "learning_rate": 1.624425048846017e-05,
+      "loss": 0.4712,
+      "step": 9822
+    },
+    {
+      "epoch": 0.9824,
+      "learning_rate": 1.6238795967093858e-05,
+      "loss": 1.1037,
+      "step": 9824
+    },
+    {
+      "epoch": 0.9826,
+      "learning_rate": 1.623333840501421e-05,
+      "loss": 0.8822,
+      "step": 9826
+    },
+    {
+      "epoch": 0.9828,
+      "learning_rate": 1.6227877804881137e-05,
+      "loss": 0.9155,
+      "step": 9828
+    },
+    {
+      "epoch": 0.983,
+      "learning_rate": 1.6222414169356063e-05,
+      "loss": 1.0002,
+      "step": 9830
+    },
+    {
+      "epoch": 0.9832,
+      "learning_rate": 1.621694750110194e-05,
+      "loss": 1.179,
+      "step": 9832
+    },
+    {
+      "epoch": 0.9834,
+      "learning_rate": 1.6211477802783102e-05,
+      "loss": 1.0244,
+      "step": 9834
+    },
+    {
+      "epoch": 0.9836,
+      "learning_rate": 1.6206005077065454e-05,
+      "loss": 0.4421,
+      "step": 9836
+    },
+    {
+      "epoch": 0.9838,
+      "learning_rate": 1.6200529326616343e-05,
+      "loss": 0.361,
+      "step": 9838
+    },
+    {
+      "epoch": 0.984,
+      "learning_rate": 1.6195050554104526e-05,
+      "loss": 0.4444,
+      "step": 9840
+    },
+    {
+      "epoch": 0.9842,
+      "learning_rate": 1.618956876220035e-05,
+      "loss": 0.4511,
+      "step": 9842
+    },
+    {
+      "epoch": 0.9844,
+      "learning_rate": 1.6184083953575557e-05,
+      "loss": 0.8795,
+      "step": 9844
+    },
+    {
+      "epoch": 0.9846,
+      "learning_rate": 1.6178596130903345e-05,
+      "loss": 0.5645,
+      "step": 9846
+    },
+    {
+      "epoch": 0.9848,
+      "learning_rate": 1.617310529685845e-05,
+      "loss": 1.0401,
+      "step": 9848
+    },
+    {
+      "epoch": 0.985,
+      "learning_rate": 1.616761145411704e-05,
+      "loss": 0.6114,
+      "step": 9850
+    },
+    {
+      "epoch": 0.9852,
+      "learning_rate": 1.6162114605356708e-05,
+      "loss": 0.7283,
+      "step": 9852
+    },
+    {
+      "epoch": 0.9854,
+      "learning_rate": 1.6156614753256587e-05,
+      "loss": 1.0077,
+      "step": 9854
+    },
+    {
+      "epoch": 0.9856,
+      "learning_rate": 1.615111190049721e-05,
+      "loss": 0.8827,
+      "step": 9856
+    },
+    {
+      "epoch": 0.9858,
+      "learning_rate": 1.6145606049760648e-05,
+      "loss": 0.6342,
+      "step": 9858
+    },
+    {
+      "epoch": 0.986,
+      "learning_rate": 1.6140097203730353e-05,
+      "loss": 1.1744,
+      "step": 9860
+    },
+    {
+      "epoch": 0.9862,
+      "learning_rate": 1.613458536509123e-05,
+      "loss": 0.6559,
+      "step": 9862
+    },
+    {
+      "epoch": 0.9864,
+      "learning_rate": 1.612907053652977e-05,
+      "loss": 0.9224,
+      "step": 9864
+    },
+    {
+      "epoch": 0.9866,
+      "learning_rate": 1.612355272073378e-05,
+      "loss": 1.0288,
+      "step": 9866
+    },
+    {
+      "epoch": 0.9868,
+      "learning_rate": 1.6118031920392562e-05,
+      "loss": 1.3632,
+      "step": 9868
+    },
+    {
+      "epoch": 0.987,
+      "learning_rate": 1.6112508138196922e-05,
+      "loss": 0.7955,
+      "step": 9870
+    },
+    {
+      "epoch": 0.9872,
+      "learning_rate": 1.6106981376839085e-05,
+      "loss": 0.6293,
+      "step": 9872
+    },
+    {
+      "epoch": 0.9874,
+      "learning_rate": 1.610145163901268e-05,
+      "loss": 0.7069,
+      "step": 9874
+    },
+    {
+      "epoch": 0.9876,
+      "learning_rate": 1.6095918927412886e-05,
+      "loss": 0.5608,
+      "step": 9876
+    },
+    {
+      "epoch": 0.9878,
+      "learning_rate": 1.6090383244736277e-05,
+      "loss": 0.4292,
+      "step": 9878
+    },
+    {
+      "epoch": 0.988,
+      "learning_rate": 1.608484459368083e-05,
+      "loss": 0.5379,
+      "step": 9880
+    },
+    {
+      "epoch": 0.9882,
+      "learning_rate": 1.6079302976946062e-05,
+      "loss": 1.0245,
+      "step": 9882
+    },
+    {
+      "epoch": 0.9884,
+      "learning_rate": 1.607375839723286e-05,
+      "loss": 0.8959,
+      "step": 9884
+    },
+    {
+      "epoch": 0.9886,
+      "learning_rate": 1.606821085724363e-05,
+      "loss": 0.473,
+      "step": 9886
+    },
+    {
+      "epoch": 0.9888,
+      "learning_rate": 1.606266035968213e-05,
+      "loss": 0.5757,
+      "step": 9888
+    },
+    {
+      "epoch": 0.989,
+      "learning_rate": 1.6057106907253607e-05,
+      "loss": 0.1738,
+      "step": 9890
+    },
+    {
+      "epoch": 0.9892,
+      "learning_rate": 1.605155050266479e-05,
+      "loss": 0.7203,
+      "step": 9892
+    },
+    {
+      "epoch": 0.9894,
+      "learning_rate": 1.6045991148623756e-05,
+      "loss": 0.2887,
+      "step": 9894
+    },
+    {
+      "epoch": 0.9896,
+      "learning_rate": 1.604042884784007e-05,
+      "loss": 0.5053,
+      "step": 9896
+    },
+    {
+      "epoch": 0.9898,
+      "learning_rate": 1.6034863603024775e-05,
+      "loss": 0.2253,
+      "step": 9898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.602929541689026e-05,
+      "loss": 1.3171,
+      "step": 9900
+    },
+    {
+      "epoch": 0.9902,
+      "learning_rate": 1.602372429215038e-05,
+      "loss": 1.1936,
+      "step": 9902
+    },
+    {
+      "epoch": 0.9904,
+      "learning_rate": 1.6018150231520496e-05,
+      "loss": 0.2954,
+      "step": 9904
+    },
+    {
+      "epoch": 0.9906,
+      "learning_rate": 1.6012573237717265e-05,
+      "loss": 0.0371,
+      "step": 9906
+    },
+    {
+      "epoch": 0.9908,
+      "learning_rate": 1.6006993313458893e-05,
+      "loss": 0.8737,
+      "step": 9908
+    },
+    {
+      "epoch": 0.991,
+      "learning_rate": 1.600141046146497e-05,
+      "loss": 0.2282,
+      "step": 9910
+    },
+    {
+      "epoch": 0.9912,
+      "learning_rate": 1.599582468445646e-05,
+      "loss": 0.075,
+      "step": 9912
+    },
+    {
+      "epoch": 0.9914,
+      "learning_rate": 1.5990235985155856e-05,
+      "loss": 0.6726,
+      "step": 9914
+    },
+    {
+      "epoch": 0.9916,
+      "learning_rate": 1.5984644366287018e-05,
+      "loss": 1.0456,
+      "step": 9916
+    },
+    {
+      "epoch": 0.9918,
+      "learning_rate": 1.597904983057519e-05,
+      "loss": 0.8383,
+      "step": 9918
+    },
+    {
+      "epoch": 0.992,
+      "learning_rate": 1.5973452380747135e-05,
+      "loss": 0.8429,
+      "step": 9920
+    },
+    {
+      "epoch": 0.9922,
+      "learning_rate": 1.5967852019530942e-05,
+      "loss": 0.4982,
+      "step": 9922
+    },
+    {
+      "epoch": 0.9924,
+      "learning_rate": 1.5962248749656158e-05,
+      "loss": 0.4071,
+      "step": 9924
+    },
+    {
+      "epoch": 0.9926,
+      "learning_rate": 1.5956642573853794e-05,
+      "loss": 0.6326,
+      "step": 9926
+    },
+    {
+      "epoch": 0.9928,
+      "learning_rate": 1.595103349485617e-05,
+      "loss": 0.516,
+      "step": 9928
+    },
+    {
+      "epoch": 0.993,
+      "learning_rate": 1.5945421515397135e-05,
+      "loss": 0.1301,
+      "step": 9930
+    },
+    {
+      "epoch": 0.9932,
+      "learning_rate": 1.5939806638211888e-05,
+      "loss": 1.2294,
+      "step": 9932
+    },
+    {
+      "epoch": 0.9934,
+      "learning_rate": 1.5934188866037014e-05,
+      "loss": 0.6136,
+      "step": 9934
+    },
+    {
+      "epoch": 0.9936,
+      "learning_rate": 1.5928568201610593e-05,
+      "loss": 0.8199,
+      "step": 9936
+    },
+    {
+      "epoch": 0.9938,
+      "learning_rate": 1.5922944647672068e-05,
+      "loss": 0.5391,
+      "step": 9938
+    },
+    {
+      "epoch": 0.994,
+      "learning_rate": 1.591731820696224e-05,
+      "loss": 0.7408,
+      "step": 9940
+    },
+    {
+      "epoch": 0.9942,
+      "learning_rate": 1.591168888222342e-05,
+      "loss": 1.2169,
+      "step": 9942
+    },
+    {
+      "epoch": 0.9944,
+      "learning_rate": 1.590605667619927e-05,
+      "loss": 0.9219,
+      "step": 9944
+    },
+    {
+      "epoch": 0.9946,
+      "learning_rate": 1.5900421591634816e-05,
+      "loss": 0.2186,
+      "step": 9946
+    },
+    {
+      "epoch": 0.9948,
+      "learning_rate": 1.589478363127657e-05,
+      "loss": 0.7159,
+      "step": 9948
+    },
+    {
+      "epoch": 0.995,
+      "learning_rate": 1.5889142797872407e-05,
+      "loss": 0.8122,
+      "step": 9950
+    },
+    {
+      "epoch": 0.9952,
+      "learning_rate": 1.5883499094171556e-05,
+      "loss": 0.8054,
+      "step": 9952
+    },
+    {
+      "epoch": 0.9954,
+      "learning_rate": 1.5877852522924736e-05,
+      "loss": 0.9825,
+      "step": 9954
+    },
+    {
+      "epoch": 0.9956,
+      "learning_rate": 1.587220308688399e-05,
+      "loss": 0.8515,
+      "step": 9956
+    },
+    {
+      "epoch": 0.9958,
+      "learning_rate": 1.5866550788802818e-05,
+      "loss": 1.1663,
+      "step": 9958
+    },
+    {
+      "epoch": 0.996,
+      "learning_rate": 1.586089563143605e-05,
+      "loss": 1.1685,
+      "step": 9960
+    },
+    {
+      "epoch": 0.9962,
+      "learning_rate": 1.5855237617539932e-05,
+      "loss": 0.6764,
+      "step": 9962
+    },
+    {
+      "epoch": 0.9964,
+      "learning_rate": 1.5849576749872162e-05,
+      "loss": 0.2241,
+      "step": 9964
+    },
+    {
+      "epoch": 0.9966,
+      "learning_rate": 1.584391303119173e-05,
+      "loss": 0.721,
+      "step": 9966
+    },
+    {
+      "epoch": 0.9968,
+      "learning_rate": 1.5838246464259063e-05,
+      "loss": 0.9108,
+      "step": 9968
+    },
+    {
+      "epoch": 0.997,
+      "learning_rate": 1.5832577051836023e-05,
+      "loss": 1.3745,
+      "step": 9970
+    },
+    {
+      "epoch": 0.9972,
+      "learning_rate": 1.582690479668577e-05,
+      "loss": 1.2047,
+      "step": 9972
+    },
+    {
+      "epoch": 0.9974,
+      "learning_rate": 1.582122970157289e-05,
+      "loss": 0.1463,
+      "step": 9974
+    },
+    {
+      "epoch": 0.9976,
+      "learning_rate": 1.5815551769263397e-05,
+      "loss": 0.4587,
+      "step": 9976
+    },
+    {
+      "epoch": 0.9978,
+      "learning_rate": 1.5809871002524592e-05,
+      "loss": 0.5188,
+      "step": 9978
+    },
+    {
+      "epoch": 0.998,
+      "learning_rate": 1.5804187404125272e-05,
+      "loss": 0.6264,
+      "step": 9980
+    },
+    {
+      "epoch": 0.9982,
+      "learning_rate": 1.5798500976835503e-05,
+      "loss": 0.6184,
+      "step": 9982
+    },
+    {
+      "epoch": 0.9984,
+      "learning_rate": 1.5792811723426784e-05,
+      "loss": 0.4618,
+      "step": 9984
+    },
+    {
+      "epoch": 0.9986,
+      "learning_rate": 1.5787119646672032e-05,
+      "loss": 1.4253,
+      "step": 9986
+    },
+    {
+      "epoch": 0.9988,
+      "learning_rate": 1.5781424749345454e-05,
+      "loss": 0.0065,
+      "step": 9988
+    },
+    {
+      "epoch": 0.999,
+      "learning_rate": 1.577572703422267e-05,
+      "loss": 0.084,
+      "step": 9990
+    },
+    {
+      "epoch": 0.9992,
+      "learning_rate": 1.5770026504080726e-05,
+      "loss": 0.3571,
+      "step": 9992
+    },
+    {
+      "epoch": 0.9994,
+      "learning_rate": 1.5764323161697946e-05,
+      "loss": 0.2175,
+      "step": 9994
+    },
+    {
+      "epoch": 0.9996,
+      "learning_rate": 1.5758617009854065e-05,
+      "loss": 0.1345,
+      "step": 9996
+    },
+    {
+      "epoch": 0.9998,
+      "learning_rate": 1.575290805133024e-05,
+      "loss": 1.0397,
+      "step": 9998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5747196288908887e-05,
+      "loss": 0.6098,
+      "step": 10000
+    },
+    {
+      "epoch": 1.0,
+      "step": 10000,
+      "total_flos": 1.2860345316068557e+17,
+      "train_loss": 0.6640387017915317,
+      "train_runtime": 28170.0443,
+      "train_samples_per_second": 5.68,
+      "train_steps_per_second": 0.355
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 10000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {},
+  "total_flos": 1.2860345316068557e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round0.pth b/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..101d4b4a9377879494bb49672b07bb42df15e129
--- /dev/null
+++ b/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b3f6ad2e5a7683790c3aba0014b64f994306e898bd07ecdd9243c08448a88e74
+size 1279587682
diff --git a/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round1.pth b/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..70e38be00f623520bbdb3e336697561684f4875c
--- /dev/null
+++ b/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:32d206dd0a6657ca89682c8133ed2e7dcaa54d2fa4e3ae583df7163256adaf2b
+size 1279587682
diff --git a/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round2.pth b/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..67da6a3e09847ff4b2b05ebeb056167075ced60b
--- /dev/null
+++ b/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7d877fea4f89a0abc71caa74b2a93b836c703ba686af4beae291e4e412747603
+size 1279587682
diff --git a/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round3.pth b/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..3dd1321391d9a45535f9c160affca070252aedae
--- /dev/null
+++ b/client_states_v7_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_125_selfsup_scenario16_10000_random0_0625_seed1/server_model_round3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c019f42f5297dce9a0ef306a07bb2f8bd79b56dbbe17b616b713f7e2f0c6c8a6
+size 1279587682