diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/0_trainer_state.json b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/0_trainer_state.json
new file mode 100644
index 0000000000000000000000000000000000000000..db73ae0742006e86d8f066c868149ea192314bb7
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/0_trainer_state.json
@@ -0,0 +1,15032 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 5000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0004,
+      "learning_rate": 2.3485380412522497e-06,
+      "loss": 0.4978,
+      "step": 2
+    },
+    {
+      "epoch": 0.0008,
+      "learning_rate": 2.357535430610912e-06,
+      "loss": 0.2848,
+      "step": 4
+    },
+    {
+      "epoch": 0.0012,
+      "learning_rate": 2.366547719345306e-06,
+      "loss": 0.5063,
+      "step": 6
+    },
+    {
+      "epoch": 0.0016,
+      "learning_rate": 2.3755748898855234e-06,
+      "loss": 0.2269,
+      "step": 8
+    },
+    {
+      "epoch": 0.002,
+      "learning_rate": 2.3846169246326332e-06,
+      "loss": 0.789,
+      "step": 10
+    },
+    {
+      "epoch": 0.0024,
+      "learning_rate": 2.3936738059587174e-06,
+      "loss": 0.357,
+      "step": 12
+    },
+    {
+      "epoch": 0.0028,
+      "learning_rate": 2.4027455162069537e-06,
+      "loss": 0.7059,
+      "step": 14
+    },
+    {
+      "epoch": 0.0032,
+      "learning_rate": 2.411832037691545e-06,
+      "loss": 0.1656,
+      "step": 16
+    },
+    {
+      "epoch": 0.0036,
+      "learning_rate": 2.420933352697865e-06,
+      "loss": 0.6619,
+      "step": 18
+    },
+    {
+      "epoch": 0.004,
+      "learning_rate": 2.430049443482434e-06,
+      "loss": 0.2676,
+      "step": 20
+    },
+    {
+      "epoch": 0.0044,
+      "learning_rate": 2.439180292272967e-06,
+      "loss": 1.0847,
+      "step": 22
+    },
+    {
+      "epoch": 0.0048,
+      "learning_rate": 2.448325881268406e-06,
+      "loss": 0.3018,
+      "step": 24
+    },
+    {
+      "epoch": 0.0052,
+      "learning_rate": 2.457486192638958e-06,
+      "loss": 0.1407,
+      "step": 26
+    },
+    {
+      "epoch": 0.0056,
+      "learning_rate": 2.4666612085261277e-06,
+      "loss": 0.4019,
+      "step": 28
+    },
+    {
+      "epoch": 0.006,
+      "learning_rate": 2.475850911042752e-06,
+      "loss": 0.3209,
+      "step": 30
+    },
+    {
+      "epoch": 0.0064,
+      "learning_rate": 2.4850552822730346e-06,
+      "loss": 0.2196,
+      "step": 32
+    },
+    {
+      "epoch": 0.0068,
+      "learning_rate": 2.4942743042725836e-06,
+      "loss": 0.4247,
+      "step": 34
+    },
+    {
+      "epoch": 0.0072,
+      "learning_rate": 2.503507959068455e-06,
+      "loss": 0.4194,
+      "step": 36
+    },
+    {
+      "epoch": 0.0076,
+      "learning_rate": 2.5127562286591313e-06,
+      "loss": 0.659,
+      "step": 38
+    },
+    {
+      "epoch": 0.008,
+      "learning_rate": 2.522019095014686e-06,
+      "loss": 0.4145,
+      "step": 40
+    },
+    {
+      "epoch": 0.0084,
+      "learning_rate": 2.531296540076638e-06,
+      "loss": 0.571,
+      "step": 42
+    },
+    {
+      "epoch": 0.0088,
+      "learning_rate": 2.5405885457581814e-06,
+      "loss": 0.3624,
+      "step": 44
+    },
+    {
+      "epoch": 0.0092,
+      "learning_rate": 2.5498950939440413e-06,
+      "loss": 0.526,
+      "step": 46
+    },
+    {
+      "epoch": 0.0096,
+      "learning_rate": 2.5592161664906243e-06,
+      "loss": 0.6059,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5685517452260587e-06,
+      "loss": 0.4984,
+      "step": 50
+    },
+    {
+      "epoch": 0.0104,
+      "learning_rate": 2.5779018119501086e-06,
+      "loss": 0.8004,
+      "step": 52
+    },
+    {
+      "epoch": 0.0108,
+      "learning_rate": 2.5872663484343887e-06,
+      "loss": 1.0366,
+      "step": 54
+    },
+    {
+      "epoch": 0.0112,
+      "learning_rate": 2.596645336422219e-06,
+      "loss": 0.4256,
+      "step": 56
+    },
+    {
+      "epoch": 0.0116,
+      "learning_rate": 2.606038757628795e-06,
+      "loss": 0.081,
+      "step": 58
+    },
+    {
+      "epoch": 0.012,
+      "learning_rate": 2.615446593741161e-06,
+      "loss": 0.1795,
+      "step": 60
+    },
+    {
+      "epoch": 0.0124,
+      "learning_rate": 2.6248688264182588e-06,
+      "loss": 0.2073,
+      "step": 62
+    },
+    {
+      "epoch": 0.0128,
+      "learning_rate": 2.6343054372909648e-06,
+      "loss": 0.6551,
+      "step": 64
+    },
+    {
+      "epoch": 0.0132,
+      "learning_rate": 2.6437564079621235e-06,
+      "loss": 0.2136,
+      "step": 66
+    },
+    {
+      "epoch": 0.0136,
+      "learning_rate": 2.6532217200065826e-06,
+      "loss": 0.4476,
+      "step": 68
+    },
+    {
+      "epoch": 0.014,
+      "learning_rate": 2.662701354971232e-06,
+      "loss": 0.8091,
+      "step": 70
+    },
+    {
+      "epoch": 0.0144,
+      "learning_rate": 2.6721952943750396e-06,
+      "loss": 0.3775,
+      "step": 72
+    },
+    {
+      "epoch": 0.0148,
+      "learning_rate": 2.6817035197090825e-06,
+      "loss": 0.5697,
+      "step": 74
+    },
+    {
+      "epoch": 0.0152,
+      "learning_rate": 2.691226012436604e-06,
+      "loss": 0.4306,
+      "step": 76
+    },
+    {
+      "epoch": 0.0156,
+      "learning_rate": 2.7007627539929783e-06,
+      "loss": 0.2091,
+      "step": 78
+    },
+    {
+      "epoch": 0.016,
+      "learning_rate": 2.7103137257858893e-06,
+      "loss": 0.5239,
+      "step": 80
+    },
+    {
+      "epoch": 0.0164,
+      "learning_rate": 2.7198789091951806e-06,
+      "loss": 0.3424,
+      "step": 82
+    },
+    {
+      "epoch": 0.0168,
+      "learning_rate": 2.7294582855730733e-06,
+      "loss": 0.4204,
+      "step": 84
+    },
+    {
+      "epoch": 0.0172,
+      "learning_rate": 2.7390518362440843e-06,
+      "loss": 0.5054,
+      "step": 86
+    },
+    {
+      "epoch": 0.0176,
+      "learning_rate": 2.7486595425050566e-06,
+      "loss": 0.1196,
+      "step": 88
+    },
+    {
+      "epoch": 0.018,
+      "learning_rate": 2.7582813856253264e-06,
+      "loss": 0.4632,
+      "step": 90
+    },
+    {
+      "epoch": 0.0184,
+      "learning_rate": 2.7679173468465813e-06,
+      "loss": 0.3133,
+      "step": 92
+    },
+    {
+      "epoch": 0.0188,
+      "learning_rate": 2.777567407383033e-06,
+      "loss": 0.2941,
+      "step": 94
+    },
+    {
+      "epoch": 0.0192,
+      "learning_rate": 2.7872315484213954e-06,
+      "loss": 0.4177,
+      "step": 96
+    },
+    {
+      "epoch": 0.0196,
+      "learning_rate": 2.796909751120931e-06,
+      "loss": 0.4319,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8066019966134873e-06,
+      "loss": 0.5043,
+      "step": 100
+    },
+    {
+      "epoch": 0.0204,
+      "learning_rate": 2.816308266003538e-06,
+      "loss": 0.4153,
+      "step": 102
+    },
+    {
+      "epoch": 0.0208,
+      "learning_rate": 2.826028540368212e-06,
+      "loss": 0.4126,
+      "step": 104
+    },
+    {
+      "epoch": 0.0212,
+      "learning_rate": 2.835762800757338e-06,
+      "loss": 0.7709,
+      "step": 106
+    },
+    {
+      "epoch": 0.0216,
+      "learning_rate": 2.845511028193477e-06,
+      "loss": 0.5181,
+      "step": 108
+    },
+    {
+      "epoch": 0.022,
+      "learning_rate": 2.855273203671962e-06,
+      "loss": 0.433,
+      "step": 110
+    },
+    {
+      "epoch": 0.0224,
+      "learning_rate": 2.865049308160931e-06,
+      "loss": 0.5872,
+      "step": 112
+    },
+    {
+      "epoch": 0.0228,
+      "learning_rate": 2.874839322601368e-06,
+      "loss": 0.5732,
+      "step": 114
+    },
+    {
+      "epoch": 0.0232,
+      "learning_rate": 2.8846432279071533e-06,
+      "loss": 0.2212,
+      "step": 116
+    },
+    {
+      "epoch": 0.0236,
+      "learning_rate": 2.8944610049650314e-06,
+      "loss": 0.6495,
+      "step": 118
+    },
+    {
+      "epoch": 0.024,
+      "learning_rate": 2.9042926346347835e-06,
+      "loss": 0.6217,
+      "step": 120
+    },
+    {
+      "epoch": 0.0244,
+      "learning_rate": 2.914138097749143e-06,
+      "loss": 0.3093,
+      "step": 122
+    },
+    {
+      "epoch": 0.0248,
+      "learning_rate": 2.9239973751138397e-06,
+      "loss": 0.3164,
+      "step": 124
+    },
+    {
+      "epoch": 0.0252,
+      "learning_rate": 2.933870447507756e-06,
+      "loss": 0.7215,
+      "step": 126
+    },
+    {
+      "epoch": 0.0256,
+      "learning_rate": 2.943757295682783e-06,
+      "loss": 0.2408,
+      "step": 128
+    },
+    {
+      "epoch": 0.026,
+      "learning_rate": 2.953657900364055e-06,
+      "loss": 0.3275,
+      "step": 130
+    },
+    {
+      "epoch": 0.0264,
+      "learning_rate": 2.9635722422497983e-06,
+      "loss": 0.6744,
+      "step": 132
+    },
+    {
+      "epoch": 0.0268,
+      "learning_rate": 2.973500302011496e-06,
+      "loss": 0.2212,
+      "step": 134
+    },
+    {
+      "epoch": 0.0272,
+      "learning_rate": 2.983442060293926e-06,
+      "loss": 0.6929,
+      "step": 136
+    },
+    {
+      "epoch": 0.0276,
+      "learning_rate": 2.9933974977150827e-06,
+      "loss": 0.1681,
+      "step": 138
+    },
+    {
+      "epoch": 0.028,
+      "learning_rate": 3.003366594866345e-06,
+      "loss": 0.4171,
+      "step": 140
+    },
+    {
+      "epoch": 0.0284,
+      "learning_rate": 3.0133493323124474e-06,
+      "loss": 0.3914,
+      "step": 142
+    },
+    {
+      "epoch": 0.0288,
+      "learning_rate": 3.0233456905915338e-06,
+      "loss": 0.1667,
+      "step": 144
+    },
+    {
+      "epoch": 0.0292,
+      "learning_rate": 3.0333556502151895e-06,
+      "loss": 0.3647,
+      "step": 146
+    },
+    {
+      "epoch": 0.0296,
+      "learning_rate": 3.0433791916684885e-06,
+      "loss": 0.843,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0534162954100234e-06,
+      "loss": 0.7318,
+      "step": 150
+    },
+    {
+      "epoch": 0.0304,
+      "learning_rate": 3.0634669418719453e-06,
+      "loss": 0.6392,
+      "step": 152
+    },
+    {
+      "epoch": 0.0308,
+      "learning_rate": 3.0735311114600064e-06,
+      "loss": 0.3443,
+      "step": 154
+    },
+    {
+      "epoch": 0.0312,
+      "learning_rate": 3.0836087845535933e-06,
+      "loss": 0.3958,
+      "step": 156
+    },
+    {
+      "epoch": 0.0316,
+      "learning_rate": 3.0936999415057645e-06,
+      "loss": 0.2644,
+      "step": 158
+    },
+    {
+      "epoch": 0.032,
+      "learning_rate": 3.1038045626432945e-06,
+      "loss": 0.5595,
+      "step": 160
+    },
+    {
+      "epoch": 0.0324,
+      "learning_rate": 3.1139226282667212e-06,
+      "loss": 0.5395,
+      "step": 162
+    },
+    {
+      "epoch": 0.0328,
+      "learning_rate": 3.1240541186503173e-06,
+      "loss": 0.3528,
+      "step": 164
+    },
+    {
+      "epoch": 0.0332,
+      "learning_rate": 3.134199014042277e-06,
+      "loss": 0.289,
+      "step": 166
+    },
+    {
+      "epoch": 0.0336,
+      "learning_rate": 3.1443572946645683e-06,
+      "loss": 1.0455,
+      "step": 168
+    },
+    {
+      "epoch": 0.034,
+      "learning_rate": 3.154528940713103e-06,
+      "loss": 0.5556,
+      "step": 170
+    },
+    {
+      "epoch": 0.0344,
+      "learning_rate": 3.164713932357776e-06,
+      "loss": 0.6263,
+      "step": 172
+    },
+    {
+      "epoch": 0.0348,
+      "learning_rate": 3.1749122497423724e-06,
+      "loss": 0.4976,
+      "step": 174
+    },
+    {
+      "epoch": 0.0352,
+      "learning_rate": 3.1851238729848033e-06,
+      "loss": 0.4342,
+      "step": 176
+    },
+    {
+      "epoch": 0.0356,
+      "learning_rate": 3.195348782176948e-06,
+      "loss": 0.2161,
+      "step": 178
+    },
+    {
+      "epoch": 0.036,
+      "learning_rate": 3.205586957384834e-06,
+      "loss": 0.189,
+      "step": 180
+    },
+    {
+      "epoch": 0.0364,
+      "learning_rate": 3.215838378648617e-06,
+      "loss": 0.2455,
+      "step": 182
+    },
+    {
+      "epoch": 0.0368,
+      "learning_rate": 3.2261030259826253e-06,
+      "loss": 0.4143,
+      "step": 184
+    },
+    {
+      "epoch": 0.0372,
+      "learning_rate": 3.2363808793754036e-06,
+      "loss": 0.2427,
+      "step": 186
+    },
+    {
+      "epoch": 0.0376,
+      "learning_rate": 3.246671918789752e-06,
+      "loss": 0.5408,
+      "step": 188
+    },
+    {
+      "epoch": 0.038,
+      "learning_rate": 3.2569761241627617e-06,
+      "loss": 1.2436,
+      "step": 190
+    },
+    {
+      "epoch": 0.0384,
+      "learning_rate": 3.267293475405858e-06,
+      "loss": 0.362,
+      "step": 192
+    },
+    {
+      "epoch": 0.0388,
+      "learning_rate": 3.277623952404835e-06,
+      "loss": 0.3556,
+      "step": 194
+    },
+    {
+      "epoch": 0.0392,
+      "learning_rate": 3.2879675350199004e-06,
+      "loss": 1.0699,
+      "step": 196
+    },
+    {
+      "epoch": 0.0396,
+      "learning_rate": 3.298324203085723e-06,
+      "loss": 0.5274,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.3086939364114113e-06,
+      "loss": 0.3893,
+      "step": 200
+    },
+    {
+      "epoch": 0.0404,
+      "learning_rate": 3.3190767147806892e-06,
+      "loss": 0.2916,
+      "step": 202
+    },
+    {
+      "epoch": 0.0408,
+      "learning_rate": 3.329472517951747e-06,
+      "loss": 0.4463,
+      "step": 204
+    },
+    {
+      "epoch": 0.0412,
+      "learning_rate": 3.3398813256574745e-06,
+      "loss": 0.6399,
+      "step": 206
+    },
+    {
+      "epoch": 0.0416,
+      "learning_rate": 3.350303117605369e-06,
+      "loss": 0.2346,
+      "step": 208
+    },
+    {
+      "epoch": 0.042,
+      "learning_rate": 3.360737873477574e-06,
+      "loss": 0.5752,
+      "step": 210
+    },
+    {
+      "epoch": 0.0424,
+      "learning_rate": 3.3711855729310503e-06,
+      "loss": 0.4147,
+      "step": 212
+    },
+    {
+      "epoch": 0.0428,
+      "learning_rate": 3.3816461955974224e-06,
+      "loss": 0.4813,
+      "step": 214
+    },
+    {
+      "epoch": 0.0432,
+      "learning_rate": 3.3921197210832235e-06,
+      "loss": 0.5642,
+      "step": 216
+    },
+    {
+      "epoch": 0.0436,
+      "learning_rate": 3.4026061289697397e-06,
+      "loss": 0.4102,
+      "step": 218
+    },
+    {
+      "epoch": 0.044,
+      "learning_rate": 3.4131053988131947e-06,
+      "loss": 0.3265,
+      "step": 220
+    },
+    {
+      "epoch": 0.0444,
+      "learning_rate": 3.4236175101447257e-06,
+      "loss": 0.422,
+      "step": 222
+    },
+    {
+      "epoch": 0.0448,
+      "learning_rate": 3.434142442470434e-06,
+      "loss": 0.4798,
+      "step": 224
+    },
+    {
+      "epoch": 0.0452,
+      "learning_rate": 3.444680175271424e-06,
+      "loss": 0.7324,
+      "step": 226
+    },
+    {
+      "epoch": 0.0456,
+      "learning_rate": 3.455230688003849e-06,
+      "loss": 0.2095,
+      "step": 228
+    },
+    {
+      "epoch": 0.046,
+      "learning_rate": 3.465793960098942e-06,
+      "loss": 0.3462,
+      "step": 230
+    },
+    {
+      "epoch": 0.0464,
+      "learning_rate": 3.476369970963065e-06,
+      "loss": 0.4735,
+      "step": 232
+    },
+    {
+      "epoch": 0.0468,
+      "learning_rate": 3.486958699977743e-06,
+      "loss": 0.7601,
+      "step": 234
+    },
+    {
+      "epoch": 0.0472,
+      "learning_rate": 3.497560126499706e-06,
+      "loss": 0.5506,
+      "step": 236
+    },
+    {
+      "epoch": 0.0476,
+      "learning_rate": 3.508174229860947e-06,
+      "loss": 0.2765,
+      "step": 238
+    },
+    {
+      "epoch": 0.048,
+      "learning_rate": 3.5188009893686836e-06,
+      "loss": 0.479,
+      "step": 240
+    },
+    {
+      "epoch": 0.0484,
+      "learning_rate": 3.5294403843055493e-06,
+      "loss": 0.468,
+      "step": 242
+    },
+    {
+      "epoch": 0.0488,
+      "learning_rate": 3.5400923939294827e-06,
+      "loss": 0.6509,
+      "step": 244
+    },
+    {
+      "epoch": 0.0492,
+      "learning_rate": 3.5507569974738477e-06,
+      "loss": 0.2147,
+      "step": 246
+    },
+    {
+      "epoch": 0.0496,
+      "learning_rate": 3.5614341741474667e-06,
+      "loss": 0.3087,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.5721239031345966e-06,
+      "loss": 0.2604,
+      "step": 250
+    },
+    {
+      "epoch": 0.0504,
+      "learning_rate": 3.5828261635951177e-06,
+      "loss": 0.5584,
+      "step": 252
+    },
+    {
+      "epoch": 0.0508,
+      "learning_rate": 3.593540934664387e-06,
+      "loss": 0.2994,
+      "step": 254
+    },
+    {
+      "epoch": 0.0512,
+      "learning_rate": 3.604268195453421e-06,
+      "loss": 0.5102,
+      "step": 256
+    },
+    {
+      "epoch": 0.0516,
+      "learning_rate": 3.6150079250488767e-06,
+      "loss": 0.44,
+      "step": 258
+    },
+    {
+      "epoch": 0.052,
+      "learning_rate": 3.6257601025130893e-06,
+      "loss": 0.2371,
+      "step": 260
+    },
+    {
+      "epoch": 0.0524,
+      "learning_rate": 3.636524706884178e-06,
+      "loss": 0.5082,
+      "step": 262
+    },
+    {
+      "epoch": 0.0528,
+      "learning_rate": 3.647301717175955e-06,
+      "loss": 0.3223,
+      "step": 264
+    },
+    {
+      "epoch": 0.0532,
+      "learning_rate": 3.6580911123781025e-06,
+      "loss": 0.1914,
+      "step": 266
+    },
+    {
+      "epoch": 0.0536,
+      "learning_rate": 3.66889287145614e-06,
+      "loss": 0.6891,
+      "step": 268
+    },
+    {
+      "epoch": 0.054,
+      "learning_rate": 3.679706973351488e-06,
+      "loss": 4.2481,
+      "step": 270
+    },
+    {
+      "epoch": 0.0544,
+      "learning_rate": 3.6905333969814995e-06,
+      "loss": 0.2092,
+      "step": 272
+    },
+    {
+      "epoch": 0.0548,
+      "learning_rate": 3.701372121239508e-06,
+      "loss": 0.6768,
+      "step": 274
+    },
+    {
+      "epoch": 0.0552,
+      "learning_rate": 3.712223124994867e-06,
+      "loss": 0.591,
+      "step": 276
+    },
+    {
+      "epoch": 0.0556,
+      "learning_rate": 3.723086387092989e-06,
+      "loss": 0.2471,
+      "step": 278
+    },
+    {
+      "epoch": 0.056,
+      "learning_rate": 3.7339618863553885e-06,
+      "loss": 0.1902,
+      "step": 280
+    },
+    {
+      "epoch": 0.0564,
+      "learning_rate": 3.744849601579722e-06,
+      "loss": 0.6435,
+      "step": 282
+    },
+    {
+      "epoch": 0.0568,
+      "learning_rate": 3.755749511539848e-06,
+      "loss": 0.3655,
+      "step": 284
+    },
+    {
+      "epoch": 0.0572,
+      "learning_rate": 3.7666615949857897e-06,
+      "loss": 0.3219,
+      "step": 286
+    },
+    {
+      "epoch": 0.0576,
+      "learning_rate": 3.7775858306439404e-06,
+      "loss": 0.4249,
+      "step": 288
+    },
+    {
+      "epoch": 0.058,
+      "learning_rate": 3.7885221972168864e-06,
+      "loss": 0.4868,
+      "step": 290
+    },
+    {
+      "epoch": 0.0584,
+      "learning_rate": 3.799470673383677e-06,
+      "loss": 0.8884,
+      "step": 292
+    },
+    {
+      "epoch": 0.0588,
+      "learning_rate": 3.810431237799657e-06,
+      "loss": 0.3619,
+      "step": 294
+    },
+    {
+      "epoch": 0.0592,
+      "learning_rate": 3.821403869096644e-06,
+      "loss": 0.2374,
+      "step": 296
+    },
+    {
+      "epoch": 0.0596,
+      "learning_rate": 3.8323885458829745e-06,
+      "loss": 0.3296,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8433852467434175e-06,
+      "loss": 0.2686,
+      "step": 300
+    },
+    {
+      "epoch": 0.0604,
+      "learning_rate": 3.854393950239355e-06,
+      "loss": 0.2137,
+      "step": 302
+    },
+    {
+      "epoch": 0.0608,
+      "learning_rate": 3.865414634908756e-06,
+      "loss": 0.4099,
+      "step": 304
+    },
+    {
+      "epoch": 0.0612,
+      "learning_rate": 3.876447279266233e-06,
+      "loss": 0.513,
+      "step": 306
+    },
+    {
+      "epoch": 0.0616,
+      "learning_rate": 3.887491861803081e-06,
+      "loss": 0.6536,
+      "step": 308
+    },
+    {
+      "epoch": 0.062,
+      "learning_rate": 3.898548360987321e-06,
+      "loss": 0.3845,
+      "step": 310
+    },
+    {
+      "epoch": 0.0624,
+      "learning_rate": 3.909616755263741e-06,
+      "loss": 0.4884,
+      "step": 312
+    },
+    {
+      "epoch": 0.0628,
+      "learning_rate": 3.920697023053941e-06,
+      "loss": 1.63,
+      "step": 314
+    },
+    {
+      "epoch": 0.0632,
+      "learning_rate": 3.9317891427563725e-06,
+      "loss": 0.4922,
+      "step": 316
+    },
+    {
+      "epoch": 0.0636,
+      "learning_rate": 3.942893092746381e-06,
+      "loss": 0.6452,
+      "step": 318
+    },
+    {
+      "epoch": 0.064,
+      "learning_rate": 3.954008851376244e-06,
+      "loss": 0.7281,
+      "step": 320
+    },
+    {
+      "epoch": 0.0644,
+      "learning_rate": 3.965136396975227e-06,
+      "loss": 0.7939,
+      "step": 322
+    },
+    {
+      "epoch": 0.0648,
+      "learning_rate": 3.976275707849619e-06,
+      "loss": 0.604,
+      "step": 324
+    },
+    {
+      "epoch": 0.0652,
+      "learning_rate": 3.987426762282726e-06,
+      "loss": 0.2484,
+      "step": 326
+    },
+    {
+      "epoch": 0.0656,
+      "learning_rate": 3.99858953853505e-06,
+      "loss": 0.2199,
+      "step": 328
+    },
+    {
+      "epoch": 0.066,
+      "learning_rate": 4.009764014844146e-06,
+      "loss": 0.1041,
+      "step": 330
+    },
+    {
+      "epoch": 0.0664,
+      "learning_rate": 4.0209501694248e-06,
+      "loss": 0.5506,
+      "step": 332
+    },
+    {
+      "epoch": 0.0668,
+      "learning_rate": 4.032147980469076e-06,
+      "loss": 0.5264,
+      "step": 334
+    },
+    {
+      "epoch": 0.0672,
+      "learning_rate": 4.043357426146209e-06,
+      "loss": 0.2475,
+      "step": 336
+    },
+    {
+      "epoch": 0.0676,
+      "learning_rate": 4.054578484602869e-06,
+      "loss": 0.7816,
+      "step": 338
+    },
+    {
+      "epoch": 0.068,
+      "learning_rate": 4.065811133962987e-06,
+      "loss": 0.4805,
+      "step": 340
+    },
+    {
+      "epoch": 0.0684,
+      "learning_rate": 4.07705535232795e-06,
+      "loss": 0.2808,
+      "step": 342
+    },
+    {
+      "epoch": 0.0688,
+      "learning_rate": 4.08831111777658e-06,
+      "loss": 0.3566,
+      "step": 344
+    },
+    {
+      "epoch": 0.0692,
+      "learning_rate": 4.0995784083651865e-06,
+      "loss": 0.3536,
+      "step": 346
+    },
+    {
+      "epoch": 0.0696,
+      "learning_rate": 4.110857202127611e-06,
+      "loss": 0.5355,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.122147477075266e-06,
+      "loss": 0.3157,
+      "step": 350
+    },
+    {
+      "epoch": 0.0704,
+      "learning_rate": 4.133449211197183e-06,
+      "loss": 0.3068,
+      "step": 352
+    },
+    {
+      "epoch": 0.0708,
+      "learning_rate": 4.144762382460055e-06,
+      "loss": 0.3316,
+      "step": 354
+    },
+    {
+      "epoch": 0.0712,
+      "learning_rate": 4.156086968808274e-06,
+      "loss": 0.8446,
+      "step": 356
+    },
+    {
+      "epoch": 0.0716,
+      "learning_rate": 4.1674229481639796e-06,
+      "loss": 0.7279,
+      "step": 358
+    },
+    {
+      "epoch": 0.072,
+      "learning_rate": 4.178770298427114e-06,
+      "loss": 0.594,
+      "step": 360
+    },
+    {
+      "epoch": 0.0724,
+      "learning_rate": 4.190128997475395e-06,
+      "loss": 0.6365,
+      "step": 362
+    },
+    {
+      "epoch": 0.0728,
+      "learning_rate": 4.201499023164515e-06,
+      "loss": 0.135,
+      "step": 364
+    },
+    {
+      "epoch": 0.0732,
+      "learning_rate": 4.212880353327968e-06,
+      "loss": 0.2319,
+      "step": 366
+    },
+    {
+      "epoch": 0.0736,
+      "learning_rate": 4.224272965777315e-06,
+      "loss": 0.5744,
+      "step": 368
+    },
+    {
+      "epoch": 0.074,
+      "learning_rate": 4.235676838302072e-06,
+      "loss": 0.2797,
+      "step": 370
+    },
+    {
+      "epoch": 0.0744,
+      "learning_rate": 4.247091948669764e-06,
+      "loss": 0.5988,
+      "step": 372
+    },
+    {
+      "epoch": 0.0748,
+      "learning_rate": 4.258518274626106e-06,
+      "loss": 0.216,
+      "step": 374
+    },
+    {
+      "epoch": 0.0752,
+      "learning_rate": 4.269955793894849e-06,
+      "loss": 0.3689,
+      "step": 376
+    },
+    {
+      "epoch": 0.0756,
+      "learning_rate": 4.281404484177978e-06,
+      "loss": 0.8246,
+      "step": 378
+    },
+    {
+      "epoch": 0.076,
+      "learning_rate": 4.292864323155684e-06,
+      "loss": 0.346,
+      "step": 380
+    },
+    {
+      "epoch": 0.0764,
+      "learning_rate": 4.304335288486412e-06,
+      "loss": 0.723,
+      "step": 382
+    },
+    {
+      "epoch": 0.0768,
+      "learning_rate": 4.3158173578069696e-06,
+      "loss": 0.5287,
+      "step": 384
+    },
+    {
+      "epoch": 0.0772,
+      "learning_rate": 4.327310508732434e-06,
+      "loss": 0.6834,
+      "step": 386
+    },
+    {
+      "epoch": 0.0776,
+      "learning_rate": 4.338814718856333e-06,
+      "loss": 0.3162,
+      "step": 388
+    },
+    {
+      "epoch": 0.078,
+      "learning_rate": 4.350329965750618e-06,
+      "loss": 0.3834,
+      "step": 390
+    },
+    {
+      "epoch": 0.0784,
+      "learning_rate": 4.3618562269657285e-06,
+      "loss": 0.1518,
+      "step": 392
+    },
+    {
+      "epoch": 0.0788,
+      "learning_rate": 4.373393480030629e-06,
+      "loss": 0.617,
+      "step": 394
+    },
+    {
+      "epoch": 0.0792,
+      "learning_rate": 4.384941702452852e-06,
+      "loss": 0.1557,
+      "step": 396
+    },
+    {
+      "epoch": 0.0796,
+      "learning_rate": 4.396500871718548e-06,
+      "loss": 0.254,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.408070965292526e-06,
+      "loss": 0.5976,
+      "step": 400
+    },
+    {
+      "epoch": 0.0804,
+      "learning_rate": 4.419651960618294e-06,
+      "loss": 0.52,
+      "step": 402
+    },
+    {
+      "epoch": 0.0808,
+      "learning_rate": 4.431243835118112e-06,
+      "loss": 0.3455,
+      "step": 404
+    },
+    {
+      "epoch": 0.0812,
+      "learning_rate": 4.442846566193041e-06,
+      "loss": 0.6479,
+      "step": 406
+    },
+    {
+      "epoch": 0.0816,
+      "learning_rate": 4.4544601312229185e-06,
+      "loss": 0.6547,
+      "step": 408
+    },
+    {
+      "epoch": 0.082,
+      "learning_rate": 4.4660845075665635e-06,
+      "loss": 0.3382,
+      "step": 410
+    },
+    {
+      "epoch": 0.0824,
+      "learning_rate": 4.477719672561602e-06,
+      "loss": 0.8742,
+      "step": 412
+    },
+    {
+      "epoch": 0.0828,
+      "learning_rate": 4.489365603524743e-06,
+      "loss": 0.3645,
+      "step": 414
+    },
+    {
+      "epoch": 0.0832,
+      "learning_rate": 4.501022277751605e-06,
+      "loss": 0.4448,
+      "step": 416
+    },
+    {
+      "epoch": 0.0836,
+      "learning_rate": 4.5126896725169025e-06,
+      "loss": 0.5673,
+      "step": 418
+    },
+    {
+      "epoch": 0.084,
+      "learning_rate": 4.524367765074499e-06,
+      "loss": 0.5706,
+      "step": 420
+    },
+    {
+      "epoch": 0.0844,
+      "learning_rate": 4.536056532657295e-06,
+      "loss": 0.423,
+      "step": 422
+    },
+    {
+      "epoch": 0.0848,
+      "learning_rate": 4.5477559524775e-06,
+      "loss": 0.3611,
+      "step": 424
+    },
+    {
+      "epoch": 0.0852,
+      "learning_rate": 4.559466001726451e-06,
+      "loss": 0.4763,
+      "step": 426
+    },
+    {
+      "epoch": 0.0856,
+      "learning_rate": 4.571186657574823e-06,
+      "loss": 1.0101,
+      "step": 428
+    },
+    {
+      "epoch": 0.086,
+      "learning_rate": 4.582917897172599e-06,
+      "loss": 0.8055,
+      "step": 430
+    },
+    {
+      "epoch": 0.0864,
+      "learning_rate": 4.5946596976491254e-06,
+      "loss": 0.1887,
+      "step": 432
+    },
+    {
+      "epoch": 0.0868,
+      "learning_rate": 4.6064120361131624e-06,
+      "loss": 0.5731,
+      "step": 434
+    },
+    {
+      "epoch": 0.0872,
+      "learning_rate": 4.618174889652924e-06,
+      "loss": 0.4192,
+      "step": 436
+    },
+    {
+      "epoch": 0.0876,
+      "learning_rate": 4.629948235336126e-06,
+      "loss": 0.575,
+      "step": 438
+    },
+    {
+      "epoch": 0.088,
+      "learning_rate": 4.6417320502100286e-06,
+      "loss": 0.6181,
+      "step": 440
+    },
+    {
+      "epoch": 0.0884,
+      "learning_rate": 4.653526311301479e-06,
+      "loss": 0.7818,
+      "step": 442
+    },
+    {
+      "epoch": 0.0888,
+      "learning_rate": 4.665330995616967e-06,
+      "loss": 0.4514,
+      "step": 444
+    },
+    {
+      "epoch": 0.0892,
+      "learning_rate": 4.677146080142667e-06,
+      "loss": 0.3001,
+      "step": 446
+    },
+    {
+      "epoch": 0.0896,
+      "learning_rate": 4.688971541844424e-06,
+      "loss": 0.6596,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.700807357667956e-06,
+      "loss": 0.3166,
+      "step": 450
+    },
+    {
+      "epoch": 0.0904,
+      "learning_rate": 4.712653504538672e-06,
+      "loss": 0.3794,
+      "step": 452
+    },
+    {
+      "epoch": 0.0908,
+      "learning_rate": 4.7245099593619495e-06,
+      "loss": 0.6721,
+      "step": 454
+    },
+    {
+      "epoch": 0.0912,
+      "learning_rate": 4.736376699023023e-06,
+      "loss": 0.8228,
+      "step": 456
+    },
+    {
+      "epoch": 0.0916,
+      "learning_rate": 4.74825370038703e-06,
+      "loss": 0.5651,
+      "step": 458
+    },
+    {
+      "epoch": 0.092,
+      "learning_rate": 4.76014094029921e-06,
+      "loss": 0.2414,
+      "step": 460
+    },
+    {
+      "epoch": 0.0924,
+      "learning_rate": 4.772038395584735e-06,
+      "loss": 0.3285,
+      "step": 462
+    },
+    {
+      "epoch": 0.0928,
+      "learning_rate": 4.7839460430489216e-06,
+      "loss": 0.5655,
+      "step": 464
+    },
+    {
+      "epoch": 0.0932,
+      "learning_rate": 4.7958638594772035e-06,
+      "loss": 0.2886,
+      "step": 466
+    },
+    {
+      "epoch": 0.0936,
+      "learning_rate": 4.807791821635185e-06,
+      "loss": 0.2587,
+      "step": 468
+    },
+    {
+      "epoch": 0.094,
+      "learning_rate": 4.8197299062686954e-06,
+      "loss": 0.3,
+      "step": 470
+    },
+    {
+      "epoch": 0.0944,
+      "learning_rate": 4.831678090103828e-06,
+      "loss": 0.5498,
+      "step": 472
+    },
+    {
+      "epoch": 0.0948,
+      "learning_rate": 4.8436363498469865e-06,
+      "loss": 0.3583,
+      "step": 474
+    },
+    {
+      "epoch": 0.0952,
+      "learning_rate": 4.855604662184931e-06,
+      "loss": 0.2844,
+      "step": 476
+    },
+    {
+      "epoch": 0.0956,
+      "learning_rate": 4.867583003784825e-06,
+      "loss": 0.428,
+      "step": 478
+    },
+    {
+      "epoch": 0.096,
+      "learning_rate": 4.8795713512942785e-06,
+      "loss": 0.3722,
+      "step": 480
+    },
+    {
+      "epoch": 0.0964,
+      "learning_rate": 4.891569681341395e-06,
+      "loss": 0.4665,
+      "step": 482
+    },
+    {
+      "epoch": 0.0968,
+      "learning_rate": 4.903577970534815e-06,
+      "loss": 0.3869,
+      "step": 484
+    },
+    {
+      "epoch": 0.0972,
+      "learning_rate": 4.91559619546378e-06,
+      "loss": 0.5019,
+      "step": 486
+    },
+    {
+      "epoch": 0.0976,
+      "learning_rate": 4.9276243326981e-06,
+      "loss": 0.2032,
+      "step": 488
+    },
+    {
+      "epoch": 0.098,
+      "learning_rate": 4.939662358788352e-06,
+      "loss": 0.6518,
+      "step": 490
+    },
+    {
+      "epoch": 0.0984,
+      "learning_rate": 4.951710250265788e-06,
+      "loss": 0.3309,
+      "step": 492
+    },
+    {
+      "epoch": 0.0988,
+      "learning_rate": 4.96376798364238e-06,
+      "loss": 0.6847,
+      "step": 494
+    },
+    {
+      "epoch": 0.0992,
+      "learning_rate": 4.975835535411023e-06,
+      "loss": 0.4409,
+      "step": 496
+    },
+    {
+      "epoch": 0.0996,
+      "learning_rate": 4.987912882045345e-06,
+      "loss": 0.8926,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.3192,
+      "step": 500
+    },
+    {
+      "epoch": 0.1004,
+      "learning_rate": 5.012096865710493e-06,
+      "loss": 0.6568,
+      "step": 502
+    },
+    {
+      "epoch": 0.1008,
+      "learning_rate": 5.024203455593375e-06,
+      "loss": 0.7142,
+      "step": 504
+    },
+    {
+      "epoch": 0.1012,
+      "learning_rate": 5.036319746046232e-06,
+      "loss": 0.3062,
+      "step": 506
+    },
+    {
+      "epoch": 0.1016,
+      "learning_rate": 5.048445713447734e-06,
+      "loss": 0.1935,
+      "step": 508
+    },
+    {
+      "epoch": 0.102,
+      "learning_rate": 5.0605813341576885e-06,
+      "loss": 0.3819,
+      "step": 510
+    },
+    {
+      "epoch": 0.1024,
+      "learning_rate": 5.072726584517083e-06,
+      "loss": 0.5336,
+      "step": 512
+    },
+    {
+      "epoch": 0.1028,
+      "learning_rate": 5.084881440848126e-06,
+      "loss": 0.3949,
+      "step": 514
+    },
+    {
+      "epoch": 0.1032,
+      "learning_rate": 5.097045879454308e-06,
+      "loss": 0.4412,
+      "step": 516
+    },
+    {
+      "epoch": 0.1036,
+      "learning_rate": 5.109219876620433e-06,
+      "loss": 0.1469,
+      "step": 518
+    },
+    {
+      "epoch": 0.104,
+      "learning_rate": 5.1214034086126685e-06,
+      "loss": 0.2903,
+      "step": 520
+    },
+    {
+      "epoch": 0.1044,
+      "learning_rate": 5.133596451678611e-06,
+      "loss": 0.6229,
+      "step": 522
+    },
+    {
+      "epoch": 0.1048,
+      "learning_rate": 5.145798982047253e-06,
+      "loss": 0.637,
+      "step": 524
+    },
+    {
+      "epoch": 0.1052,
+      "learning_rate": 5.158010975929185e-06,
+      "loss": 0.3293,
+      "step": 526
+    },
+    {
+      "epoch": 0.1056,
+      "learning_rate": 5.170232409516483e-06,
+      "loss": 0.213,
+      "step": 528
+    },
+    {
+      "epoch": 0.106,
+      "learning_rate": 5.182463258982837e-06,
+      "loss": 0.4401,
+      "step": 530
+    },
+    {
+      "epoch": 0.1064,
+      "learning_rate": 5.194703500483597e-06,
+      "loss": 0.2414,
+      "step": 532
+    },
+    {
+      "epoch": 0.1068,
+      "learning_rate": 5.2069531101557395e-06,
+      "loss": 0.2141,
+      "step": 534
+    },
+    {
+      "epoch": 0.1072,
+      "learning_rate": 5.219212064118082e-06,
+      "loss": 0.3803,
+      "step": 536
+    },
+    {
+      "epoch": 0.1076,
+      "learning_rate": 5.231480338471124e-06,
+      "loss": 0.1411,
+      "step": 538
+    },
+    {
+      "epoch": 0.108,
+      "learning_rate": 5.24375790929725e-06,
+      "loss": 0.323,
+      "step": 540
+    },
+    {
+      "epoch": 0.1084,
+      "learning_rate": 5.256044752660709e-06,
+      "loss": 0.2705,
+      "step": 542
+    },
+    {
+      "epoch": 0.1088,
+      "learning_rate": 5.268340844607653e-06,
+      "loss": 0.4348,
+      "step": 544
+    },
+    {
+      "epoch": 0.1092,
+      "learning_rate": 5.2806461611662725e-06,
+      "loss": 0.517,
+      "step": 546
+    },
+    {
+      "epoch": 0.1096,
+      "learning_rate": 5.2929606783466735e-06,
+      "loss": 0.3316,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 5.305284372141091e-06,
+      "loss": 0.474,
+      "step": 550
+    },
+    {
+      "epoch": 0.1104,
+      "learning_rate": 5.317617218523853e-06,
+      "loss": 0.5303,
+      "step": 552
+    },
+    {
+      "epoch": 0.1108,
+      "learning_rate": 5.3299591934514435e-06,
+      "loss": 0.6748,
+      "step": 554
+    },
+    {
+      "epoch": 0.1112,
+      "learning_rate": 5.342310272862553e-06,
+      "loss": 0.636,
+      "step": 556
+    },
+    {
+      "epoch": 0.1116,
+      "learning_rate": 5.354670432678119e-06,
+      "loss": 0.2121,
+      "step": 558
+    },
+    {
+      "epoch": 0.112,
+      "learning_rate": 5.367039648801377e-06,
+      "loss": 0.4483,
+      "step": 560
+    },
+    {
+      "epoch": 0.1124,
+      "learning_rate": 5.379417897117909e-06,
+      "loss": 0.2498,
+      "step": 562
+    },
+    {
+      "epoch": 0.1128,
+      "learning_rate": 5.391805153495684e-06,
+      "loss": 0.3031,
+      "step": 564
+    },
+    {
+      "epoch": 0.1132,
+      "learning_rate": 5.404201393785113e-06,
+      "loss": 0.4915,
+      "step": 566
+    },
+    {
+      "epoch": 0.1136,
+      "learning_rate": 5.416606593819109e-06,
+      "loss": 0.2193,
+      "step": 568
+    },
+    {
+      "epoch": 0.114,
+      "learning_rate": 5.429020729413049e-06,
+      "loss": 0.2939,
+      "step": 570
+    },
+    {
+      "epoch": 0.1144,
+      "learning_rate": 5.441443776365005e-06,
+      "loss": 0.52,
+      "step": 572
+    },
+    {
+      "epoch": 0.1148,
+      "learning_rate": 5.453875710455549e-06,
+      "loss": 0.2586,
+      "step": 574
+    },
+    {
+      "epoch": 0.1152,
+      "learning_rate": 5.466316507448053e-06,
+      "loss": 0.4845,
+      "step": 576
+    },
+    {
+      "epoch": 0.1156,
+      "learning_rate": 5.478766143088497e-06,
+      "loss": 0.4885,
+      "step": 578
+    },
+    {
+      "epoch": 0.116,
+      "learning_rate": 5.49122459310568e-06,
+      "loss": 0.5363,
+      "step": 580
+    },
+    {
+      "epoch": 0.1164,
+      "learning_rate": 5.503691833211264e-06,
+      "loss": 0.4225,
+      "step": 582
+    },
+    {
+      "epoch": 0.1168,
+      "learning_rate": 5.516167839099662e-06,
+      "loss": 0.3044,
+      "step": 584
+    },
+    {
+      "epoch": 0.1172,
+      "learning_rate": 5.5286525864483285e-06,
+      "loss": 0.5066,
+      "step": 586
+    },
+    {
+      "epoch": 0.1176,
+      "learning_rate": 5.5411460509175605e-06,
+      "loss": 0.3234,
+      "step": 588
+    },
+    {
+      "epoch": 0.118,
+      "learning_rate": 5.553648208150724e-06,
+      "loss": 0.3873,
+      "step": 590
+    },
+    {
+      "epoch": 0.1184,
+      "learning_rate": 5.5661590337742255e-06,
+      "loss": 0.2609,
+      "step": 592
+    },
+    {
+      "epoch": 0.1188,
+      "learning_rate": 5.57867850339757e-06,
+      "loss": 0.424,
+      "step": 594
+    },
+    {
+      "epoch": 0.1192,
+      "learning_rate": 5.591206592613412e-06,
+      "loss": 0.36,
+      "step": 596
+    },
+    {
+      "epoch": 0.1196,
+      "learning_rate": 5.603743276997597e-06,
+      "loss": 0.5183,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 5.616288532109221e-06,
+      "loss": 0.4514,
+      "step": 600
+    },
+    {
+      "epoch": 0.1204,
+      "learning_rate": 5.628842333490665e-06,
+      "loss": 0.0299,
+      "step": 602
+    },
+    {
+      "epoch": 0.1208,
+      "learning_rate": 5.641404656667652e-06,
+      "loss": 0.5307,
+      "step": 604
+    },
+    {
+      "epoch": 0.1212,
+      "learning_rate": 5.653975477149289e-06,
+      "loss": 0.2468,
+      "step": 606
+    },
+    {
+      "epoch": 0.1216,
+      "learning_rate": 5.666554770428136e-06,
+      "loss": 0.3562,
+      "step": 608
+    },
+    {
+      "epoch": 0.122,
+      "learning_rate": 5.679142511980168e-06,
+      "loss": 0.419,
+      "step": 610
+    },
+    {
+      "epoch": 0.1224,
+      "learning_rate": 5.6917386772650015e-06,
+      "loss": 0.4426,
+      "step": 612
+    },
+    {
+      "epoch": 0.1228,
+      "learning_rate": 5.7043432417257076e-06,
+      "loss": 0.1631,
+      "step": 614
+    },
+    {
+      "epoch": 0.1232,
+      "learning_rate": 5.716956180789086e-06,
+      "loss": 0.1677,
+      "step": 616
+    },
+    {
+      "epoch": 0.1236,
+      "learning_rate": 5.729577469865569e-06,
+      "loss": 0.2451,
+      "step": 618
+    },
+    {
+      "epoch": 0.124,
+      "learning_rate": 5.74220708434926e-06,
+      "loss": 0.4916,
+      "step": 620
+    },
+    {
+      "epoch": 0.1244,
+      "learning_rate": 5.754844999618143e-06,
+      "loss": 0.4607,
+      "step": 622
+    },
+    {
+      "epoch": 0.1248,
+      "learning_rate": 5.767491191033909e-06,
+      "loss": 0.1815,
+      "step": 624
+    },
+    {
+      "epoch": 0.1252,
+      "learning_rate": 5.780145633942173e-06,
+      "loss": 0.6035,
+      "step": 626
+    },
+    {
+      "epoch": 0.1256,
+      "learning_rate": 5.7928083036724535e-06,
+      "loss": 0.489,
+      "step": 628
+    },
+    {
+      "epoch": 0.126,
+      "learning_rate": 5.8054791755382125e-06,
+      "loss": 0.2393,
+      "step": 630
+    },
+    {
+      "epoch": 0.1264,
+      "learning_rate": 5.818158224836983e-06,
+      "loss": 0.4386,
+      "step": 632
+    },
+    {
+      "epoch": 0.1268,
+      "learning_rate": 5.830845426850263e-06,
+      "loss": 0.2538,
+      "step": 634
+    },
+    {
+      "epoch": 0.1272,
+      "learning_rate": 5.8435407568437194e-06,
+      "loss": 0.1257,
+      "step": 636
+    },
+    {
+      "epoch": 0.1276,
+      "learning_rate": 5.856244190067155e-06,
+      "loss": 1.1604,
+      "step": 638
+    },
+    {
+      "epoch": 0.128,
+      "learning_rate": 5.868955701754577e-06,
+      "loss": 0.7365,
+      "step": 640
+    },
+    {
+      "epoch": 0.1284,
+      "learning_rate": 5.881675267124245e-06,
+      "loss": 0.2199,
+      "step": 642
+    },
+    {
+      "epoch": 0.1288,
+      "learning_rate": 5.894402861378714e-06,
+      "loss": 0.3825,
+      "step": 644
+    },
+    {
+      "epoch": 0.1292,
+      "learning_rate": 5.907138459704886e-06,
+      "loss": 0.1964,
+      "step": 646
+    },
+    {
+      "epoch": 0.1296,
+      "learning_rate": 5.919882037274065e-06,
+      "loss": 0.2918,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 5.932633569241989e-06,
+      "loss": 0.355,
+      "step": 650
+    },
+    {
+      "epoch": 0.1304,
+      "learning_rate": 5.9453930307488985e-06,
+      "loss": 0.2742,
+      "step": 652
+    },
+    {
+      "epoch": 0.1308,
+      "learning_rate": 5.958160396919584e-06,
+      "loss": 0.5656,
+      "step": 654
+    },
+    {
+      "epoch": 0.1312,
+      "learning_rate": 5.970935642863362e-06,
+      "loss": 1.0181,
+      "step": 656
+    },
+    {
+      "epoch": 0.1316,
+      "learning_rate": 5.983718743674305e-06,
+      "loss": 0.4758,
+      "step": 658
+    },
+    {
+      "epoch": 0.132,
+      "learning_rate": 5.996509674431038e-06,
+      "loss": 0.7185,
+      "step": 660
+    },
+    {
+      "epoch": 0.1324,
+      "learning_rate": 6.00930841019705e-06,
+      "loss": 0.186,
+      "step": 662
+    },
+    {
+      "epoch": 0.1328,
+      "learning_rate": 6.022114926020505e-06,
+      "loss": 0.4552,
+      "step": 664
+    },
+    {
+      "epoch": 0.1332,
+      "learning_rate": 6.0349291969344426e-06,
+      "loss": 0.7673,
+      "step": 666
+    },
+    {
+      "epoch": 0.1336,
+      "learning_rate": 6.047751197956836e-06,
+      "loss": 0.0948,
+      "step": 668
+    },
+    {
+      "epoch": 0.134,
+      "learning_rate": 6.060580904090489e-06,
+      "loss": 0.4452,
+      "step": 670
+    },
+    {
+      "epoch": 0.1344,
+      "learning_rate": 6.0734182903232475e-06,
+      "loss": 0.2749,
+      "step": 672
+    },
+    {
+      "epoch": 0.1348,
+      "learning_rate": 6.086263331627974e-06,
+      "loss": 0.4888,
+      "step": 674
+    },
+    {
+      "epoch": 0.1352,
+      "learning_rate": 6.0991160029626e-06,
+      "loss": 0.5217,
+      "step": 676
+    },
+    {
+      "epoch": 0.1356,
+      "learning_rate": 6.111976279270187e-06,
+      "loss": 0.2678,
+      "step": 678
+    },
+    {
+      "epoch": 0.136,
+      "learning_rate": 6.124844135478966e-06,
+      "loss": 0.7836,
+      "step": 680
+    },
+    {
+      "epoch": 0.1364,
+      "learning_rate": 6.137719546502394e-06,
+      "loss": 0.3438,
+      "step": 682
+    },
+    {
+      "epoch": 0.1368,
+      "learning_rate": 6.1506024872392e-06,
+      "loss": 0.4353,
+      "step": 684
+    },
+    {
+      "epoch": 0.1372,
+      "learning_rate": 6.163492932573429e-06,
+      "loss": 0.4931,
+      "step": 686
+    },
+    {
+      "epoch": 0.1376,
+      "learning_rate": 6.176390857374501e-06,
+      "loss": 0.544,
+      "step": 688
+    },
+    {
+      "epoch": 0.138,
+      "learning_rate": 6.189296236497251e-06,
+      "loss": 0.3892,
+      "step": 690
+    },
+    {
+      "epoch": 0.1384,
+      "learning_rate": 6.202209044781979e-06,
+      "loss": 0.4959,
+      "step": 692
+    },
+    {
+      "epoch": 0.1388,
+      "learning_rate": 6.215129257054525e-06,
+      "loss": 0.2829,
+      "step": 694
+    },
+    {
+      "epoch": 0.1392,
+      "learning_rate": 6.228056848126223e-06,
+      "loss": 0.2992,
+      "step": 696
+    },
+    {
+      "epoch": 0.1396,
+      "learning_rate": 6.240991792794137e-06,
+      "loss": 0.2395,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 6.253934065840883e-06,
+      "loss": 0.1361,
+      "step": 700
+    },
+    {
+      "epoch": 0.1404,
+      "learning_rate": 6.2668836420348374e-06,
+      "loss": 0.532,
+      "step": 702
+    },
+    {
+      "epoch": 0.1408,
+      "learning_rate": 6.279840496130188e-06,
+      "loss": 0.2938,
+      "step": 704
+    },
+    {
+      "epoch": 0.1412,
+      "learning_rate": 6.2928046028668185e-06,
+      "loss": 0.3802,
+      "step": 706
+    },
+    {
+      "epoch": 0.1416,
+      "learning_rate": 6.305775936970606e-06,
+      "loss": 0.6094,
+      "step": 708
+    },
+    {
+      "epoch": 0.142,
+      "learning_rate": 6.3187544731532205e-06,
+      "loss": 0.4248,
+      "step": 710
+    },
+    {
+      "epoch": 0.1424,
+      "learning_rate": 6.331740186112359e-06,
+      "loss": 0.0424,
+      "step": 712
+    },
+    {
+      "epoch": 0.1428,
+      "learning_rate": 6.344733050531709e-06,
+      "loss": 0.5293,
+      "step": 714
+    },
+    {
+      "epoch": 0.1432,
+      "learning_rate": 6.357733041081015e-06,
+      "loss": 0.2113,
+      "step": 716
+    },
+    {
+      "epoch": 0.1436,
+      "learning_rate": 6.370740132416133e-06,
+      "loss": 0.4702,
+      "step": 718
+    },
+    {
+      "epoch": 0.144,
+      "learning_rate": 6.383754299179072e-06,
+      "loss": 0.5836,
+      "step": 720
+    },
+    {
+      "epoch": 0.1444,
+      "learning_rate": 6.3967755159980485e-06,
+      "loss": 0.3205,
+      "step": 722
+    },
+    {
+      "epoch": 0.1448,
+      "learning_rate": 6.409803757487532e-06,
+      "loss": 0.2761,
+      "step": 724
+    },
+    {
+      "epoch": 0.1452,
+      "learning_rate": 6.422838998248301e-06,
+      "loss": 0.5086,
+      "step": 726
+    },
+    {
+      "epoch": 0.1456,
+      "learning_rate": 6.435881212867485e-06,
+      "loss": 0.2804,
+      "step": 728
+    },
+    {
+      "epoch": 0.146,
+      "learning_rate": 6.4489303759186385e-06,
+      "loss": 0.1072,
+      "step": 730
+    },
+    {
+      "epoch": 0.1464,
+      "learning_rate": 6.4619864619616975e-06,
+      "loss": 0.3754,
+      "step": 732
+    },
+    {
+      "epoch": 0.1468,
+      "learning_rate": 6.475049445543222e-06,
+      "loss": 0.4856,
+      "step": 734
+    },
+    {
+      "epoch": 0.1472,
+      "learning_rate": 6.48811930119619e-06,
+      "loss": 0.2702,
+      "step": 736
+    },
+    {
+      "epoch": 0.1476,
+      "learning_rate": 6.5011960034403e-06,
+      "loss": 0.2769,
+      "step": 738
+    },
+    {
+      "epoch": 0.148,
+      "learning_rate": 6.514279526781853e-06,
+      "loss": 0.3243,
+      "step": 740
+    },
+    {
+      "epoch": 0.1484,
+      "learning_rate": 6.5273698457137965e-06,
+      "loss": 0.2323,
+      "step": 742
+    },
+    {
+      "epoch": 0.1488,
+      "learning_rate": 6.540466934715955e-06,
+      "loss": 0.5303,
+      "step": 744
+    },
+    {
+      "epoch": 0.1492,
+      "learning_rate": 6.553570768254831e-06,
+      "loss": 0.4626,
+      "step": 746
+    },
+    {
+      "epoch": 0.1496,
+      "learning_rate": 6.566681320783848e-06,
+      "loss": 0.2484,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 6.579798566743313e-06,
+      "loss": 0.5957,
+      "step": 750
+    },
+    {
+      "epoch": 0.1504,
+      "learning_rate": 6.592922480560483e-06,
+      "loss": 0.3162,
+      "step": 752
+    },
+    {
+      "epoch": 0.1508,
+      "learning_rate": 6.606053036649618e-06,
+      "loss": 0.2676,
+      "step": 754
+    },
+    {
+      "epoch": 0.1512,
+      "learning_rate": 6.619190209412025e-06,
+      "loss": 0.3487,
+      "step": 756
+    },
+    {
+      "epoch": 0.1516,
+      "learning_rate": 6.632333973236113e-06,
+      "loss": 0.3227,
+      "step": 758
+    },
+    {
+      "epoch": 0.152,
+      "learning_rate": 6.6454843024974465e-06,
+      "loss": 0.4049,
+      "step": 760
+    },
+    {
+      "epoch": 0.1524,
+      "learning_rate": 6.6586411715587805e-06,
+      "loss": 0.7003,
+      "step": 762
+    },
+    {
+      "epoch": 0.1528,
+      "learning_rate": 6.671804554770128e-06,
+      "loss": 0.4246,
+      "step": 764
+    },
+    {
+      "epoch": 0.1532,
+      "learning_rate": 6.6849744264688e-06,
+      "loss": 0.3153,
+      "step": 766
+    },
+    {
+      "epoch": 0.1536,
+      "learning_rate": 6.698150760979456e-06,
+      "loss": 0.2446,
+      "step": 768
+    },
+    {
+      "epoch": 0.154,
+      "learning_rate": 6.711333532614177e-06,
+      "loss": 0.4019,
+      "step": 770
+    },
+    {
+      "epoch": 0.1544,
+      "learning_rate": 6.724522715672421e-06,
+      "loss": 0.45,
+      "step": 772
+    },
+    {
+      "epoch": 0.1548,
+      "learning_rate": 6.737718284441256e-06,
+      "loss": 0.17,
+      "step": 774
+    },
+    {
+      "epoch": 0.1552,
+      "learning_rate": 6.750920213195242e-06,
+      "loss": 0.6598,
+      "step": 776
+    },
+    {
+      "epoch": 0.1556,
+      "learning_rate": 6.764128476196494e-06,
+      "loss": 0.1592,
+      "step": 778
+    },
+    {
+      "epoch": 0.156,
+      "learning_rate": 6.777343047694894e-06,
+      "loss": 0.7457,
+      "step": 780
+    },
+    {
+      "epoch": 0.1564,
+      "learning_rate": 6.7905639019278925e-06,
+      "loss": 0.4014,
+      "step": 782
+    },
+    {
+      "epoch": 0.1568,
+      "learning_rate": 6.803791013120824e-06,
+      "loss": 0.5695,
+      "step": 784
+    },
+    {
+      "epoch": 0.1572,
+      "learning_rate": 6.817024355486707e-06,
+      "loss": 0.3665,
+      "step": 786
+    },
+    {
+      "epoch": 0.1576,
+      "learning_rate": 6.8302639032264836e-06,
+      "loss": 0.6023,
+      "step": 788
+    },
+    {
+      "epoch": 0.158,
+      "learning_rate": 6.8435096305289765e-06,
+      "loss": 0.4421,
+      "step": 790
+    },
+    {
+      "epoch": 0.1584,
+      "learning_rate": 6.856761511570944e-06,
+      "loss": 0.1436,
+      "step": 792
+    },
+    {
+      "epoch": 0.1588,
+      "learning_rate": 6.870019520517217e-06,
+      "loss": 0.695,
+      "step": 794
+    },
+    {
+      "epoch": 0.1592,
+      "learning_rate": 6.883283631520579e-06,
+      "loss": 0.2095,
+      "step": 796
+    },
+    {
+      "epoch": 0.1596,
+      "learning_rate": 6.896553818721985e-06,
+      "loss": 0.3553,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 6.909830056250522e-06,
+      "loss": 0.3617,
+      "step": 800
+    },
+    {
+      "epoch": 0.1604,
+      "learning_rate": 6.9231123182234895e-06,
+      "loss": 0.4206,
+      "step": 802
+    },
+    {
+      "epoch": 0.1608,
+      "learning_rate": 6.936400578746436e-06,
+      "loss": 0.7615,
+      "step": 804
+    },
+    {
+      "epoch": 0.1612,
+      "learning_rate": 6.949694811913237e-06,
+      "loss": 0.3431,
+      "step": 806
+    },
+    {
+      "epoch": 0.1616,
+      "learning_rate": 6.96299499180605e-06,
+      "loss": 0.7773,
+      "step": 808
+    },
+    {
+      "epoch": 0.162,
+      "learning_rate": 6.976301092495548e-06,
+      "loss": 0.4262,
+      "step": 810
+    },
+    {
+      "epoch": 0.1624,
+      "learning_rate": 6.989613088040787e-06,
+      "loss": 0.5774,
+      "step": 812
+    },
+    {
+      "epoch": 0.1628,
+      "learning_rate": 7.002930952489353e-06,
+      "loss": 0.3014,
+      "step": 814
+    },
+    {
+      "epoch": 0.1632,
+      "learning_rate": 7.016254659877404e-06,
+      "loss": 0.5273,
+      "step": 816
+    },
+    {
+      "epoch": 0.1636,
+      "learning_rate": 7.029584184229641e-06,
+      "loss": 0.4042,
+      "step": 818
+    },
+    {
+      "epoch": 0.164,
+      "learning_rate": 7.042919499559539e-06,
+      "loss": 1.0242,
+      "step": 820
+    },
+    {
+      "epoch": 0.1644,
+      "learning_rate": 7.056260579869152e-06,
+      "loss": 0.3114,
+      "step": 822
+    },
+    {
+      "epoch": 0.1648,
+      "learning_rate": 7.06960739914943e-06,
+      "loss": 0.4621,
+      "step": 824
+    },
+    {
+      "epoch": 0.1652,
+      "learning_rate": 7.082959931380013e-06,
+      "loss": 0.2639,
+      "step": 826
+    },
+    {
+      "epoch": 0.1656,
+      "learning_rate": 7.09631815052946e-06,
+      "loss": 0.3438,
+      "step": 828
+    },
+    {
+      "epoch": 0.166,
+      "learning_rate": 7.109682030555285e-06,
+      "loss": 0.3132,
+      "step": 830
+    },
+    {
+      "epoch": 0.1664,
+      "learning_rate": 7.123051545403873e-06,
+      "loss": 0.4491,
+      "step": 832
+    },
+    {
+      "epoch": 0.1668,
+      "learning_rate": 7.136426669010686e-06,
+      "loss": 0.3661,
+      "step": 834
+    },
+    {
+      "epoch": 0.1672,
+      "learning_rate": 7.1498073753002375e-06,
+      "loss": 0.4484,
+      "step": 836
+    },
+    {
+      "epoch": 0.1676,
+      "learning_rate": 7.1631936381861544e-06,
+      "loss": 0.3452,
+      "step": 838
+    },
+    {
+      "epoch": 0.168,
+      "learning_rate": 7.1765854315712325e-06,
+      "loss": 0.5085,
+      "step": 840
+    },
+    {
+      "epoch": 0.1684,
+      "learning_rate": 7.189982729347485e-06,
+      "loss": 0.3839,
+      "step": 842
+    },
+    {
+      "epoch": 0.1688,
+      "learning_rate": 7.203385505396197e-06,
+      "loss": 0.2722,
+      "step": 844
+    },
+    {
+      "epoch": 0.1692,
+      "learning_rate": 7.216793733587966e-06,
+      "loss": 0.1636,
+      "step": 846
+    },
+    {
+      "epoch": 0.1696,
+      "learning_rate": 7.230207387782771e-06,
+      "loss": 0.3665,
+      "step": 848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 7.243626441830001e-06,
+      "loss": 0.3125,
+      "step": 850
+    },
+    {
+      "epoch": 0.1704,
+      "learning_rate": 7.257050869568527e-06,
+      "loss": 0.3114,
+      "step": 852
+    },
+    {
+      "epoch": 0.1708,
+      "learning_rate": 7.270480644826739e-06,
+      "loss": 0.4676,
+      "step": 854
+    },
+    {
+      "epoch": 0.1712,
+      "learning_rate": 7.28391574142262e-06,
+      "loss": 0.6019,
+      "step": 856
+    },
+    {
+      "epoch": 0.1716,
+      "learning_rate": 7.297356133163711e-06,
+      "loss": 0.5212,
+      "step": 858
+    },
+    {
+      "epoch": 0.172,
+      "learning_rate": 7.3108017938473485e-06,
+      "loss": 0.0636,
+      "step": 860
+    },
+    {
+      "epoch": 0.1724,
+      "learning_rate": 7.324252697260479e-06,
+      "loss": 0.122,
+      "step": 862
+    },
+    {
+      "epoch": 0.1728,
+      "learning_rate": 7.337708817179875e-06,
+      "loss": 0.2703,
+      "step": 864
+    },
+    {
+      "epoch": 0.1732,
+      "learning_rate": 7.351170127372196e-06,
+      "loss": 0.6272,
+      "step": 866
+    },
+    {
+      "epoch": 0.1736,
+      "learning_rate": 7.36463660159386e-06,
+      "loss": 0.8853,
+      "step": 868
+    },
+    {
+      "epoch": 0.174,
+      "learning_rate": 7.378108213591355e-06,
+      "loss": 0.3086,
+      "step": 870
+    },
+    {
+      "epoch": 0.1744,
+      "learning_rate": 7.39158493710103e-06,
+      "loss": 0.4001,
+      "step": 872
+    },
+    {
+      "epoch": 0.1748,
+      "learning_rate": 7.405066745849345e-06,
+      "loss": 0.3119,
+      "step": 874
+    },
+    {
+      "epoch": 0.1752,
+      "learning_rate": 7.418553613552822e-06,
+      "loss": 0.3373,
+      "step": 876
+    },
+    {
+      "epoch": 0.1756,
+      "learning_rate": 7.432045513918121e-06,
+      "loss": 0.3995,
+      "step": 878
+    },
+    {
+      "epoch": 0.176,
+      "learning_rate": 7.445542420642091e-06,
+      "loss": 0.3247,
+      "step": 880
+    },
+    {
+      "epoch": 0.1764,
+      "learning_rate": 7.459044307411826e-06,
+      "loss": 0.504,
+      "step": 882
+    },
+    {
+      "epoch": 0.1768,
+      "learning_rate": 7.472551147904703e-06,
+      "loss": 0.9268,
+      "step": 884
+    },
+    {
+      "epoch": 0.1772,
+      "learning_rate": 7.486062915788446e-06,
+      "loss": 0.1192,
+      "step": 886
+    },
+    {
+      "epoch": 0.1776,
+      "learning_rate": 7.499579584721173e-06,
+      "loss": 0.2752,
+      "step": 888
+    },
+    {
+      "epoch": 0.178,
+      "learning_rate": 7.513101128351446e-06,
+      "loss": 0.407,
+      "step": 890
+    },
+    {
+      "epoch": 0.1784,
+      "learning_rate": 7.5266275203183395e-06,
+      "loss": 0.4311,
+      "step": 892
+    },
+    {
+      "epoch": 0.1788,
+      "learning_rate": 7.540158734251412e-06,
+      "loss": 0.4275,
+      "step": 894
+    },
+    {
+      "epoch": 0.1792,
+      "learning_rate": 7.553694743770917e-06,
+      "loss": 1.0431,
+      "step": 896
+    },
+    {
+      "epoch": 0.1796,
+      "learning_rate": 7.567235522487698e-06,
+      "loss": 0.2753,
+      "step": 898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 7.580781044003312e-06,
+      "loss": 0.4615,
+      "step": 900
+    },
+    {
+      "epoch": 0.1804,
+      "learning_rate": 7.5943312819100875e-06,
+      "loss": 0.3807,
+      "step": 902
+    },
+    {
+      "epoch": 0.1808,
+      "learning_rate": 7.607886209791095e-06,
+      "loss": 1.419,
+      "step": 904
+    },
+    {
+      "epoch": 0.1812,
+      "learning_rate": 7.6214458012203726e-06,
+      "loss": 0.6087,
+      "step": 906
+    },
+    {
+      "epoch": 0.1816,
+      "learning_rate": 7.635010029762755e-06,
+      "loss": 0.3881,
+      "step": 908
+    },
+    {
+      "epoch": 0.182,
+      "learning_rate": 7.648578868974102e-06,
+      "loss": 0.3662,
+      "step": 910
+    },
+    {
+      "epoch": 0.1824,
+      "learning_rate": 7.662152292401265e-06,
+      "loss": 0.5038,
+      "step": 912
+    },
+    {
+      "epoch": 0.1828,
+      "learning_rate": 7.675730273582142e-06,
+      "loss": 1.2598,
+      "step": 914
+    },
+    {
+      "epoch": 0.1832,
+      "learning_rate": 7.689312786045822e-06,
+      "loss": 0.3086,
+      "step": 916
+    },
+    {
+      "epoch": 0.1836,
+      "learning_rate": 7.702899803312443e-06,
+      "loss": 0.3597,
+      "step": 918
+    },
+    {
+      "epoch": 0.184,
+      "learning_rate": 7.716491298893441e-06,
+      "loss": 0.4188,
+      "step": 920
+    },
+    {
+      "epoch": 0.1844,
+      "learning_rate": 7.730087246291498e-06,
+      "loss": 0.4932,
+      "step": 922
+    },
+    {
+      "epoch": 0.1848,
+      "learning_rate": 7.74368761900062e-06,
+      "loss": 0.5425,
+      "step": 924
+    },
+    {
+      "epoch": 0.1852,
+      "learning_rate": 7.757292390506184e-06,
+      "loss": 0.6096,
+      "step": 926
+    },
+    {
+      "epoch": 0.1856,
+      "learning_rate": 7.770901534284991e-06,
+      "loss": 0.6611,
+      "step": 928
+    },
+    {
+      "epoch": 0.186,
+      "learning_rate": 7.78451502380532e-06,
+      "loss": 0.4042,
+      "step": 930
+    },
+    {
+      "epoch": 0.1864,
+      "learning_rate": 7.798132832526976e-06,
+      "loss": 0.3537,
+      "step": 932
+    },
+    {
+      "epoch": 0.1868,
+      "learning_rate": 7.811754933901346e-06,
+      "loss": 0.6119,
+      "step": 934
+    },
+    {
+      "epoch": 0.1872,
+      "learning_rate": 7.825381301371444e-06,
+      "loss": 0.6406,
+      "step": 936
+    },
+    {
+      "epoch": 0.1876,
+      "learning_rate": 7.839011908371987e-06,
+      "loss": 0.7218,
+      "step": 938
+    },
+    {
+      "epoch": 0.188,
+      "learning_rate": 7.852646728329358e-06,
+      "loss": 0.348,
+      "step": 940
+    },
+    {
+      "epoch": 0.1884,
+      "learning_rate": 7.866285734661845e-06,
+      "loss": 0.1599,
+      "step": 942
+    },
+    {
+      "epoch": 0.1888,
+      "learning_rate": 7.879928900779441e-06,
+      "loss": 0.4448,
+      "step": 944
+    },
+    {
+      "epoch": 0.1892,
+      "learning_rate": 7.893576200084164e-06,
+      "loss": 0.1563,
+      "step": 946
+    },
+    {
+      "epoch": 0.1896,
+      "learning_rate": 7.907227605969852e-06,
+      "loss": 0.4625,
+      "step": 948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 7.92088309182239e-06,
+      "loss": 0.6913,
+      "step": 950
+    },
+    {
+      "epoch": 0.1904,
+      "learning_rate": 7.934542631019767e-06,
+      "loss": 0.4542,
+      "step": 952
+    },
+    {
+      "epoch": 0.1908,
+      "learning_rate": 7.948206196931937e-06,
+      "loss": 0.2976,
+      "step": 954
+    },
+    {
+      "epoch": 0.1912,
+      "learning_rate": 7.961873762921153e-06,
+      "loss": 0.3839,
+      "step": 956
+    },
+    {
+      "epoch": 0.1916,
+      "learning_rate": 7.97554530234174e-06,
+      "loss": 0.6925,
+      "step": 958
+    },
+    {
+      "epoch": 0.192,
+      "learning_rate": 7.989220788540351e-06,
+      "loss": 0.6349,
+      "step": 960
+    },
+    {
+      "epoch": 0.1924,
+      "learning_rate": 8.002900194855927e-06,
+      "loss": 0.4547,
+      "step": 962
+    },
+    {
+      "epoch": 0.1928,
+      "learning_rate": 8.016583494619764e-06,
+      "loss": 0.2978,
+      "step": 964
+    },
+    {
+      "epoch": 0.1932,
+      "learning_rate": 8.03027066115557e-06,
+      "loss": 0.2427,
+      "step": 966
+    },
+    {
+      "epoch": 0.1936,
+      "learning_rate": 8.043961667779511e-06,
+      "loss": 0.4806,
+      "step": 968
+    },
+    {
+      "epoch": 0.194,
+      "learning_rate": 8.057656487800274e-06,
+      "loss": 0.6885,
+      "step": 970
+    },
+    {
+      "epoch": 0.1944,
+      "learning_rate": 8.071355094519103e-06,
+      "loss": 0.5551,
+      "step": 972
+    },
+    {
+      "epoch": 0.1948,
+      "learning_rate": 8.085057461229862e-06,
+      "loss": 0.4917,
+      "step": 974
+    },
+    {
+      "epoch": 0.1952,
+      "learning_rate": 8.098763561219089e-06,
+      "loss": 0.1577,
+      "step": 976
+    },
+    {
+      "epoch": 0.1956,
+      "learning_rate": 8.112473367766056e-06,
+      "loss": 0.3893,
+      "step": 978
+    },
+    {
+      "epoch": 0.196,
+      "learning_rate": 8.126186854142744e-06,
+      "loss": 0.566,
+      "step": 980
+    },
+    {
+      "epoch": 0.1964,
+      "learning_rate": 8.139903993614075e-06,
+      "loss": 0.3859,
+      "step": 982
+    },
+    {
+      "epoch": 0.1968,
+      "learning_rate": 8.153624759437718e-06,
+      "loss": 0.2921,
+      "step": 984
+    },
+    {
+      "epoch": 0.1972,
+      "learning_rate": 8.167349124864389e-06,
+      "loss": 0.1714,
+      "step": 986
+    },
+    {
+      "epoch": 0.1976,
+      "learning_rate": 8.181077063137735e-06,
+      "loss": 0.6299,
+      "step": 988
+    },
+    {
+      "epoch": 0.198,
+      "learning_rate": 8.194808547494386e-06,
+      "loss": 0.478,
+      "step": 990
+    },
+    {
+      "epoch": 0.1984,
+      "learning_rate": 8.208543551164178e-06,
+      "loss": 0.6152,
+      "step": 992
+    },
+    {
+      "epoch": 0.1988,
+      "learning_rate": 8.22228204736997e-06,
+      "loss": 0.3155,
+      "step": 994
+    },
+    {
+      "epoch": 0.1992,
+      "learning_rate": 8.236024009327877e-06,
+      "loss": 0.3155,
+      "step": 996
+    },
+    {
+      "epoch": 0.1996,
+      "learning_rate": 8.249769410247239e-06,
+      "loss": 0.4525,
+      "step": 998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 8.263518223330695e-06,
+      "loss": 0.3657,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2004,
+      "learning_rate": 8.277270421774231e-06,
+      "loss": 0.5314,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2008,
+      "learning_rate": 8.29102597876723e-06,
+      "loss": 0.3303,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2012,
+      "learning_rate": 8.304784867492532e-06,
+      "loss": 0.2775,
+      "step": 1006
+    },
+    {
+      "epoch": 0.2016,
+      "learning_rate": 8.31854706112648e-06,
+      "loss": 0.297,
+      "step": 1008
+    },
+    {
+      "epoch": 0.202,
+      "learning_rate": 8.332312532838972e-06,
+      "loss": 0.6559,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2024,
+      "learning_rate": 8.346081255793516e-06,
+      "loss": 0.4066,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2028,
+      "learning_rate": 8.359853203147282e-06,
+      "loss": 0.6603,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2032,
+      "learning_rate": 8.373628348051156e-06,
+      "loss": 2.3754,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2036,
+      "learning_rate": 8.387406663649803e-06,
+      "loss": 0.5686,
+      "step": 1018
+    },
+    {
+      "epoch": 0.204,
+      "learning_rate": 8.401188123081642e-06,
+      "loss": 0.384,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2044,
+      "learning_rate": 8.414972699479062e-06,
+      "loss": 0.2191,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2048,
+      "learning_rate": 8.428760365968329e-06,
+      "loss": 1.2355,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2052,
+      "learning_rate": 8.442551095669627e-06,
+      "loss": 0.5,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2056,
+      "learning_rate": 8.456344861697293e-06,
+      "loss": 0.6364,
+      "step": 1028
+    },
+    {
+      "epoch": 0.206,
+      "learning_rate": 8.470141637159605e-06,
+      "loss": 0.6276,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2064,
+      "learning_rate": 8.483941395159114e-06,
+      "loss": 0.2992,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2068,
+      "learning_rate": 8.497744108792431e-06,
+      "loss": 0.7278,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2072,
+      "learning_rate": 8.511549751150478e-06,
+      "loss": 0.2421,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2076,
+      "learning_rate": 8.52535829531845e-06,
+      "loss": 0.4927,
+      "step": 1038
+    },
+    {
+      "epoch": 0.208,
+      "learning_rate": 8.539169714375883e-06,
+      "loss": 0.2617,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2084,
+      "learning_rate": 8.552983981396707e-06,
+      "loss": 0.335,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2088,
+      "learning_rate": 8.566801069449304e-06,
+      "loss": 0.2865,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2092,
+      "learning_rate": 8.580620951596553e-06,
+      "loss": 0.4233,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2096,
+      "learning_rate": 8.594443600895886e-06,
+      "loss": 0.2044,
+      "step": 1048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.60826899039934e-06,
+      "loss": 0.3534,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2104,
+      "learning_rate": 8.622097093153612e-06,
+      "loss": 0.3784,
+      "step": 1052
+    },
+    {
+      "epoch": 0.2108,
+      "learning_rate": 8.635927882200128e-06,
+      "loss": 0.3752,
+      "step": 1054
+    },
+    {
+      "epoch": 0.2112,
+      "learning_rate": 8.649761330575e-06,
+      "loss": 0.4934,
+      "step": 1056
+    },
+    {
+      "epoch": 0.2116,
+      "learning_rate": 8.663597411309268e-06,
+      "loss": 0.4367,
+      "step": 1058
+    },
+    {
+      "epoch": 0.212,
+      "learning_rate": 8.677436097428766e-06,
+      "loss": 0.2832,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2124,
+      "learning_rate": 8.691277361954266e-06,
+      "loss": 0.2921,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2128,
+      "learning_rate": 8.705121177901537e-06,
+      "loss": 0.4773,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2132,
+      "learning_rate": 8.718967518281292e-06,
+      "loss": 0.2715,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2136,
+      "learning_rate": 8.732816356099459e-06,
+      "loss": 0.5513,
+      "step": 1068
+    },
+    {
+      "epoch": 0.214,
+      "learning_rate": 8.746667664356962e-06,
+      "loss": 0.3792,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2144,
+      "learning_rate": 8.760521416049986e-06,
+      "loss": 0.2928,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2148,
+      "learning_rate": 8.774377584169934e-06,
+      "loss": 0.3655,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2152,
+      "learning_rate": 8.788236141703477e-06,
+      "loss": 0.4638,
+      "step": 1076
+    },
+    {
+      "epoch": 0.2156,
+      "learning_rate": 8.802097061632706e-06,
+      "loss": 0.4798,
+      "step": 1078
+    },
+    {
+      "epoch": 0.216,
+      "learning_rate": 8.81596031693499e-06,
+      "loss": 0.3537,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2164,
+      "learning_rate": 8.829825880583224e-06,
+      "loss": 0.2643,
+      "step": 1082
+    },
+    {
+      "epoch": 0.2168,
+      "learning_rate": 8.84369372554578e-06,
+      "loss": 0.527,
+      "step": 1084
+    },
+    {
+      "epoch": 0.2172,
+      "learning_rate": 8.85756382478659e-06,
+      "loss": 0.5204,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2176,
+      "learning_rate": 8.87143615126518e-06,
+      "loss": 0.3133,
+      "step": 1088
+    },
+    {
+      "epoch": 0.218,
+      "learning_rate": 8.88531067793674e-06,
+      "loss": 0.4401,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2184,
+      "learning_rate": 8.899187377752173e-06,
+      "loss": 0.6687,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2188,
+      "learning_rate": 8.913066223658141e-06,
+      "loss": 0.5779,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2192,
+      "learning_rate": 8.926947188597127e-06,
+      "loss": 0.494,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2196,
+      "learning_rate": 8.940830245507473e-06,
+      "loss": 0.4146,
+      "step": 1098
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.954715367323473e-06,
+      "loss": 0.3231,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2204,
+      "learning_rate": 8.968602526975317e-06,
+      "loss": 0.2961,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2208,
+      "learning_rate": 8.982491697389344e-06,
+      "loss": 0.9067,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2212,
+      "learning_rate": 8.996382851487839e-06,
+      "loss": 0.2304,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2216,
+      "learning_rate": 9.010275962189356e-06,
+      "loss": 0.4528,
+      "step": 1108
+    },
+    {
+      "epoch": 0.222,
+      "learning_rate": 9.024171002408509e-06,
+      "loss": 0.5356,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2224,
+      "learning_rate": 9.03806794505621e-06,
+      "loss": 0.28,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2228,
+      "learning_rate": 9.051966763039708e-06,
+      "loss": 0.3268,
+      "step": 1114
+    },
+    {
+      "epoch": 0.2232,
+      "learning_rate": 9.065867429262497e-06,
+      "loss": 0.3966,
+      "step": 1116
+    },
+    {
+      "epoch": 0.2236,
+      "learning_rate": 9.07976991662453e-06,
+      "loss": 0.8257,
+      "step": 1118
+    },
+    {
+      "epoch": 0.224,
+      "learning_rate": 9.093674198022198e-06,
+      "loss": 0.3123,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2244,
+      "learning_rate": 9.107580246348395e-06,
+      "loss": 0.4364,
+      "step": 1122
+    },
+    {
+      "epoch": 0.2248,
+      "learning_rate": 9.121488034492567e-06,
+      "loss": 0.7562,
+      "step": 1124
+    },
+    {
+      "epoch": 0.2252,
+      "learning_rate": 9.135397535340768e-06,
+      "loss": 0.3469,
+      "step": 1126
+    },
+    {
+      "epoch": 0.2256,
+      "learning_rate": 9.149308721775717e-06,
+      "loss": 1.2449,
+      "step": 1128
+    },
+    {
+      "epoch": 0.226,
+      "learning_rate": 9.16322156667684e-06,
+      "loss": 0.4778,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2264,
+      "learning_rate": 9.177136042920338e-06,
+      "loss": 0.1348,
+      "step": 1132
+    },
+    {
+      "epoch": 0.2268,
+      "learning_rate": 9.191052123379227e-06,
+      "loss": 0.2866,
+      "step": 1134
+    },
+    {
+      "epoch": 0.2272,
+      "learning_rate": 9.204969780923396e-06,
+      "loss": 0.3545,
+      "step": 1136
+    },
+    {
+      "epoch": 0.2276,
+      "learning_rate": 9.218888988419656e-06,
+      "loss": 0.3384,
+      "step": 1138
+    },
+    {
+      "epoch": 0.228,
+      "learning_rate": 9.232809718731822e-06,
+      "loss": 0.5363,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2284,
+      "learning_rate": 9.246731944720663e-06,
+      "loss": 0.5801,
+      "step": 1142
+    },
+    {
+      "epoch": 0.2288,
+      "learning_rate": 9.26065563924414e-06,
+      "loss": 0.3464,
+      "step": 1144
+    },
+    {
+      "epoch": 0.2292,
+      "learning_rate": 9.274580775157299e-06,
+      "loss": 0.3997,
+      "step": 1146
+    },
+    {
+      "epoch": 0.2296,
+      "learning_rate": 9.288507325312319e-06,
+      "loss": 0.7596,
+      "step": 1148
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.302435262558752e-06,
+      "loss": 0.452,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2304,
+      "learning_rate": 9.316364559743298e-06,
+      "loss": 0.2643,
+      "step": 1152
+    },
+    {
+      "epoch": 0.2308,
+      "learning_rate": 9.330295189710153e-06,
+      "loss": 0.4243,
+      "step": 1154
+    },
+    {
+      "epoch": 0.2312,
+      "learning_rate": 9.344227125300788e-06,
+      "loss": 0.2009,
+      "step": 1156
+    },
+    {
+      "epoch": 0.2316,
+      "learning_rate": 9.358160339354196e-06,
+      "loss": 0.3203,
+      "step": 1158
+    },
+    {
+      "epoch": 0.232,
+      "learning_rate": 9.372094804706867e-06,
+      "loss": 0.9398,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2324,
+      "learning_rate": 9.386030494192826e-06,
+      "loss": 0.2386,
+      "step": 1162
+    },
+    {
+      "epoch": 0.2328,
+      "learning_rate": 9.39996738064379e-06,
+      "loss": 0.3563,
+      "step": 1164
+    },
+    {
+      "epoch": 0.2332,
+      "learning_rate": 9.413905436889032e-06,
+      "loss": 0.5222,
+      "step": 1166
+    },
+    {
+      "epoch": 0.2336,
+      "learning_rate": 9.427844635755615e-06,
+      "loss": 0.7201,
+      "step": 1168
+    },
+    {
+      "epoch": 0.234,
+      "learning_rate": 9.441784950068357e-06,
+      "loss": 0.3855,
+      "step": 1170
+    },
+    {
+      "epoch": 0.2344,
+      "learning_rate": 9.455726352649904e-06,
+      "loss": 0.4182,
+      "step": 1172
+    },
+    {
+      "epoch": 0.2348,
+      "learning_rate": 9.469668816320777e-06,
+      "loss": 0.5191,
+      "step": 1174
+    },
+    {
+      "epoch": 0.2352,
+      "learning_rate": 9.483612313899446e-06,
+      "loss": 0.3424,
+      "step": 1176
+    },
+    {
+      "epoch": 0.2356,
+      "learning_rate": 9.497556818202297e-06,
+      "loss": 0.2497,
+      "step": 1178
+    },
+    {
+      "epoch": 0.236,
+      "learning_rate": 9.511502302043859e-06,
+      "loss": 0.4109,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2364,
+      "learning_rate": 9.52544873823668e-06,
+      "loss": 0.5524,
+      "step": 1182
+    },
+    {
+      "epoch": 0.2368,
+      "learning_rate": 9.539396099591469e-06,
+      "loss": 0.0892,
+      "step": 1184
+    },
+    {
+      "epoch": 0.2372,
+      "learning_rate": 9.553344358917146e-06,
+      "loss": 0.654,
+      "step": 1186
+    },
+    {
+      "epoch": 0.2376,
+      "learning_rate": 9.567293489020816e-06,
+      "loss": 0.2993,
+      "step": 1188
+    },
+    {
+      "epoch": 0.238,
+      "learning_rate": 9.581243462708009e-06,
+      "loss": 0.3409,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2384,
+      "learning_rate": 9.595194252782461e-06,
+      "loss": 0.6568,
+      "step": 1192
+    },
+    {
+      "epoch": 0.2388,
+      "learning_rate": 9.609145832046469e-06,
+      "loss": 0.1691,
+      "step": 1194
+    },
+    {
+      "epoch": 0.2392,
+      "learning_rate": 9.623098173300656e-06,
+      "loss": 0.6078,
+      "step": 1196
+    },
+    {
+      "epoch": 0.2396,
+      "learning_rate": 9.637051249344225e-06,
+      "loss": 0.3729,
+      "step": 1198
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.651005032974991e-06,
+      "loss": 0.33,
+      "step": 1200
+    },
+    {
+      "epoch": 0.2404,
+      "learning_rate": 9.664959496989285e-06,
+      "loss": 0.5501,
+      "step": 1202
+    },
+    {
+      "epoch": 0.2408,
+      "learning_rate": 9.678914614182184e-06,
+      "loss": 0.341,
+      "step": 1204
+    },
+    {
+      "epoch": 0.2412,
+      "learning_rate": 9.69287035734747e-06,
+      "loss": 0.6727,
+      "step": 1206
+    },
+    {
+      "epoch": 0.2416,
+      "learning_rate": 9.706826699277714e-06,
+      "loss": 1.3072,
+      "step": 1208
+    },
+    {
+      "epoch": 0.242,
+      "learning_rate": 9.720783612764307e-06,
+      "loss": 0.6377,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2424,
+      "learning_rate": 9.734741070597535e-06,
+      "loss": 0.8198,
+      "step": 1212
+    },
+    {
+      "epoch": 0.2428,
+      "learning_rate": 9.74869904556662e-06,
+      "loss": 0.316,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2432,
+      "learning_rate": 9.762657510459774e-06,
+      "loss": 0.3971,
+      "step": 1216
+    },
+    {
+      "epoch": 0.2436,
+      "learning_rate": 9.776616438064255e-06,
+      "loss": 0.24,
+      "step": 1218
+    },
+    {
+      "epoch": 0.244,
+      "learning_rate": 9.790575801166422e-06,
+      "loss": 0.575,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2444,
+      "learning_rate": 9.804535572551782e-06,
+      "loss": 0.2665,
+      "step": 1222
+    },
+    {
+      "epoch": 0.2448,
+      "learning_rate": 9.818495725005043e-06,
+      "loss": 0.4151,
+      "step": 1224
+    },
+    {
+      "epoch": 0.2452,
+      "learning_rate": 9.832456231310194e-06,
+      "loss": 0.1338,
+      "step": 1226
+    },
+    {
+      "epoch": 0.2456,
+      "learning_rate": 9.846417064250459e-06,
+      "loss": 0.2552,
+      "step": 1228
+    },
+    {
+      "epoch": 0.246,
+      "learning_rate": 9.860378196608552e-06,
+      "loss": 0.733,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2464,
+      "learning_rate": 9.874339601166479e-06,
+      "loss": 0.7393,
+      "step": 1232
+    },
+    {
+      "epoch": 0.2468,
+      "learning_rate": 9.888301250705765e-06,
+      "loss": 0.5403,
+      "step": 1234
+    },
+    {
+      "epoch": 0.2472,
+      "learning_rate": 9.902263118007513e-06,
+      "loss": 0.1973,
+      "step": 1236
+    },
+    {
+      "epoch": 0.2476,
+      "learning_rate": 9.916225175852278e-06,
+      "loss": 0.2178,
+      "step": 1238
+    },
+    {
+      "epoch": 0.248,
+      "learning_rate": 9.930187397020385e-06,
+      "loss": 0.3829,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2484,
+      "learning_rate": 9.944149754291716e-06,
+      "loss": 0.3331,
+      "step": 1242
+    },
+    {
+      "epoch": 0.2488,
+      "learning_rate": 9.95811222044596e-06,
+      "loss": 0.2738,
+      "step": 1244
+    },
+    {
+      "epoch": 0.2492,
+      "learning_rate": 9.972074768262572e-06,
+      "loss": 0.4324,
+      "step": 1246
+    },
+    {
+      "epoch": 0.2496,
+      "learning_rate": 9.986037370520855e-06,
+      "loss": 0.4488,
+      "step": 1248
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.999999999999996e-06,
+      "loss": 0.4487,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2504,
+      "learning_rate": 1.0013962629479139e-05,
+      "loss": 0.3213,
+      "step": 1252
+    },
+    {
+      "epoch": 0.2508,
+      "learning_rate": 1.0027925231737419e-05,
+      "loss": 0.2961,
+      "step": 1254
+    },
+    {
+      "epoch": 0.2512,
+      "learning_rate": 1.0041887779554034e-05,
+      "loss": 0.2068,
+      "step": 1256
+    },
+    {
+      "epoch": 0.2516,
+      "learning_rate": 1.0055850245708276e-05,
+      "loss": 0.3443,
+      "step": 1258
+    },
+    {
+      "epoch": 0.252,
+      "learning_rate": 1.0069812602979607e-05,
+      "loss": 0.5035,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2524,
+      "learning_rate": 1.0083774824147717e-05,
+      "loss": 0.1745,
+      "step": 1262
+    },
+    {
+      "epoch": 0.2528,
+      "learning_rate": 1.0097736881992482e-05,
+      "loss": 0.596,
+      "step": 1264
+    },
+    {
+      "epoch": 0.2532,
+      "learning_rate": 1.011169874929423e-05,
+      "loss": 0.2749,
+      "step": 1266
+    },
+    {
+      "epoch": 0.2536,
+      "learning_rate": 1.0125660398833514e-05,
+      "loss": 0.5535,
+      "step": 1268
+    },
+    {
+      "epoch": 0.254,
+      "learning_rate": 1.013962180339144e-05,
+      "loss": 0.3698,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2544,
+      "learning_rate": 1.0153582935749533e-05,
+      "loss": 0.47,
+      "step": 1272
+    },
+    {
+      "epoch": 0.2548,
+      "learning_rate": 1.01675437686898e-05,
+      "loss": 0.4238,
+      "step": 1274
+    },
+    {
+      "epoch": 0.2552,
+      "learning_rate": 1.0181504274994952e-05,
+      "loss": 1.0414,
+      "step": 1276
+    },
+    {
+      "epoch": 0.2556,
+      "learning_rate": 1.0195464427448212e-05,
+      "loss": 0.2627,
+      "step": 1278
+    },
+    {
+      "epoch": 0.256,
+      "learning_rate": 1.0209424198833571e-05,
+      "loss": 0.5201,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2564,
+      "learning_rate": 1.0223383561935738e-05,
+      "loss": 0.2721,
+      "step": 1282
+    },
+    {
+      "epoch": 0.2568,
+      "learning_rate": 1.0237342489540218e-05,
+      "loss": 0.1794,
+      "step": 1284
+    },
+    {
+      "epoch": 0.2572,
+      "learning_rate": 1.0251300954433374e-05,
+      "loss": 0.4827,
+      "step": 1286
+    },
+    {
+      "epoch": 0.2576,
+      "learning_rate": 1.0265258929402458e-05,
+      "loss": 0.2802,
+      "step": 1288
+    },
+    {
+      "epoch": 0.258,
+      "learning_rate": 1.0279216387235686e-05,
+      "loss": 0.169,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2584,
+      "learning_rate": 1.029317330072228e-05,
+      "loss": 0.3722,
+      "step": 1292
+    },
+    {
+      "epoch": 0.2588,
+      "learning_rate": 1.0307129642652523e-05,
+      "loss": 0.4635,
+      "step": 1294
+    },
+    {
+      "epoch": 0.2592,
+      "learning_rate": 1.0321085385817811e-05,
+      "loss": 0.6744,
+      "step": 1296
+    },
+    {
+      "epoch": 0.2596,
+      "learning_rate": 1.033504050301071e-05,
+      "loss": 0.3408,
+      "step": 1298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.0348994967025004e-05,
+      "loss": 0.6595,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2604,
+      "learning_rate": 1.0362948750655768e-05,
+      "loss": 0.4341,
+      "step": 1302
+    },
+    {
+      "epoch": 0.2608,
+      "learning_rate": 1.0376901826699337e-05,
+      "loss": 0.4674,
+      "step": 1304
+    },
+    {
+      "epoch": 0.2612,
+      "learning_rate": 1.0390854167953526e-05,
+      "loss": 0.7926,
+      "step": 1306
+    },
+    {
+      "epoch": 0.2616,
+      "learning_rate": 1.0404805747217532e-05,
+      "loss": 0.8566,
+      "step": 1308
+    },
+    {
+      "epoch": 0.262,
+      "learning_rate": 1.0418756537291984e-05,
+      "loss": 0.6801,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2624,
+      "learning_rate": 1.0432706510979175e-05,
+      "loss": 0.2952,
+      "step": 1312
+    },
+    {
+      "epoch": 0.2628,
+      "learning_rate": 1.0446655641082846e-05,
+      "loss": 0.4488,
+      "step": 1314
+    },
+    {
+      "epoch": 0.2632,
+      "learning_rate": 1.0460603900408526e-05,
+      "loss": 0.2719,
+      "step": 1316
+    },
+    {
+      "epoch": 0.2636,
+      "learning_rate": 1.0474551261763312e-05,
+      "loss": 0.186,
+      "step": 1318
+    },
+    {
+      "epoch": 0.264,
+      "learning_rate": 1.0488497697956134e-05,
+      "loss": 0.4073,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2644,
+      "learning_rate": 1.0502443181797696e-05,
+      "loss": 0.3498,
+      "step": 1322
+    },
+    {
+      "epoch": 0.2648,
+      "learning_rate": 1.0516387686100549e-05,
+      "loss": 0.7824,
+      "step": 1324
+    },
+    {
+      "epoch": 0.2652,
+      "learning_rate": 1.0530331183679216e-05,
+      "loss": 0.3453,
+      "step": 1326
+    },
+    {
+      "epoch": 0.2656,
+      "learning_rate": 1.054427364735009e-05,
+      "loss": 0.5599,
+      "step": 1328
+    },
+    {
+      "epoch": 0.266,
+      "learning_rate": 1.0558215049931634e-05,
+      "loss": 0.3058,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2664,
+      "learning_rate": 1.0572155364244378e-05,
+      "loss": 1.0564,
+      "step": 1332
+    },
+    {
+      "epoch": 0.2668,
+      "learning_rate": 1.058609456311096e-05,
+      "loss": 0.2901,
+      "step": 1334
+    },
+    {
+      "epoch": 0.2672,
+      "learning_rate": 1.0600032619356203e-05,
+      "loss": 2.8422,
+      "step": 1336
+    },
+    {
+      "epoch": 0.2676,
+      "learning_rate": 1.0613969505807167e-05,
+      "loss": 0.272,
+      "step": 1338
+    },
+    {
+      "epoch": 0.268,
+      "learning_rate": 1.0627905195293127e-05,
+      "loss": 0.192,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2684,
+      "learning_rate": 1.0641839660645795e-05,
+      "loss": 0.4333,
+      "step": 1342
+    },
+    {
+      "epoch": 0.2688,
+      "learning_rate": 1.0655772874699206e-05,
+      "loss": 0.2916,
+      "step": 1344
+    },
+    {
+      "epoch": 0.2692,
+      "learning_rate": 1.066970481028984e-05,
+      "loss": 0.3684,
+      "step": 1346
+    },
+    {
+      "epoch": 0.2696,
+      "learning_rate": 1.0683635440256694e-05,
+      "loss": 0.2411,
+      "step": 1348
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.0697564737441242e-05,
+      "loss": 0.255,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2704,
+      "learning_rate": 1.0711492674687674e-05,
+      "loss": 2.549,
+      "step": 1352
+    },
+    {
+      "epoch": 0.2708,
+      "learning_rate": 1.0725419224842695e-05,
+      "loss": 0.2908,
+      "step": 1354
+    },
+    {
+      "epoch": 0.2712,
+      "learning_rate": 1.0739344360755855e-05,
+      "loss": 0.5812,
+      "step": 1356
+    },
+    {
+      "epoch": 0.2716,
+      "learning_rate": 1.0753268055279332e-05,
+      "loss": 0.1264,
+      "step": 1358
+    },
+    {
+      "epoch": 0.272,
+      "learning_rate": 1.0767190281268171e-05,
+      "loss": 0.4746,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2724,
+      "learning_rate": 1.0781111011580336e-05,
+      "loss": 0.2499,
+      "step": 1362
+    },
+    {
+      "epoch": 0.2728,
+      "learning_rate": 1.07950302190766e-05,
+      "loss": 0.5753,
+      "step": 1364
+    },
+    {
+      "epoch": 0.2732,
+      "learning_rate": 1.0808947876620766e-05,
+      "loss": 0.2242,
+      "step": 1366
+    },
+    {
+      "epoch": 0.2736,
+      "learning_rate": 1.0822863957079654e-05,
+      "loss": 0.5205,
+      "step": 1368
+    },
+    {
+      "epoch": 0.274,
+      "learning_rate": 1.0836778433323153e-05,
+      "loss": 0.5005,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2744,
+      "learning_rate": 1.0850691278224277e-05,
+      "loss": 0.2949,
+      "step": 1372
+    },
+    {
+      "epoch": 0.2748,
+      "learning_rate": 1.0864602464659227e-05,
+      "loss": 0.1662,
+      "step": 1374
+    },
+    {
+      "epoch": 0.2752,
+      "learning_rate": 1.0878511965507428e-05,
+      "loss": 0.3944,
+      "step": 1376
+    },
+    {
+      "epoch": 0.2756,
+      "learning_rate": 1.0892419753651598e-05,
+      "loss": 0.4994,
+      "step": 1378
+    },
+    {
+      "epoch": 0.276,
+      "learning_rate": 1.0906325801977795e-05,
+      "loss": 0.6333,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2764,
+      "learning_rate": 1.0920230083375465e-05,
+      "loss": 0.5979,
+      "step": 1382
+    },
+    {
+      "epoch": 0.2768,
+      "learning_rate": 1.0934132570737497e-05,
+      "loss": 0.6516,
+      "step": 1384
+    },
+    {
+      "epoch": 0.2772,
+      "learning_rate": 1.0948033236960285e-05,
+      "loss": 0.5031,
+      "step": 1386
+    },
+    {
+      "epoch": 0.2776,
+      "learning_rate": 1.0961932054943785e-05,
+      "loss": 0.5321,
+      "step": 1388
+    },
+    {
+      "epoch": 0.278,
+      "learning_rate": 1.0975828997591484e-05,
+      "loss": 0.5573,
+      "step": 1390
+    },
+    {
+      "epoch": 0.2784,
+      "learning_rate": 1.098972403781064e-05,
+      "loss": 0.5062,
+      "step": 1392
+    },
+    {
+      "epoch": 0.2788,
+      "learning_rate": 1.1003617148512154e-05,
+      "loss": 0.1223,
+      "step": 1394
+    },
+    {
+      "epoch": 0.2792,
+      "learning_rate": 1.101750830261065e-05,
+      "loss": 0.158,
+      "step": 1396
+    },
+    {
+      "epoch": 0.2796,
+      "learning_rate": 1.1031397473024676e-05,
+      "loss": 0.2081,
+      "step": 1398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.104528463267652e-05,
+      "loss": 0.3427,
+      "step": 1400
+    },
+    {
+      "epoch": 0.2804,
+      "learning_rate": 1.1059169754492518e-05,
+      "loss": 0.3411,
+      "step": 1402
+    },
+    {
+      "epoch": 0.2808,
+      "learning_rate": 1.1073052811402867e-05,
+      "loss": 0.787,
+      "step": 1404
+    },
+    {
+      "epoch": 0.2812,
+      "learning_rate": 1.108693377634185e-05,
+      "loss": 0.8588,
+      "step": 1406
+    },
+    {
+      "epoch": 0.2816,
+      "learning_rate": 1.1100812622247821e-05,
+      "loss": 0.4741,
+      "step": 1408
+    },
+    {
+      "epoch": 0.282,
+      "learning_rate": 1.1114689322063252e-05,
+      "loss": 0.3054,
+      "step": 1410
+    },
+    {
+      "epoch": 0.2824,
+      "learning_rate": 1.1128563848734815e-05,
+      "loss": 0.2285,
+      "step": 1412
+    },
+    {
+      "epoch": 0.2828,
+      "learning_rate": 1.1142436175213404e-05,
+      "loss": 0.4861,
+      "step": 1414
+    },
+    {
+      "epoch": 0.2832,
+      "learning_rate": 1.1156306274454211e-05,
+      "loss": 0.4535,
+      "step": 1416
+    },
+    {
+      "epoch": 0.2836,
+      "learning_rate": 1.117017411941677e-05,
+      "loss": 0.3787,
+      "step": 1418
+    },
+    {
+      "epoch": 0.284,
+      "learning_rate": 1.1184039683065002e-05,
+      "loss": 0.3924,
+      "step": 1420
+    },
+    {
+      "epoch": 0.2844,
+      "learning_rate": 1.1197902938367289e-05,
+      "loss": 0.4588,
+      "step": 1422
+    },
+    {
+      "epoch": 0.2848,
+      "learning_rate": 1.1211763858296516e-05,
+      "loss": 0.4571,
+      "step": 1424
+    },
+    {
+      "epoch": 0.2852,
+      "learning_rate": 1.122562241583006e-05,
+      "loss": 0.4542,
+      "step": 1426
+    },
+    {
+      "epoch": 0.2856,
+      "learning_rate": 1.1239478583950007e-05,
+      "loss": 0.1352,
+      "step": 1428
+    },
+    {
+      "epoch": 0.286,
+      "learning_rate": 1.1253332335643033e-05,
+      "loss": 1.1907,
+      "step": 1430
+    },
+    {
+      "epoch": 0.2864,
+      "learning_rate": 1.1267183643900534e-05,
+      "loss": 0.617,
+      "step": 1432
+    },
+    {
+      "epoch": 0.2868,
+      "learning_rate": 1.1281032481718701e-05,
+      "loss": 0.7864,
+      "step": 1434
+    },
+    {
+      "epoch": 0.2872,
+      "learning_rate": 1.1294878822098456e-05,
+      "loss": 0.5383,
+      "step": 1436
+    },
+    {
+      "epoch": 0.2876,
+      "learning_rate": 1.1308722638045725e-05,
+      "loss": 0.469,
+      "step": 1438
+    },
+    {
+      "epoch": 0.288,
+      "learning_rate": 1.1322563902571227e-05,
+      "loss": 0.8571,
+      "step": 1440
+    },
+    {
+      "epoch": 0.2884,
+      "learning_rate": 1.1336402588690725e-05,
+      "loss": 0.1857,
+      "step": 1442
+    },
+    {
+      "epoch": 0.2888,
+      "learning_rate": 1.1350238669424993e-05,
+      "loss": 0.5524,
+      "step": 1444
+    },
+    {
+      "epoch": 0.2892,
+      "learning_rate": 1.1364072117799864e-05,
+      "loss": 0.5513,
+      "step": 1446
+    },
+    {
+      "epoch": 0.2896,
+      "learning_rate": 1.137790290684638e-05,
+      "loss": 0.2028,
+      "step": 1448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.1391731009600652e-05,
+      "loss": 0.6303,
+      "step": 1450
+    },
+    {
+      "epoch": 0.2904,
+      "learning_rate": 1.1405556399104108e-05,
+      "loss": 0.6124,
+      "step": 1452
+    },
+    {
+      "epoch": 0.2908,
+      "learning_rate": 1.141937904840344e-05,
+      "loss": 0.6582,
+      "step": 1454
+    },
+    {
+      "epoch": 0.2912,
+      "learning_rate": 1.143319893055069e-05,
+      "loss": 0.4317,
+      "step": 1456
+    },
+    {
+      "epoch": 0.2916,
+      "learning_rate": 1.1447016018603286e-05,
+      "loss": 0.3871,
+      "step": 1458
+    },
+    {
+      "epoch": 0.292,
+      "learning_rate": 1.1460830285624112e-05,
+      "loss": 0.5083,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2924,
+      "learning_rate": 1.1474641704681541e-05,
+      "loss": 0.1262,
+      "step": 1462
+    },
+    {
+      "epoch": 0.2928,
+      "learning_rate": 1.1488450248849515e-05,
+      "loss": 0.1934,
+      "step": 1464
+    },
+    {
+      "epoch": 0.2932,
+      "learning_rate": 1.150225589120756e-05,
+      "loss": 0.7433,
+      "step": 1466
+    },
+    {
+      "epoch": 0.2936,
+      "learning_rate": 1.1516058604840881e-05,
+      "loss": 0.5863,
+      "step": 1468
+    },
+    {
+      "epoch": 0.294,
+      "learning_rate": 1.1529858362840388e-05,
+      "loss": 0.143,
+      "step": 1470
+    },
+    {
+      "epoch": 0.2944,
+      "learning_rate": 1.15436551383027e-05,
+      "loss": 0.2944,
+      "step": 1472
+    },
+    {
+      "epoch": 0.2948,
+      "learning_rate": 1.1557448904330366e-05,
+      "loss": 0.2327,
+      "step": 1474
+    },
+    {
+      "epoch": 0.2952,
+      "learning_rate": 1.1571239634031666e-05,
+      "loss": 0.3026,
+      "step": 1476
+    },
+    {
+      "epoch": 0.2956,
+      "learning_rate": 1.158502730052093e-05,
+      "loss": 0.1145,
+      "step": 1478
+    },
+    {
+      "epoch": 0.296,
+      "learning_rate": 1.1598811876918352e-05,
+      "loss": 0.4268,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2964,
+      "learning_rate": 1.161259333635019e-05,
+      "loss": 0.2187,
+      "step": 1482
+    },
+    {
+      "epoch": 0.2968,
+      "learning_rate": 1.1626371651948839e-05,
+      "loss": 0.2586,
+      "step": 1484
+    },
+    {
+      "epoch": 0.2972,
+      "learning_rate": 1.1640146796852711e-05,
+      "loss": 0.659,
+      "step": 1486
+    },
+    {
+      "epoch": 0.2976,
+      "learning_rate": 1.1653918744206478e-05,
+      "loss": 0.2556,
+      "step": 1488
+    },
+    {
+      "epoch": 0.298,
+      "learning_rate": 1.1667687467161021e-05,
+      "loss": 1.9373,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2984,
+      "learning_rate": 1.1681452938873515e-05,
+      "loss": 0.7823,
+      "step": 1492
+    },
+    {
+      "epoch": 0.2988,
+      "learning_rate": 1.169521513250746e-05,
+      "loss": 0.3833,
+      "step": 1494
+    },
+    {
+      "epoch": 0.2992,
+      "learning_rate": 1.1708974021232763e-05,
+      "loss": 0.7123,
+      "step": 1496
+    },
+    {
+      "epoch": 0.2996,
+      "learning_rate": 1.1722729578225762e-05,
+      "loss": 0.1582,
+      "step": 1498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.1736481776669297e-05,
+      "loss": 0.2081,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3004,
+      "learning_rate": 1.1750230589752753e-05,
+      "loss": 0.2514,
+      "step": 1502
+    },
+    {
+      "epoch": 0.3008,
+      "learning_rate": 1.1763975990672116e-05,
+      "loss": 0.459,
+      "step": 1504
+    },
+    {
+      "epoch": 0.3012,
+      "learning_rate": 1.1777717952630023e-05,
+      "loss": 0.3441,
+      "step": 1506
+    },
+    {
+      "epoch": 0.3016,
+      "learning_rate": 1.1791456448835815e-05,
+      "loss": 0.1616,
+      "step": 1508
+    },
+    {
+      "epoch": 0.302,
+      "learning_rate": 1.180519145250561e-05,
+      "loss": 0.2462,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3024,
+      "learning_rate": 1.1818922936862258e-05,
+      "loss": 0.4865,
+      "step": 1512
+    },
+    {
+      "epoch": 0.3028,
+      "learning_rate": 1.1832650875135606e-05,
+      "loss": 0.3133,
+      "step": 1514
+    },
+    {
+      "epoch": 0.3032,
+      "learning_rate": 1.1846375240562274e-05,
+      "loss": 0.2671,
+      "step": 1516
+    },
+    {
+      "epoch": 0.3036,
+      "learning_rate": 1.1860096006385918e-05,
+      "loss": 0.3608,
+      "step": 1518
+    },
+    {
+      "epoch": 0.304,
+      "learning_rate": 1.187381314585725e-05,
+      "loss": 0.3268,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3044,
+      "learning_rate": 1.1887526632233937e-05,
+      "loss": 0.2715,
+      "step": 1522
+    },
+    {
+      "epoch": 0.3048,
+      "learning_rate": 1.1901236438780906e-05,
+      "loss": 0.2828,
+      "step": 1524
+    },
+    {
+      "epoch": 0.3052,
+      "learning_rate": 1.191494253877013e-05,
+      "loss": 0.2979,
+      "step": 1526
+    },
+    {
+      "epoch": 0.3056,
+      "learning_rate": 1.192864490548089e-05,
+      "loss": 0.8562,
+      "step": 1528
+    },
+    {
+      "epoch": 0.306,
+      "learning_rate": 1.1942343512199719e-05,
+      "loss": 0.3044,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3064,
+      "learning_rate": 1.195603833222048e-05,
+      "loss": 0.9839,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3068,
+      "learning_rate": 1.1969729338844422e-05,
+      "loss": 0.4034,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3072,
+      "learning_rate": 1.198341650538023e-05,
+      "loss": 0.8108,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3076,
+      "learning_rate": 1.1997099805144066e-05,
+      "loss": 0.1834,
+      "step": 1538
+    },
+    {
+      "epoch": 0.308,
+      "learning_rate": 1.2010779211459642e-05,
+      "loss": 0.3162,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3084,
+      "learning_rate": 1.2024454697658254e-05,
+      "loss": 0.4109,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3088,
+      "learning_rate": 1.203812623707884e-05,
+      "loss": 0.4498,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3092,
+      "learning_rate": 1.2051793803068054e-05,
+      "loss": 0.1773,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3096,
+      "learning_rate": 1.2065457368980227e-05,
+      "loss": 0.3029,
+      "step": 1548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.20791169081776e-05,
+      "loss": 0.5723,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3104,
+      "learning_rate": 1.2092772394030141e-05,
+      "loss": 0.3939,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3108,
+      "learning_rate": 1.210642379991583e-05,
+      "loss": 0.6722,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3112,
+      "learning_rate": 1.2120071099220552e-05,
+      "loss": 0.6192,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3116,
+      "learning_rate": 1.2133714265338148e-05,
+      "loss": 0.6173,
+      "step": 1558
+    },
+    {
+      "epoch": 0.312,
+      "learning_rate": 1.2147353271670637e-05,
+      "loss": 0.4244,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3124,
+      "learning_rate": 1.2160988091628006e-05,
+      "loss": 0.2055,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3128,
+      "learning_rate": 1.217461869862855e-05,
+      "loss": 0.4073,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3132,
+      "learning_rate": 1.2188245066098647e-05,
+      "loss": 0.4476,
+      "step": 1566
+    },
+    {
+      "epoch": 0.3136,
+      "learning_rate": 1.2201867167473015e-05,
+      "loss": 0.6565,
+      "step": 1568
+    },
+    {
+      "epoch": 0.314,
+      "learning_rate": 1.2215484976194673e-05,
+      "loss": 0.628,
+      "step": 1570
+    },
+    {
+      "epoch": 0.3144,
+      "learning_rate": 1.2229098465715002e-05,
+      "loss": 0.3125,
+      "step": 1572
+    },
+    {
+      "epoch": 0.3148,
+      "learning_rate": 1.2242707609493809e-05,
+      "loss": 1.4024,
+      "step": 1574
+    },
+    {
+      "epoch": 0.3152,
+      "learning_rate": 1.2256312380999373e-05,
+      "loss": 0.6397,
+      "step": 1576
+    },
+    {
+      "epoch": 0.3156,
+      "learning_rate": 1.2269912753708496e-05,
+      "loss": 0.4433,
+      "step": 1578
+    },
+    {
+      "epoch": 0.316,
+      "learning_rate": 1.2283508701106552e-05,
+      "loss": 0.3786,
+      "step": 1580
+    },
+    {
+      "epoch": 0.3164,
+      "learning_rate": 1.229710019668755e-05,
+      "loss": 0.7145,
+      "step": 1582
+    },
+    {
+      "epoch": 0.3168,
+      "learning_rate": 1.2310687213954173e-05,
+      "loss": 0.3936,
+      "step": 1584
+    },
+    {
+      "epoch": 0.3172,
+      "learning_rate": 1.232426972641785e-05,
+      "loss": 0.324,
+      "step": 1586
+    },
+    {
+      "epoch": 0.3176,
+      "learning_rate": 1.233784770759873e-05,
+      "loss": 0.4066,
+      "step": 1588
+    },
+    {
+      "epoch": 0.318,
+      "learning_rate": 1.2351421131025891e-05,
+      "loss": 0.095,
+      "step": 1590
+    },
+    {
+      "epoch": 0.3184,
+      "learning_rate": 1.2364989970237238e-05,
+      "loss": 0.3485,
+      "step": 1592
+    },
+    {
+      "epoch": 0.3188,
+      "learning_rate": 1.237855419877962e-05,
+      "loss": 0.4817,
+      "step": 1594
+    },
+    {
+      "epoch": 0.3192,
+      "learning_rate": 1.23921137902089e-05,
+      "loss": 0.2275,
+      "step": 1596
+    },
+    {
+      "epoch": 0.3196,
+      "learning_rate": 1.2405668718089906e-05,
+      "loss": 0.2994,
+      "step": 1598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.241921895599668e-05,
+      "loss": 0.3392,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3204,
+      "learning_rate": 1.2432764477512295e-05,
+      "loss": 0.2856,
+      "step": 1602
+    },
+    {
+      "epoch": 0.3208,
+      "learning_rate": 1.2446305256229076e-05,
+      "loss": 0.1886,
+      "step": 1604
+    },
+    {
+      "epoch": 0.3212,
+      "learning_rate": 1.2459841265748582e-05,
+      "loss": 0.326,
+      "step": 1606
+    },
+    {
+      "epoch": 0.3216,
+      "learning_rate": 1.2473372479681653e-05,
+      "loss": 0.3892,
+      "step": 1608
+    },
+    {
+      "epoch": 0.322,
+      "learning_rate": 1.2486898871648547e-05,
+      "loss": 0.2951,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3224,
+      "learning_rate": 1.2500420415278822e-05,
+      "loss": 0.2612,
+      "step": 1612
+    },
+    {
+      "epoch": 0.3228,
+      "learning_rate": 1.2513937084211546e-05,
+      "loss": 0.6412,
+      "step": 1614
+    },
+    {
+      "epoch": 0.3232,
+      "learning_rate": 1.2527448852095292e-05,
+      "loss": 0.6221,
+      "step": 1616
+    },
+    {
+      "epoch": 0.3236,
+      "learning_rate": 1.2540955692588167e-05,
+      "loss": 0.4591,
+      "step": 1618
+    },
+    {
+      "epoch": 0.324,
+      "learning_rate": 1.2554457579357902e-05,
+      "loss": 0.3751,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3244,
+      "learning_rate": 1.2567954486081873e-05,
+      "loss": 0.3578,
+      "step": 1622
+    },
+    {
+      "epoch": 0.3248,
+      "learning_rate": 1.2581446386447171e-05,
+      "loss": 0.4868,
+      "step": 1624
+    },
+    {
+      "epoch": 0.3252,
+      "learning_rate": 1.2594933254150647e-05,
+      "loss": 0.1881,
+      "step": 1626
+    },
+    {
+      "epoch": 0.3256,
+      "learning_rate": 1.2608415062898963e-05,
+      "loss": 0.1632,
+      "step": 1628
+    },
+    {
+      "epoch": 0.326,
+      "learning_rate": 1.262189178640864e-05,
+      "loss": 0.1633,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3264,
+      "learning_rate": 1.2635363398406133e-05,
+      "loss": 0.3786,
+      "step": 1632
+    },
+    {
+      "epoch": 0.3268,
+      "learning_rate": 1.2648829872627797e-05,
+      "loss": 0.2799,
+      "step": 1634
+    },
+    {
+      "epoch": 0.3272,
+      "learning_rate": 1.266229118282012e-05,
+      "loss": 0.2151,
+      "step": 1636
+    },
+    {
+      "epoch": 0.3276,
+      "learning_rate": 1.2675747302739516e-05,
+      "loss": 0.2396,
+      "step": 1638
+    },
+    {
+      "epoch": 0.328,
+      "learning_rate": 1.2689198206152644e-05,
+      "loss": 0.3094,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3284,
+      "learning_rate": 1.2702643866836281e-05,
+      "loss": 0.2299,
+      "step": 1642
+    },
+    {
+      "epoch": 0.3288,
+      "learning_rate": 1.2716084258577373e-05,
+      "loss": 1.506,
+      "step": 1644
+    },
+    {
+      "epoch": 0.3292,
+      "learning_rate": 1.2729519355173254e-05,
+      "loss": 0.5759,
+      "step": 1646
+    },
+    {
+      "epoch": 0.3296,
+      "learning_rate": 1.2742949130431468e-05,
+      "loss": 0.3715,
+      "step": 1648
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.2756373558169992e-05,
+      "loss": 0.2455,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3304,
+      "learning_rate": 1.2769792612217224e-05,
+      "loss": 0.3149,
+      "step": 1652
+    },
+    {
+      "epoch": 0.3308,
+      "learning_rate": 1.2783206266412028e-05,
+      "loss": 0.7631,
+      "step": 1654
+    },
+    {
+      "epoch": 0.3312,
+      "learning_rate": 1.2796614494603795e-05,
+      "loss": 0.4574,
+      "step": 1656
+    },
+    {
+      "epoch": 0.3316,
+      "learning_rate": 1.2810017270652508e-05,
+      "loss": 0.0546,
+      "step": 1658
+    },
+    {
+      "epoch": 0.332,
+      "learning_rate": 1.282341456842876e-05,
+      "loss": 0.8595,
+      "step": 1660
+    },
+    {
+      "epoch": 0.3324,
+      "learning_rate": 1.283680636181384e-05,
+      "loss": 0.3372,
+      "step": 1662
+    },
+    {
+      "epoch": 0.3328,
+      "learning_rate": 1.2850192624699756e-05,
+      "loss": 0.3071,
+      "step": 1664
+    },
+    {
+      "epoch": 0.3332,
+      "learning_rate": 1.2863573330989308e-05,
+      "loss": 0.3267,
+      "step": 1666
+    },
+    {
+      "epoch": 0.3336,
+      "learning_rate": 1.2876948454596122e-05,
+      "loss": 0.4947,
+      "step": 1668
+    },
+    {
+      "epoch": 0.334,
+      "learning_rate": 1.2890317969444708e-05,
+      "loss": 0.6155,
+      "step": 1670
+    },
+    {
+      "epoch": 0.3344,
+      "learning_rate": 1.2903681849470535e-05,
+      "loss": 0.3785,
+      "step": 1672
+    },
+    {
+      "epoch": 0.3348,
+      "learning_rate": 1.291704006861998e-05,
+      "loss": 0.646,
+      "step": 1674
+    },
+    {
+      "epoch": 0.3352,
+      "learning_rate": 1.2930392600850565e-05,
+      "loss": 0.6758,
+      "step": 1676
+    },
+    {
+      "epoch": 0.3356,
+      "learning_rate": 1.2943739420130843e-05,
+      "loss": 0.5161,
+      "step": 1678
+    },
+    {
+      "epoch": 0.336,
+      "learning_rate": 1.2957080500440455e-05,
+      "loss": 0.211,
+      "step": 1680
+    },
+    {
+      "epoch": 0.3364,
+      "learning_rate": 1.2970415815770353e-05,
+      "loss": 1.1864,
+      "step": 1682
+    },
+    {
+      "epoch": 0.3368,
+      "learning_rate": 1.2983745340122589e-05,
+      "loss": 0.6206,
+      "step": 1684
+    },
+    {
+      "epoch": 0.3372,
+      "learning_rate": 1.299706904751064e-05,
+      "loss": 0.8598,
+      "step": 1686
+    },
+    {
+      "epoch": 0.3376,
+      "learning_rate": 1.3010386911959205e-05,
+      "loss": 0.5634,
+      "step": 1688
+    },
+    {
+      "epoch": 0.338,
+      "learning_rate": 1.3023698907504447e-05,
+      "loss": 0.451,
+      "step": 1690
+    },
+    {
+      "epoch": 0.3384,
+      "learning_rate": 1.3037005008193944e-05,
+      "loss": 0.2313,
+      "step": 1692
+    },
+    {
+      "epoch": 0.3388,
+      "learning_rate": 1.3050305188086757e-05,
+      "loss": 0.6209,
+      "step": 1694
+    },
+    {
+      "epoch": 0.3392,
+      "learning_rate": 1.3063599421253556e-05,
+      "loss": 0.3399,
+      "step": 1696
+    },
+    {
+      "epoch": 0.3396,
+      "learning_rate": 1.3076887681776504e-05,
+      "loss": 0.3016,
+      "step": 1698
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.309016994374947e-05,
+      "loss": 0.2639,
+      "step": 1700
+    },
+    {
+      "epoch": 0.3404,
+      "learning_rate": 1.310344618127801e-05,
+      "loss": 1.0176,
+      "step": 1702
+    },
+    {
+      "epoch": 0.3408,
+      "learning_rate": 1.3116716368479415e-05,
+      "loss": 0.7978,
+      "step": 1704
+    },
+    {
+      "epoch": 0.3412,
+      "learning_rate": 1.3129980479482776e-05,
+      "loss": 0.374,
+      "step": 1706
+    },
+    {
+      "epoch": 0.3416,
+      "learning_rate": 1.3143238488429049e-05,
+      "loss": 0.4477,
+      "step": 1708
+    },
+    {
+      "epoch": 0.342,
+      "learning_rate": 1.3156490369471018e-05,
+      "loss": 0.1268,
+      "step": 1710
+    },
+    {
+      "epoch": 0.3424,
+      "learning_rate": 1.316973609677351e-05,
+      "loss": 0.1999,
+      "step": 1712
+    },
+    {
+      "epoch": 0.3428,
+      "learning_rate": 1.3182975644513286e-05,
+      "loss": 0.9926,
+      "step": 1714
+    },
+    {
+      "epoch": 0.3432,
+      "learning_rate": 1.319620898687917e-05,
+      "loss": 0.1662,
+      "step": 1716
+    },
+    {
+      "epoch": 0.3436,
+      "learning_rate": 1.3209436098072102e-05,
+      "loss": 0.5184,
+      "step": 1718
+    },
+    {
+      "epoch": 0.344,
+      "learning_rate": 1.32226569523051e-05,
+      "loss": 0.3454,
+      "step": 1720
+    },
+    {
+      "epoch": 0.3444,
+      "learning_rate": 1.3235871523803501e-05,
+      "loss": 0.2766,
+      "step": 1722
+    },
+    {
+      "epoch": 0.3448,
+      "learning_rate": 1.324907978680475e-05,
+      "loss": 1.7348,
+      "step": 1724
+    },
+    {
+      "epoch": 0.3452,
+      "learning_rate": 1.3262281715558738e-05,
+      "loss": 0.1671,
+      "step": 1726
+    },
+    {
+      "epoch": 0.3456,
+      "learning_rate": 1.3275477284327572e-05,
+      "loss": 0.3433,
+      "step": 1728
+    },
+    {
+      "epoch": 0.346,
+      "learning_rate": 1.3288666467385815e-05,
+      "loss": 0.3995,
+      "step": 1730
+    },
+    {
+      "epoch": 0.3464,
+      "learning_rate": 1.3301849239020537e-05,
+      "loss": 0.408,
+      "step": 1732
+    },
+    {
+      "epoch": 0.3468,
+      "learning_rate": 1.3315025573531193e-05,
+      "loss": 0.4914,
+      "step": 1734
+    },
+    {
+      "epoch": 0.3472,
+      "learning_rate": 1.3328195445229865e-05,
+      "loss": 0.4791,
+      "step": 1736
+    },
+    {
+      "epoch": 0.3476,
+      "learning_rate": 1.3341358828441214e-05,
+      "loss": 0.4036,
+      "step": 1738
+    },
+    {
+      "epoch": 0.348,
+      "learning_rate": 1.3354515697502548e-05,
+      "loss": 0.4437,
+      "step": 1740
+    },
+    {
+      "epoch": 0.3484,
+      "learning_rate": 1.3367666026763879e-05,
+      "loss": 0.1742,
+      "step": 1742
+    },
+    {
+      "epoch": 0.3488,
+      "learning_rate": 1.338080979058797e-05,
+      "loss": 0.6327,
+      "step": 1744
+    },
+    {
+      "epoch": 0.3492,
+      "learning_rate": 1.3393946963350378e-05,
+      "loss": 0.2109,
+      "step": 1746
+    },
+    {
+      "epoch": 0.3496,
+      "learning_rate": 1.340707751943951e-05,
+      "loss": 0.2445,
+      "step": 1748
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.3420201433256682e-05,
+      "loss": 0.3987,
+      "step": 1750
+    },
+    {
+      "epoch": 0.3504,
+      "learning_rate": 1.3433318679216145e-05,
+      "loss": 0.3949,
+      "step": 1752
+    },
+    {
+      "epoch": 0.3508,
+      "learning_rate": 1.3446429231745162e-05,
+      "loss": 0.2094,
+      "step": 1754
+    },
+    {
+      "epoch": 0.3512,
+      "learning_rate": 1.3459533065284039e-05,
+      "loss": 0.1445,
+      "step": 1756
+    },
+    {
+      "epoch": 0.3516,
+      "learning_rate": 1.3472630154286197e-05,
+      "loss": 0.2929,
+      "step": 1758
+    },
+    {
+      "epoch": 0.352,
+      "learning_rate": 1.348572047321814e-05,
+      "loss": 0.3592,
+      "step": 1760
+    },
+    {
+      "epoch": 0.3524,
+      "learning_rate": 1.3498803996559692e-05,
+      "loss": 0.2513,
+      "step": 1762
+    },
+    {
+      "epoch": 0.3528,
+      "learning_rate": 1.3511880698803803e-05,
+      "loss": 0.525,
+      "step": 1764
+    },
+    {
+      "epoch": 0.3532,
+      "learning_rate": 1.3524950554456773e-05,
+      "loss": 0.8528,
+      "step": 1766
+    },
+    {
+      "epoch": 0.3536,
+      "learning_rate": 1.3538013538038296e-05,
+      "loss": 1.1819,
+      "step": 1768
+    },
+    {
+      "epoch": 0.354,
+      "learning_rate": 1.3551069624081356e-05,
+      "loss": 0.3931,
+      "step": 1770
+    },
+    {
+      "epoch": 0.3544,
+      "learning_rate": 1.3564118787132507e-05,
+      "loss": 0.4501,
+      "step": 1772
+    },
+    {
+      "epoch": 0.3548,
+      "learning_rate": 1.3577161001751692e-05,
+      "loss": 0.2557,
+      "step": 1774
+    },
+    {
+      "epoch": 0.3552,
+      "learning_rate": 1.3590196242512461e-05,
+      "loss": 0.6927,
+      "step": 1776
+    },
+    {
+      "epoch": 0.3556,
+      "learning_rate": 1.3603224484001944e-05,
+      "loss": 0.7696,
+      "step": 1778
+    },
+    {
+      "epoch": 0.356,
+      "learning_rate": 1.361624570082092e-05,
+      "loss": 0.5271,
+      "step": 1780
+    },
+    {
+      "epoch": 0.3564,
+      "learning_rate": 1.362925986758386e-05,
+      "loss": 0.7454,
+      "step": 1782
+    },
+    {
+      "epoch": 0.3568,
+      "learning_rate": 1.364226695891898e-05,
+      "loss": 0.2002,
+      "step": 1784
+    },
+    {
+      "epoch": 0.3572,
+      "learning_rate": 1.3655266949468287e-05,
+      "loss": 0.3121,
+      "step": 1786
+    },
+    {
+      "epoch": 0.3576,
+      "learning_rate": 1.3668259813887637e-05,
+      "loss": 0.1468,
+      "step": 1788
+    },
+    {
+      "epoch": 0.358,
+      "learning_rate": 1.3681245526846773e-05,
+      "loss": 1.1164,
+      "step": 1790
+    },
+    {
+      "epoch": 0.3584,
+      "learning_rate": 1.3694224063029386e-05,
+      "loss": 0.4382,
+      "step": 1792
+    },
+    {
+      "epoch": 0.3588,
+      "learning_rate": 1.3707195397133176e-05,
+      "loss": 0.5762,
+      "step": 1794
+    },
+    {
+      "epoch": 0.3592,
+      "learning_rate": 1.3720159503869806e-05,
+      "loss": 0.3584,
+      "step": 1796
+    },
+    {
+      "epoch": 0.3596,
+      "learning_rate": 1.3733116357965156e-05,
+      "loss": 0.5824,
+      "step": 1798
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.374606593415911e-05,
+      "loss": 0.7021,
+      "step": 1800
+    },
+    {
+      "epoch": 0.3604,
+      "learning_rate": 1.3759008207205855e-05,
+      "loss": 2.1981,
+      "step": 1802
+    },
+    {
+      "epoch": 0.3608,
+      "learning_rate": 1.377194315187377e-05,
+      "loss": 0.3726,
+      "step": 1804
+    },
+    {
+      "epoch": 0.3612,
+      "learning_rate": 1.3784870742945468e-05,
+      "loss": 0.4626,
+      "step": 1806
+    },
+    {
+      "epoch": 0.3616,
+      "learning_rate": 1.3797790955218014e-05,
+      "loss": 0.3737,
+      "step": 1808
+    },
+    {
+      "epoch": 0.362,
+      "learning_rate": 1.3810703763502744e-05,
+      "loss": 0.314,
+      "step": 1810
+    },
+    {
+      "epoch": 0.3624,
+      "learning_rate": 1.3823609142625492e-05,
+      "loss": 0.2949,
+      "step": 1812
+    },
+    {
+      "epoch": 0.3628,
+      "learning_rate": 1.3836507067426563e-05,
+      "loss": 0.2606,
+      "step": 1814
+    },
+    {
+      "epoch": 0.3632,
+      "learning_rate": 1.3849397512760793e-05,
+      "loss": 0.2114,
+      "step": 1816
+    },
+    {
+      "epoch": 0.3636,
+      "learning_rate": 1.38622804534976e-05,
+      "loss": 0.4446,
+      "step": 1818
+    },
+    {
+      "epoch": 0.364,
+      "learning_rate": 1.3875155864521027e-05,
+      "loss": 0.2529,
+      "step": 1820
+    },
+    {
+      "epoch": 0.3644,
+      "learning_rate": 1.3888023720729806e-05,
+      "loss": 0.7536,
+      "step": 1822
+    },
+    {
+      "epoch": 0.3648,
+      "learning_rate": 1.3900883997037393e-05,
+      "loss": 0.2468,
+      "step": 1824
+    },
+    {
+      "epoch": 0.3652,
+      "learning_rate": 1.391373666837202e-05,
+      "loss": 0.3663,
+      "step": 1826
+    },
+    {
+      "epoch": 0.3656,
+      "learning_rate": 1.3926581709676746e-05,
+      "loss": 0.4399,
+      "step": 1828
+    },
+    {
+      "epoch": 0.366,
+      "learning_rate": 1.3939419095909506e-05,
+      "loss": 0.3395,
+      "step": 1830
+    },
+    {
+      "epoch": 0.3664,
+      "learning_rate": 1.3952248802043158e-05,
+      "loss": 0.482,
+      "step": 1832
+    },
+    {
+      "epoch": 0.3668,
+      "learning_rate": 1.396507080306555e-05,
+      "loss": 0.2361,
+      "step": 1834
+    },
+    {
+      "epoch": 0.3672,
+      "learning_rate": 1.397788507397949e-05,
+      "loss": 0.4983,
+      "step": 1836
+    },
+    {
+      "epoch": 0.3676,
+      "learning_rate": 1.3990691589802943e-05,
+      "loss": 0.1828,
+      "step": 1838
+    },
+    {
+      "epoch": 0.368,
+      "learning_rate": 1.4003490325568956e-05,
+      "loss": 0.2123,
+      "step": 1840
+    },
+    {
+      "epoch": 0.3684,
+      "learning_rate": 1.4016281256325688e-05,
+      "loss": 0.1257,
+      "step": 1842
+    },
+    {
+      "epoch": 0.3688,
+      "learning_rate": 1.4029064357136632e-05,
+      "loss": 0.3976,
+      "step": 1844
+    },
+    {
+      "epoch": 0.3692,
+      "learning_rate": 1.4041839603080411e-05,
+      "loss": 0.4432,
+      "step": 1846
+    },
+    {
+      "epoch": 0.3696,
+      "learning_rate": 1.4054606969251096e-05,
+      "loss": 0.4912,
+      "step": 1848
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4067366430758004e-05,
+      "loss": 0.2336,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3704,
+      "learning_rate": 1.4080117962725929e-05,
+      "loss": 0.5297,
+      "step": 1852
+    },
+    {
+      "epoch": 0.3708,
+      "learning_rate": 1.4092861540295107e-05,
+      "loss": 0.3135,
+      "step": 1854
+    },
+    {
+      "epoch": 0.3712,
+      "learning_rate": 1.4105597138621281e-05,
+      "loss": 0.8366,
+      "step": 1856
+    },
+    {
+      "epoch": 0.3716,
+      "learning_rate": 1.411832473287575e-05,
+      "loss": 0.5591,
+      "step": 1858
+    },
+    {
+      "epoch": 0.372,
+      "learning_rate": 1.4131044298245416e-05,
+      "loss": 0.6536,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3724,
+      "learning_rate": 1.414375580993284e-05,
+      "loss": 0.4745,
+      "step": 1862
+    },
+    {
+      "epoch": 0.3728,
+      "learning_rate": 1.4156459243156275e-05,
+      "loss": 0.2517,
+      "step": 1864
+    },
+    {
+      "epoch": 0.3732,
+      "learning_rate": 1.416915457314973e-05,
+      "loss": 0.2352,
+      "step": 1866
+    },
+    {
+      "epoch": 0.3736,
+      "learning_rate": 1.418184177516301e-05,
+      "loss": 0.3512,
+      "step": 1868
+    },
+    {
+      "epoch": 0.374,
+      "learning_rate": 1.4194520824461782e-05,
+      "loss": 0.1351,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3744,
+      "learning_rate": 1.420719169632754e-05,
+      "loss": 0.2865,
+      "step": 1872
+    },
+    {
+      "epoch": 0.3748,
+      "learning_rate": 1.4219854366057821e-05,
+      "loss": 0.5766,
+      "step": 1874
+    },
+    {
+      "epoch": 0.3752,
+      "learning_rate": 1.4232508808966085e-05,
+      "loss": 0.2683,
+      "step": 1876
+    },
+    {
+      "epoch": 0.3756,
+      "learning_rate": 1.424515500038185e-05,
+      "loss": 0.3567,
+      "step": 1878
+    },
+    {
+      "epoch": 0.376,
+      "learning_rate": 1.4257792915650735e-05,
+      "loss": 0.5921,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3764,
+      "learning_rate": 1.4270422530134425e-05,
+      "loss": 0.1803,
+      "step": 1882
+    },
+    {
+      "epoch": 0.3768,
+      "learning_rate": 1.4283043819210906e-05,
+      "loss": 0.4073,
+      "step": 1884
+    },
+    {
+      "epoch": 0.3772,
+      "learning_rate": 1.4295656758274288e-05,
+      "loss": 0.2364,
+      "step": 1886
+    },
+    {
+      "epoch": 0.3776,
+      "learning_rate": 1.430826132273499e-05,
+      "loss": 0.3634,
+      "step": 1888
+    },
+    {
+      "epoch": 0.378,
+      "learning_rate": 1.4320857488019826e-05,
+      "loss": 0.3056,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3784,
+      "learning_rate": 1.4333445229571857e-05,
+      "loss": 0.4522,
+      "step": 1892
+    },
+    {
+      "epoch": 0.3788,
+      "learning_rate": 1.4346024522850704e-05,
+      "loss": 0.536,
+      "step": 1894
+    },
+    {
+      "epoch": 0.3792,
+      "learning_rate": 1.4358595343332342e-05,
+      "loss": 0.3937,
+      "step": 1896
+    },
+    {
+      "epoch": 0.3796,
+      "learning_rate": 1.437115766650933e-05,
+      "loss": 0.5523,
+      "step": 1898
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4383711467890772e-05,
+      "loss": 0.4545,
+      "step": 1900
+    },
+    {
+      "epoch": 0.3804,
+      "learning_rate": 1.4396256723002398e-05,
+      "loss": 0.5919,
+      "step": 1902
+    },
+    {
+      "epoch": 0.3808,
+      "learning_rate": 1.4408793407386584e-05,
+      "loss": 0.0782,
+      "step": 1904
+    },
+    {
+      "epoch": 0.3812,
+      "learning_rate": 1.4421321496602423e-05,
+      "loss": 0.2252,
+      "step": 1906
+    },
+    {
+      "epoch": 0.3816,
+      "learning_rate": 1.4433840966225767e-05,
+      "loss": 0.6248,
+      "step": 1908
+    },
+    {
+      "epoch": 0.382,
+      "learning_rate": 1.444635179184927e-05,
+      "loss": 0.4717,
+      "step": 1910
+    },
+    {
+      "epoch": 0.3824,
+      "learning_rate": 1.4458853949082434e-05,
+      "loss": 0.0647,
+      "step": 1912
+    },
+    {
+      "epoch": 0.3828,
+      "learning_rate": 1.4471347413551665e-05,
+      "loss": 0.4582,
+      "step": 1914
+    },
+    {
+      "epoch": 0.3832,
+      "learning_rate": 1.4483832160900332e-05,
+      "loss": 0.3907,
+      "step": 1916
+    },
+    {
+      "epoch": 0.3836,
+      "learning_rate": 1.4496308166788731e-05,
+      "loss": 0.2284,
+      "step": 1918
+    },
+    {
+      "epoch": 0.384,
+      "learning_rate": 1.4508775406894315e-05,
+      "loss": 0.2488,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3844,
+      "learning_rate": 1.4521233856911499e-05,
+      "loss": 0.4022,
+      "step": 1922
+    },
+    {
+      "epoch": 0.3848,
+      "learning_rate": 1.4533683492551942e-05,
+      "loss": 0.7484,
+      "step": 1924
+    },
+    {
+      "epoch": 0.3852,
+      "learning_rate": 1.4546124289544446e-05,
+      "loss": 0.1981,
+      "step": 1926
+    },
+    {
+      "epoch": 0.3856,
+      "learning_rate": 1.4558556223634988e-05,
+      "loss": 0.4014,
+      "step": 1928
+    },
+    {
+      "epoch": 0.386,
+      "learning_rate": 1.4570979270586944e-05,
+      "loss": 0.5458,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3864,
+      "learning_rate": 1.4583393406180886e-05,
+      "loss": 0.2788,
+      "step": 1932
+    },
+    {
+      "epoch": 0.3868,
+      "learning_rate": 1.4595798606214882e-05,
+      "loss": 0.1315,
+      "step": 1934
+    },
+    {
+      "epoch": 0.3872,
+      "learning_rate": 1.460819484650431e-05,
+      "loss": 0.3745,
+      "step": 1936
+    },
+    {
+      "epoch": 0.3876,
+      "learning_rate": 1.4620582102882086e-05,
+      "loss": 0.506,
+      "step": 1938
+    },
+    {
+      "epoch": 0.388,
+      "learning_rate": 1.4632960351198618e-05,
+      "loss": 0.6673,
+      "step": 1940
+    },
+    {
+      "epoch": 0.3884,
+      "learning_rate": 1.4645329567321875e-05,
+      "loss": 0.2244,
+      "step": 1942
+    },
+    {
+      "epoch": 0.3888,
+      "learning_rate": 1.4657689727137441e-05,
+      "loss": 0.4186,
+      "step": 1944
+    },
+    {
+      "epoch": 0.3892,
+      "learning_rate": 1.4670040806548551e-05,
+      "loss": 0.2661,
+      "step": 1946
+    },
+    {
+      "epoch": 0.3896,
+      "learning_rate": 1.468238278147614e-05,
+      "loss": 0.2438,
+      "step": 1948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4694715627858904e-05,
+      "loss": 0.7098,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3904,
+      "learning_rate": 1.470703932165332e-05,
+      "loss": 0.652,
+      "step": 1952
+    },
+    {
+      "epoch": 0.3908,
+      "learning_rate": 1.471935383883372e-05,
+      "loss": 0.4836,
+      "step": 1954
+    },
+    {
+      "epoch": 0.3912,
+      "learning_rate": 1.4731659155392339e-05,
+      "loss": 0.1309,
+      "step": 1956
+    },
+    {
+      "epoch": 0.3916,
+      "learning_rate": 1.4743955247339286e-05,
+      "loss": 0.0943,
+      "step": 1958
+    },
+    {
+      "epoch": 0.392,
+      "learning_rate": 1.4756242090702744e-05,
+      "loss": 0.5174,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3924,
+      "learning_rate": 1.476851966152887e-05,
+      "loss": 0.1902,
+      "step": 1962
+    },
+    {
+      "epoch": 0.3928,
+      "learning_rate": 1.4780787935881913e-05,
+      "loss": 0.4181,
+      "step": 1964
+    },
+    {
+      "epoch": 0.3932,
+      "learning_rate": 1.4793046889844255e-05,
+      "loss": 1.2857,
+      "step": 1966
+    },
+    {
+      "epoch": 0.3936,
+      "learning_rate": 1.4805296499516397e-05,
+      "loss": 0.4257,
+      "step": 1968
+    },
+    {
+      "epoch": 0.394,
+      "learning_rate": 1.4817536741017155e-05,
+      "loss": 0.5827,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3944,
+      "learning_rate": 1.482976759048351e-05,
+      "loss": 0.4568,
+      "step": 1972
+    },
+    {
+      "epoch": 0.3948,
+      "learning_rate": 1.4841989024070809e-05,
+      "loss": 0.4093,
+      "step": 1974
+    },
+    {
+      "epoch": 0.3952,
+      "learning_rate": 1.485420101795274e-05,
+      "loss": 0.3353,
+      "step": 1976
+    },
+    {
+      "epoch": 0.3956,
+      "learning_rate": 1.4866403548321385e-05,
+      "loss": 0.3903,
+      "step": 1978
+    },
+    {
+      "epoch": 0.396,
+      "learning_rate": 1.4878596591387327e-05,
+      "loss": 0.2244,
+      "step": 1980
+    },
+    {
+      "epoch": 0.3964,
+      "learning_rate": 1.4890780123379563e-05,
+      "loss": 0.7337,
+      "step": 1982
+    },
+    {
+      "epoch": 0.3968,
+      "learning_rate": 1.4902954120545686e-05,
+      "loss": 0.4356,
+      "step": 1984
+    },
+    {
+      "epoch": 0.3972,
+      "learning_rate": 1.491511855915187e-05,
+      "loss": 0.4108,
+      "step": 1986
+    },
+    {
+      "epoch": 0.3976,
+      "learning_rate": 1.4927273415482913e-05,
+      "loss": 0.2177,
+      "step": 1988
+    },
+    {
+      "epoch": 0.398,
+      "learning_rate": 1.4939418665842307e-05,
+      "loss": 0.163,
+      "step": 1990
+    },
+    {
+      "epoch": 0.3984,
+      "learning_rate": 1.4951554286552261e-05,
+      "loss": 0.2953,
+      "step": 1992
+    },
+    {
+      "epoch": 0.3988,
+      "learning_rate": 1.4963680253953763e-05,
+      "loss": 0.716,
+      "step": 1994
+    },
+    {
+      "epoch": 0.3992,
+      "learning_rate": 1.4975796544406617e-05,
+      "loss": 0.6106,
+      "step": 1996
+    },
+    {
+      "epoch": 0.3996,
+      "learning_rate": 1.49879031342895e-05,
+      "loss": 0.1675,
+      "step": 1998
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.4999999999999992e-05,
+      "loss": 0.5611,
+      "step": 2000
+    },
+    {
+      "epoch": 0.4004,
+      "learning_rate": 1.501208711795465e-05,
+      "loss": 0.1502,
+      "step": 2002
+    },
+    {
+      "epoch": 0.4008,
+      "learning_rate": 1.502416446458897e-05,
+      "loss": 0.3786,
+      "step": 2004
+    },
+    {
+      "epoch": 0.4012,
+      "learning_rate": 1.5036232016357613e-05,
+      "loss": 0.4304,
+      "step": 2006
+    },
+    {
+      "epoch": 0.4016,
+      "learning_rate": 1.5048289749734206e-05,
+      "loss": 0.471,
+      "step": 2008
+    },
+    {
+      "epoch": 0.402,
+      "learning_rate": 1.5060337641211642e-05,
+      "loss": 0.6029,
+      "step": 2010
+    },
+    {
+      "epoch": 0.4024,
+      "learning_rate": 1.5072375667301895e-05,
+      "loss": 0.2245,
+      "step": 2012
+    },
+    {
+      "epoch": 0.4028,
+      "learning_rate": 1.5084403804536214e-05,
+      "loss": 0.2382,
+      "step": 2014
+    },
+    {
+      "epoch": 0.4032,
+      "learning_rate": 1.5096422029465178e-05,
+      "loss": 0.5211,
+      "step": 2016
+    },
+    {
+      "epoch": 0.4036,
+      "learning_rate": 1.5108430318658597e-05,
+      "loss": 0.1845,
+      "step": 2018
+    },
+    {
+      "epoch": 0.404,
+      "learning_rate": 1.5120428648705714e-05,
+      "loss": 0.1657,
+      "step": 2020
+    },
+    {
+      "epoch": 0.4044,
+      "learning_rate": 1.513241699621517e-05,
+      "loss": 0.2888,
+      "step": 2022
+    },
+    {
+      "epoch": 0.4048,
+      "learning_rate": 1.5144395337815064e-05,
+      "loss": 0.3165,
+      "step": 2024
+    },
+    {
+      "epoch": 0.4052,
+      "learning_rate": 1.5156363650153008e-05,
+      "loss": 0.3731,
+      "step": 2026
+    },
+    {
+      "epoch": 0.4056,
+      "learning_rate": 1.5168321909896166e-05,
+      "loss": 0.383,
+      "step": 2028
+    },
+    {
+      "epoch": 0.406,
+      "learning_rate": 1.51802700937313e-05,
+      "loss": 0.6397,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4064,
+      "learning_rate": 1.5192208178364808e-05,
+      "loss": 0.2375,
+      "step": 2032
+    },
+    {
+      "epoch": 0.4068,
+      "learning_rate": 1.5204136140522792e-05,
+      "loss": 0.5406,
+      "step": 2034
+    },
+    {
+      "epoch": 0.4072,
+      "learning_rate": 1.521605395695107e-05,
+      "loss": 0.1518,
+      "step": 2036
+    },
+    {
+      "epoch": 0.4076,
+      "learning_rate": 1.522796160441526e-05,
+      "loss": 1.0324,
+      "step": 2038
+    },
+    {
+      "epoch": 0.408,
+      "learning_rate": 1.5239859059700784e-05,
+      "loss": 0.6884,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4084,
+      "learning_rate": 1.5251746299612964e-05,
+      "loss": 0.337,
+      "step": 2042
+    },
+    {
+      "epoch": 0.4088,
+      "learning_rate": 1.526362330097697e-05,
+      "loss": 0.404,
+      "step": 2044
+    },
+    {
+      "epoch": 0.4092,
+      "learning_rate": 1.5275490040638044e-05,
+      "loss": 0.396,
+      "step": 2046
+    },
+    {
+      "epoch": 0.4096,
+      "learning_rate": 1.5287346495461322e-05,
+      "loss": 0.2327,
+      "step": 2048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.529919264233204e-05,
+      "loss": 0.4314,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4104,
+      "learning_rate": 1.531102845815557e-05,
+      "loss": 0.5474,
+      "step": 2052
+    },
+    {
+      "epoch": 0.4108,
+      "learning_rate": 1.5322853919857327e-05,
+      "loss": 0.7187,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4112,
+      "learning_rate": 1.5334669004383025e-05,
+      "loss": 0.5364,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4116,
+      "learning_rate": 1.5346473688698514e-05,
+      "loss": 0.0958,
+      "step": 2058
+    },
+    {
+      "epoch": 0.412,
+      "learning_rate": 1.5358267949789968e-05,
+      "loss": 0.3095,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4124,
+      "learning_rate": 1.537005176466387e-05,
+      "loss": 0.3301,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4128,
+      "learning_rate": 1.5381825110347072e-05,
+      "loss": 0.4175,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4132,
+      "learning_rate": 1.539358796388683e-05,
+      "loss": 0.4154,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4136,
+      "learning_rate": 1.540534030235087e-05,
+      "loss": 0.5686,
+      "step": 2068
+    },
+    {
+      "epoch": 0.414,
+      "learning_rate": 1.5417082102827397e-05,
+      "loss": 0.5586,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4144,
+      "learning_rate": 1.542881334242517e-05,
+      "loss": 0.6943,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4148,
+      "learning_rate": 1.5440533998273542e-05,
+      "loss": 0.5523,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4152,
+      "learning_rate": 1.5452244047522493e-05,
+      "loss": 0.6599,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4156,
+      "learning_rate": 1.54639434673427e-05,
+      "loss": 0.4307,
+      "step": 2078
+    },
+    {
+      "epoch": 0.416,
+      "learning_rate": 1.5475632234925495e-05,
+      "loss": 0.3563,
+      "step": 2080
+    },
+    {
+      "epoch": 0.4164,
+      "learning_rate": 1.548731032748309e-05,
+      "loss": 0.2965,
+      "step": 2082
+    },
+    {
+      "epoch": 0.4168,
+      "learning_rate": 1.5498977722248388e-05,
+      "loss": 0.2128,
+      "step": 2084
+    },
+    {
+      "epoch": 0.4172,
+      "learning_rate": 1.551063439647525e-05,
+      "loss": 0.5431,
+      "step": 2086
+    },
+    {
+      "epoch": 0.4176,
+      "learning_rate": 1.552228032743839e-05,
+      "loss": 0.3336,
+      "step": 2088
+    },
+    {
+      "epoch": 0.418,
+      "learning_rate": 1.553391549243343e-05,
+      "loss": 0.3082,
+      "step": 2090
+    },
+    {
+      "epoch": 0.4184,
+      "learning_rate": 1.5545539868777075e-05,
+      "loss": 0.6196,
+      "step": 2092
+    },
+    {
+      "epoch": 0.4188,
+      "learning_rate": 1.5557153433806954e-05,
+      "loss": 0.7747,
+      "step": 2094
+    },
+    {
+      "epoch": 0.4192,
+      "learning_rate": 1.556875616488188e-05,
+      "loss": 0.1838,
+      "step": 2096
+    },
+    {
+      "epoch": 0.4196,
+      "learning_rate": 1.55803480393817e-05,
+      "loss": 0.9596,
+      "step": 2098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.5591929034707468e-05,
+      "loss": 0.0986,
+      "step": 2100
+    },
+    {
+      "epoch": 0.4204,
+      "learning_rate": 1.5603499128281447e-05,
+      "loss": 0.4402,
+      "step": 2102
+    },
+    {
+      "epoch": 0.4208,
+      "learning_rate": 1.5615058297547144e-05,
+      "loss": 0.056,
+      "step": 2104
+    },
+    {
+      "epoch": 0.4212,
+      "learning_rate": 1.5626606519969366e-05,
+      "loss": 0.1684,
+      "step": 2106
+    },
+    {
+      "epoch": 0.4216,
+      "learning_rate": 1.5638143773034268e-05,
+      "loss": 0.2706,
+      "step": 2108
+    },
+    {
+      "epoch": 0.422,
+      "learning_rate": 1.5649670034249376e-05,
+      "loss": 0.4148,
+      "step": 2110
+    },
+    {
+      "epoch": 0.4224,
+      "learning_rate": 1.5661185281143663e-05,
+      "loss": 0.4329,
+      "step": 2112
+    },
+    {
+      "epoch": 0.4228,
+      "learning_rate": 1.5672689491267562e-05,
+      "loss": 0.2438,
+      "step": 2114
+    },
+    {
+      "epoch": 0.4232,
+      "learning_rate": 1.5684182642193024e-05,
+      "loss": 0.0583,
+      "step": 2116
+    },
+    {
+      "epoch": 0.4236,
+      "learning_rate": 1.5695664711513582e-05,
+      "loss": 0.9311,
+      "step": 2118
+    },
+    {
+      "epoch": 0.424,
+      "learning_rate": 1.5707135676844312e-05,
+      "loss": 0.2396,
+      "step": 2120
+    },
+    {
+      "epoch": 0.4244,
+      "learning_rate": 1.5718595515822016e-05,
+      "loss": 0.5753,
+      "step": 2122
+    },
+    {
+      "epoch": 0.4248,
+      "learning_rate": 1.5730044206105146e-05,
+      "loss": 0.4207,
+      "step": 2124
+    },
+    {
+      "epoch": 0.4252,
+      "learning_rate": 1.574148172537389e-05,
+      "loss": 0.5289,
+      "step": 2126
+    },
+    {
+      "epoch": 0.4256,
+      "learning_rate": 1.5752908051330232e-05,
+      "loss": 0.9355,
+      "step": 2128
+    },
+    {
+      "epoch": 0.426,
+      "learning_rate": 1.5764323161697923e-05,
+      "loss": 0.4498,
+      "step": 2130
+    },
+    {
+      "epoch": 0.4264,
+      "learning_rate": 1.577572703422268e-05,
+      "loss": 0.3783,
+      "step": 2132
+    },
+    {
+      "epoch": 0.4268,
+      "learning_rate": 1.5787119646672025e-05,
+      "loss": 0.4763,
+      "step": 2134
+    },
+    {
+      "epoch": 0.4272,
+      "learning_rate": 1.579850097683548e-05,
+      "loss": 0.0663,
+      "step": 2136
+    },
+    {
+      "epoch": 0.4276,
+      "learning_rate": 1.58098710025246e-05,
+      "loss": 0.3966,
+      "step": 2138
+    },
+    {
+      "epoch": 0.428,
+      "learning_rate": 1.582122970157288e-05,
+      "loss": 0.6732,
+      "step": 2140
+    },
+    {
+      "epoch": 0.4284,
+      "learning_rate": 1.5832577051836016e-05,
+      "loss": 0.2451,
+      "step": 2142
+    },
+    {
+      "epoch": 0.4288,
+      "learning_rate": 1.5843913031191722e-05,
+      "loss": 0.3867,
+      "step": 2144
+    },
+    {
+      "epoch": 0.4292,
+      "learning_rate": 1.585523761753994e-05,
+      "loss": 0.8255,
+      "step": 2146
+    },
+    {
+      "epoch": 0.4296,
+      "learning_rate": 1.586655078880281e-05,
+      "loss": 0.2558,
+      "step": 2148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.587785252292473e-05,
+      "loss": 0.1291,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4304,
+      "learning_rate": 1.5889142797872383e-05,
+      "loss": 0.314,
+      "step": 2152
+    },
+    {
+      "epoch": 0.4308,
+      "learning_rate": 1.5900421591634806e-05,
+      "loss": 0.7029,
+      "step": 2154
+    },
+    {
+      "epoch": 0.4312,
+      "learning_rate": 1.5911688882223415e-05,
+      "loss": 0.7267,
+      "step": 2156
+    },
+    {
+      "epoch": 0.4316,
+      "learning_rate": 1.5922944647672044e-05,
+      "loss": 0.4953,
+      "step": 2158
+    },
+    {
+      "epoch": 0.432,
+      "learning_rate": 1.5934188866037007e-05,
+      "loss": 0.1341,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4324,
+      "learning_rate": 1.5945421515397125e-05,
+      "loss": 0.1748,
+      "step": 2162
+    },
+    {
+      "epoch": 0.4328,
+      "learning_rate": 1.5956642573853787e-05,
+      "loss": 0.1322,
+      "step": 2164
+    },
+    {
+      "epoch": 0.4332,
+      "learning_rate": 1.5967852019530918e-05,
+      "loss": 0.3274,
+      "step": 2166
+    },
+    {
+      "epoch": 0.4336,
+      "learning_rate": 1.5979049830575193e-05,
+      "loss": 0.6823,
+      "step": 2168
+    },
+    {
+      "epoch": 0.434,
+      "learning_rate": 1.599023598515585e-05,
+      "loss": 0.1735,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4344,
+      "learning_rate": 1.6001410461464945e-05,
+      "loss": 0.326,
+      "step": 2172
+    },
+    {
+      "epoch": 0.4348,
+      "learning_rate": 1.601257323771727e-05,
+      "loss": 0.3176,
+      "step": 2174
+    },
+    {
+      "epoch": 0.4352,
+      "learning_rate": 1.6023724292150377e-05,
+      "loss": 0.4639,
+      "step": 2176
+    },
+    {
+      "epoch": 0.4356,
+      "learning_rate": 1.6034863603024768e-05,
+      "loss": 0.2781,
+      "step": 2178
+    },
+    {
+      "epoch": 0.436,
+      "learning_rate": 1.604599114862375e-05,
+      "loss": 0.7703,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4364,
+      "learning_rate": 1.6057106907253614e-05,
+      "loss": 0.154,
+      "step": 2182
+    },
+    {
+      "epoch": 0.4368,
+      "learning_rate": 1.606821085724362e-05,
+      "loss": 0.4011,
+      "step": 2184
+    },
+    {
+      "epoch": 0.4372,
+      "learning_rate": 1.6079302976946052e-05,
+      "loss": 0.3101,
+      "step": 2186
+    },
+    {
+      "epoch": 0.4376,
+      "learning_rate": 1.6090383244736253e-05,
+      "loss": 0.2479,
+      "step": 2188
+    },
+    {
+      "epoch": 0.438,
+      "learning_rate": 1.6101451639012675e-05,
+      "loss": 0.3892,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4384,
+      "learning_rate": 1.6112508138196912e-05,
+      "loss": 0.4567,
+      "step": 2192
+    },
+    {
+      "epoch": 0.4388,
+      "learning_rate": 1.6123552720733763e-05,
+      "loss": 0.4153,
+      "step": 2194
+    },
+    {
+      "epoch": 0.4392,
+      "learning_rate": 1.613458536509124e-05,
+      "loss": 0.461,
+      "step": 2196
+    },
+    {
+      "epoch": 0.4396,
+      "learning_rate": 1.614560604976064e-05,
+      "loss": 0.4327,
+      "step": 2198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.615661475325658e-05,
+      "loss": 0.3008,
+      "step": 2200
+    },
+    {
+      "epoch": 0.4404,
+      "learning_rate": 1.616761145411702e-05,
+      "loss": 0.2992,
+      "step": 2202
+    },
+    {
+      "epoch": 0.4408,
+      "learning_rate": 1.6178596130903352e-05,
+      "loss": 0.1946,
+      "step": 2204
+    },
+    {
+      "epoch": 0.4412,
+      "learning_rate": 1.618956876220034e-05,
+      "loss": 0.5753,
+      "step": 2206
+    },
+    {
+      "epoch": 0.4416,
+      "learning_rate": 1.620052932661632e-05,
+      "loss": 0.126,
+      "step": 2208
+    },
+    {
+      "epoch": 0.442,
+      "learning_rate": 1.621147780278311e-05,
+      "loss": 0.2166,
+      "step": 2210
+    },
+    {
+      "epoch": 0.4424,
+      "learning_rate": 1.6222414169356056e-05,
+      "loss": 0.1903,
+      "step": 2212
+    },
+    {
+      "epoch": 0.4428,
+      "learning_rate": 1.6233338405014204e-05,
+      "loss": 0.2995,
+      "step": 2214
+    },
+    {
+      "epoch": 0.4432,
+      "learning_rate": 1.6244250488460146e-05,
+      "loss": 0.2454,
+      "step": 2216
+    },
+    {
+      "epoch": 0.4436,
+      "learning_rate": 1.6255150398420273e-05,
+      "loss": 0.2782,
+      "step": 2218
+    },
+    {
+      "epoch": 0.444,
+      "learning_rate": 1.6266038113644605e-05,
+      "loss": 0.1758,
+      "step": 2220
+    },
+    {
+      "epoch": 0.4444,
+      "learning_rate": 1.6276913612907005e-05,
+      "loss": 0.773,
+      "step": 2222
+    },
+    {
+      "epoch": 0.4448,
+      "learning_rate": 1.6287776875005127e-05,
+      "loss": 0.7094,
+      "step": 2224
+    },
+    {
+      "epoch": 0.4452,
+      "learning_rate": 1.6298627878760488e-05,
+      "loss": 0.3022,
+      "step": 2226
+    },
+    {
+      "epoch": 0.4456,
+      "learning_rate": 1.6309466603018497e-05,
+      "loss": 0.1311,
+      "step": 2228
+    },
+    {
+      "epoch": 0.446,
+      "learning_rate": 1.6320293026648508e-05,
+      "loss": 0.2427,
+      "step": 2230
+    },
+    {
+      "epoch": 0.4464,
+      "learning_rate": 1.6331107128543856e-05,
+      "loss": 0.5957,
+      "step": 2232
+    },
+    {
+      "epoch": 0.4468,
+      "learning_rate": 1.634190888762189e-05,
+      "loss": 0.6103,
+      "step": 2234
+    },
+    {
+      "epoch": 0.4472,
+      "learning_rate": 1.635269828282404e-05,
+      "loss": 0.199,
+      "step": 2236
+    },
+    {
+      "epoch": 0.4476,
+      "learning_rate": 1.6363475293115818e-05,
+      "loss": 0.3281,
+      "step": 2238
+    },
+    {
+      "epoch": 0.448,
+      "learning_rate": 1.6374239897486905e-05,
+      "loss": 0.409,
+      "step": 2240
+    },
+    {
+      "epoch": 0.4484,
+      "learning_rate": 1.6384992074951118e-05,
+      "loss": 0.302,
+      "step": 2242
+    },
+    {
+      "epoch": 0.4488,
+      "learning_rate": 1.6395731804546575e-05,
+      "loss": 0.6546,
+      "step": 2244
+    },
+    {
+      "epoch": 0.4492,
+      "learning_rate": 1.640645906533561e-05,
+      "loss": 0.5806,
+      "step": 2246
+    },
+    {
+      "epoch": 0.4496,
+      "learning_rate": 1.6417173836404878e-05,
+      "loss": 0.7055,
+      "step": 2248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6427876096865397e-05,
+      "loss": 0.214,
+      "step": 2250
+    },
+    {
+      "epoch": 0.4504,
+      "learning_rate": 1.643856582585253e-05,
+      "loss": 0.0262,
+      "step": 2252
+    },
+    {
+      "epoch": 0.4508,
+      "learning_rate": 1.6449243002526146e-05,
+      "loss": 0.1717,
+      "step": 2254
+    },
+    {
+      "epoch": 0.4512,
+      "learning_rate": 1.6459907606070513e-05,
+      "loss": 0.2994,
+      "step": 2256
+    },
+    {
+      "epoch": 0.4516,
+      "learning_rate": 1.6470559615694445e-05,
+      "loss": 0.4484,
+      "step": 2258
+    },
+    {
+      "epoch": 0.452,
+      "learning_rate": 1.6481199010631312e-05,
+      "loss": 0.7119,
+      "step": 2260
+    },
+    {
+      "epoch": 0.4524,
+      "learning_rate": 1.649182577013905e-05,
+      "loss": 0.2169,
+      "step": 2262
+    },
+    {
+      "epoch": 0.4528,
+      "learning_rate": 1.650243987350029e-05,
+      "loss": 0.3641,
+      "step": 2264
+    },
+    {
+      "epoch": 0.4532,
+      "learning_rate": 1.6513041300022253e-05,
+      "loss": 0.5216,
+      "step": 2266
+    },
+    {
+      "epoch": 0.4536,
+      "learning_rate": 1.652363002903693e-05,
+      "loss": 0.2793,
+      "step": 2268
+    },
+    {
+      "epoch": 0.454,
+      "learning_rate": 1.6534206039901054e-05,
+      "loss": 0.3332,
+      "step": 2270
+    },
+    {
+      "epoch": 0.4544,
+      "learning_rate": 1.6544769311996146e-05,
+      "loss": 0.2331,
+      "step": 2272
+    },
+    {
+      "epoch": 0.4548,
+      "learning_rate": 1.655531982472857e-05,
+      "loss": 0.6003,
+      "step": 2274
+    },
+    {
+      "epoch": 0.4552,
+      "learning_rate": 1.656585755752956e-05,
+      "loss": 0.6675,
+      "step": 2276
+    },
+    {
+      "epoch": 0.4556,
+      "learning_rate": 1.657638248985527e-05,
+      "loss": 0.7415,
+      "step": 2278
+    },
+    {
+      "epoch": 0.456,
+      "learning_rate": 1.65868946011868e-05,
+      "loss": 0.2189,
+      "step": 2280
+    },
+    {
+      "epoch": 0.4564,
+      "learning_rate": 1.6597393871030257e-05,
+      "loss": 0.2262,
+      "step": 2282
+    },
+    {
+      "epoch": 0.4568,
+      "learning_rate": 1.660788027891677e-05,
+      "loss": 0.2986,
+      "step": 2284
+    },
+    {
+      "epoch": 0.4572,
+      "learning_rate": 1.6618353804402573e-05,
+      "loss": 0.0859,
+      "step": 2286
+    },
+    {
+      "epoch": 0.4576,
+      "learning_rate": 1.6628814427068944e-05,
+      "loss": 0.4054,
+      "step": 2288
+    },
+    {
+      "epoch": 0.458,
+      "learning_rate": 1.663926212652242e-05,
+      "loss": 0.4217,
+      "step": 2290
+    },
+    {
+      "epoch": 0.4584,
+      "learning_rate": 1.6649696882394625e-05,
+      "loss": 0.2112,
+      "step": 2292
+    },
+    {
+      "epoch": 0.4588,
+      "learning_rate": 1.666011867434252e-05,
+      "loss": 0.3053,
+      "step": 2294
+    },
+    {
+      "epoch": 0.4592,
+      "learning_rate": 1.667052748204825e-05,
+      "loss": 0.7078,
+      "step": 2296
+    },
+    {
+      "epoch": 0.4596,
+      "learning_rate": 1.6680923285219308e-05,
+      "loss": 0.5955,
+      "step": 2298
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6691306063588583e-05,
+      "loss": 0.6973,
+      "step": 2300
+    },
+    {
+      "epoch": 0.4604,
+      "learning_rate": 1.6701675796914273e-05,
+      "loss": 0.2172,
+      "step": 2302
+    },
+    {
+      "epoch": 0.4608,
+      "learning_rate": 1.6712032464980094e-05,
+      "loss": 0.6912,
+      "step": 2304
+    },
+    {
+      "epoch": 0.4612,
+      "learning_rate": 1.672237604759516e-05,
+      "loss": 0.33,
+      "step": 2306
+    },
+    {
+      "epoch": 0.4616,
+      "learning_rate": 1.6732706524594138e-05,
+      "loss": 0.3499,
+      "step": 2308
+    },
+    {
+      "epoch": 0.462,
+      "learning_rate": 1.6743023875837233e-05,
+      "loss": 0.5602,
+      "step": 2310
+    },
+    {
+      "epoch": 0.4624,
+      "learning_rate": 1.6753328081210244e-05,
+      "loss": 0.5344,
+      "step": 2312
+    },
+    {
+      "epoch": 0.4628,
+      "learning_rate": 1.6763619120624592e-05,
+      "loss": 0.1335,
+      "step": 2314
+    },
+    {
+      "epoch": 0.4632,
+      "learning_rate": 1.6773896974017373e-05,
+      "loss": 0.1973,
+      "step": 2316
+    },
+    {
+      "epoch": 0.4636,
+      "learning_rate": 1.6784161621351377e-05,
+      "loss": 0.4247,
+      "step": 2318
+    },
+    {
+      "epoch": 0.464,
+      "learning_rate": 1.679441304261516e-05,
+      "loss": 0.3894,
+      "step": 2320
+    },
+    {
+      "epoch": 0.4644,
+      "learning_rate": 1.6804651217823048e-05,
+      "loss": 0.4548,
+      "step": 2322
+    },
+    {
+      "epoch": 0.4648,
+      "learning_rate": 1.681487612701519e-05,
+      "loss": 0.3654,
+      "step": 2324
+    },
+    {
+      "epoch": 0.4652,
+      "learning_rate": 1.6825087750257624e-05,
+      "loss": 0.3532,
+      "step": 2326
+    },
+    {
+      "epoch": 0.4656,
+      "learning_rate": 1.683528606764222e-05,
+      "loss": 0.1247,
+      "step": 2328
+    },
+    {
+      "epoch": 0.466,
+      "learning_rate": 1.6845471059286893e-05,
+      "loss": 0.3859,
+      "step": 2330
+    },
+    {
+      "epoch": 0.4664,
+      "learning_rate": 1.6855642705335428e-05,
+      "loss": 0.332,
+      "step": 2332
+    },
+    {
+      "epoch": 0.4668,
+      "learning_rate": 1.6865800985957718e-05,
+      "loss": 0.4711,
+      "step": 2334
+    },
+    {
+      "epoch": 0.4672,
+      "learning_rate": 1.687594588134968e-05,
+      "loss": 0.0668,
+      "step": 2336
+    },
+    {
+      "epoch": 0.4676,
+      "learning_rate": 1.6886077371733275e-05,
+      "loss": 0.4968,
+      "step": 2338
+    },
+    {
+      "epoch": 0.468,
+      "learning_rate": 1.68961954373567e-05,
+      "loss": 0.6808,
+      "step": 2340
+    },
+    {
+      "epoch": 0.4684,
+      "learning_rate": 1.690630005849423e-05,
+      "loss": 0.335,
+      "step": 2342
+    },
+    {
+      "epoch": 0.4688,
+      "learning_rate": 1.6916391215446403e-05,
+      "loss": 0.7946,
+      "step": 2344
+    },
+    {
+      "epoch": 0.4692,
+      "learning_rate": 1.6926468888539988e-05,
+      "loss": 0.538,
+      "step": 2346
+    },
+    {
+      "epoch": 0.4696,
+      "learning_rate": 1.693653305812805e-05,
+      "loss": 0.3641,
+      "step": 2348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6946583704589973e-05,
+      "loss": 0.4414,
+      "step": 2350
+    },
+    {
+      "epoch": 0.4704,
+      "learning_rate": 1.6956620808331505e-05,
+      "loss": 0.8821,
+      "step": 2352
+    },
+    {
+      "epoch": 0.4708,
+      "learning_rate": 1.6966644349784805e-05,
+      "loss": 0.6062,
+      "step": 2354
+    },
+    {
+      "epoch": 0.4712,
+      "learning_rate": 1.697665430940846e-05,
+      "loss": 0.4634,
+      "step": 2356
+    },
+    {
+      "epoch": 0.4716,
+      "learning_rate": 1.698665066768755e-05,
+      "loss": 0.4623,
+      "step": 2358
+    },
+    {
+      "epoch": 0.472,
+      "learning_rate": 1.699663340513365e-05,
+      "loss": 0.2445,
+      "step": 2360
+    },
+    {
+      "epoch": 0.4724,
+      "learning_rate": 1.7006602502284913e-05,
+      "loss": 0.4793,
+      "step": 2362
+    },
+    {
+      "epoch": 0.4728,
+      "learning_rate": 1.7016557939706068e-05,
+      "loss": 0.6003,
+      "step": 2364
+    },
+    {
+      "epoch": 0.4732,
+      "learning_rate": 1.70264996979885e-05,
+      "loss": 0.218,
+      "step": 2366
+    },
+    {
+      "epoch": 0.4736,
+      "learning_rate": 1.7036427757750198e-05,
+      "loss": 0.3163,
+      "step": 2368
+    },
+    {
+      "epoch": 0.474,
+      "learning_rate": 1.7046342099635938e-05,
+      "loss": 0.8093,
+      "step": 2370
+    },
+    {
+      "epoch": 0.4744,
+      "learning_rate": 1.7056242704317212e-05,
+      "loss": 0.4662,
+      "step": 2372
+    },
+    {
+      "epoch": 0.4748,
+      "learning_rate": 1.706612955249224e-05,
+      "loss": 0.3372,
+      "step": 2374
+    },
+    {
+      "epoch": 0.4752,
+      "learning_rate": 1.7076002624886156e-05,
+      "loss": 0.4513,
+      "step": 2376
+    },
+    {
+      "epoch": 0.4756,
+      "learning_rate": 1.708586190225085e-05,
+      "loss": 0.1814,
+      "step": 2378
+    },
+    {
+      "epoch": 0.476,
+      "learning_rate": 1.709570736536521e-05,
+      "loss": 0.7883,
+      "step": 2380
+    },
+    {
+      "epoch": 0.4764,
+      "learning_rate": 1.710553899503496e-05,
+      "loss": 0.4103,
+      "step": 2382
+    },
+    {
+      "epoch": 0.4768,
+      "learning_rate": 1.7115356772092844e-05,
+      "loss": 0.4666,
+      "step": 2384
+    },
+    {
+      "epoch": 0.4772,
+      "learning_rate": 1.7125160677398625e-05,
+      "loss": 0.0833,
+      "step": 2386
+    },
+    {
+      "epoch": 0.4776,
+      "learning_rate": 1.7134950691839063e-05,
+      "loss": 0.253,
+      "step": 2388
+    },
+    {
+      "epoch": 0.478,
+      "learning_rate": 1.7144726796328034e-05,
+      "loss": 0.624,
+      "step": 2390
+    },
+    {
+      "epoch": 0.4784,
+      "learning_rate": 1.7154488971806518e-05,
+      "loss": 0.5618,
+      "step": 2392
+    },
+    {
+      "epoch": 0.4788,
+      "learning_rate": 1.716423719924266e-05,
+      "loss": 1.4912,
+      "step": 2394
+    },
+    {
+      "epoch": 0.4792,
+      "learning_rate": 1.7173971459631783e-05,
+      "loss": 0.3042,
+      "step": 2396
+    },
+    {
+      "epoch": 0.4796,
+      "learning_rate": 1.718369173399646e-05,
+      "loss": 0.477,
+      "step": 2398
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.7193398003386507e-05,
+      "loss": 0.1874,
+      "step": 2400
+    },
+    {
+      "epoch": 0.4804,
+      "learning_rate": 1.7203090248879063e-05,
+      "loss": 0.2845,
+      "step": 2402
+    },
+    {
+      "epoch": 0.4808,
+      "learning_rate": 1.7212768451578602e-05,
+      "loss": 0.3802,
+      "step": 2404
+    },
+    {
+      "epoch": 0.4812,
+      "learning_rate": 1.7222432592616963e-05,
+      "loss": 0.1641,
+      "step": 2406
+    },
+    {
+      "epoch": 0.4816,
+      "learning_rate": 1.7232082653153416e-05,
+      "loss": 0.3955,
+      "step": 2408
+    },
+    {
+      "epoch": 0.482,
+      "learning_rate": 1.724171861437467e-05,
+      "loss": 0.3767,
+      "step": 2410
+    },
+    {
+      "epoch": 0.4824,
+      "learning_rate": 1.7251340457494937e-05,
+      "loss": 0.6858,
+      "step": 2412
+    },
+    {
+      "epoch": 0.4828,
+      "learning_rate": 1.726094816375591e-05,
+      "loss": 0.2491,
+      "step": 2414
+    },
+    {
+      "epoch": 0.4832,
+      "learning_rate": 1.7270541714426923e-05,
+      "loss": 0.4403,
+      "step": 2416
+    },
+    {
+      "epoch": 0.4836,
+      "learning_rate": 1.7280121090804817e-05,
+      "loss": 0.684,
+      "step": 2418
+    },
+    {
+      "epoch": 0.484,
+      "learning_rate": 1.7289686274214106e-05,
+      "loss": 0.3063,
+      "step": 2420
+    },
+    {
+      "epoch": 0.4844,
+      "learning_rate": 1.7299237246007018e-05,
+      "loss": 0.6492,
+      "step": 2422
+    },
+    {
+      "epoch": 0.4848,
+      "learning_rate": 1.7308773987563393e-05,
+      "loss": 0.6284,
+      "step": 2424
+    },
+    {
+      "epoch": 0.4852,
+      "learning_rate": 1.7318296480290912e-05,
+      "loss": 0.1235,
+      "step": 2426
+    },
+    {
+      "epoch": 0.4856,
+      "learning_rate": 1.732780470562496e-05,
+      "loss": 0.4292,
+      "step": 2428
+    },
+    {
+      "epoch": 0.486,
+      "learning_rate": 1.7337298645028764e-05,
+      "loss": 0.4949,
+      "step": 2430
+    },
+    {
+      "epoch": 0.4864,
+      "learning_rate": 1.7346778279993413e-05,
+      "loss": 0.4539,
+      "step": 2432
+    },
+    {
+      "epoch": 0.4868,
+      "learning_rate": 1.7356243592037872e-05,
+      "loss": 0.267,
+      "step": 2434
+    },
+    {
+      "epoch": 0.4872,
+      "learning_rate": 1.736569456270903e-05,
+      "loss": 0.5412,
+      "step": 2436
+    },
+    {
+      "epoch": 0.4876,
+      "learning_rate": 1.7375131173581737e-05,
+      "loss": 0.8155,
+      "step": 2438
+    },
+    {
+      "epoch": 0.488,
+      "learning_rate": 1.7384553406258836e-05,
+      "loss": 0.2868,
+      "step": 2440
+    },
+    {
+      "epoch": 0.4884,
+      "learning_rate": 1.73939612423712e-05,
+      "loss": 0.5432,
+      "step": 2442
+    },
+    {
+      "epoch": 0.4888,
+      "learning_rate": 1.740335466357778e-05,
+      "loss": 0.1585,
+      "step": 2444
+    },
+    {
+      "epoch": 0.4892,
+      "learning_rate": 1.7412733651565607e-05,
+      "loss": 0.4756,
+      "step": 2446
+    },
+    {
+      "epoch": 0.4896,
+      "learning_rate": 1.7422098188049888e-05,
+      "loss": 0.3802,
+      "step": 2448
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.7431448254773936e-05,
+      "loss": 0.3062,
+      "step": 2450
+    },
+    {
+      "epoch": 0.4904,
+      "learning_rate": 1.7440783833509373e-05,
+      "loss": 0.3937,
+      "step": 2452
+    },
+    {
+      "epoch": 0.4908,
+      "learning_rate": 1.7450104906055956e-05,
+      "loss": 0.4779,
+      "step": 2454
+    },
+    {
+      "epoch": 0.4912,
+      "learning_rate": 1.7459411454241816e-05,
+      "loss": 0.5218,
+      "step": 2456
+    },
+    {
+      "epoch": 0.4916,
+      "learning_rate": 1.746870345992336e-05,
+      "loss": 0.1275,
+      "step": 2458
+    },
+    {
+      "epoch": 0.492,
+      "learning_rate": 1.747798090498531e-05,
+      "loss": 0.2544,
+      "step": 2460
+    },
+    {
+      "epoch": 0.4924,
+      "learning_rate": 1.7487243771340865e-05,
+      "loss": 0.3223,
+      "step": 2462
+    },
+    {
+      "epoch": 0.4928,
+      "learning_rate": 1.749649204093154e-05,
+      "loss": 0.4431,
+      "step": 2464
+    },
+    {
+      "epoch": 0.4932,
+      "learning_rate": 1.750572569572741e-05,
+      "loss": 0.4967,
+      "step": 2466
+    },
+    {
+      "epoch": 0.4936,
+      "learning_rate": 1.7514944717726962e-05,
+      "loss": 0.1758,
+      "step": 2468
+    },
+    {
+      "epoch": 0.494,
+      "learning_rate": 1.7524149088957244e-05,
+      "loss": 0.5168,
+      "step": 2470
+    },
+    {
+      "epoch": 0.4944,
+      "learning_rate": 1.753333879147387e-05,
+      "loss": 0.344,
+      "step": 2472
+    },
+    {
+      "epoch": 0.4948,
+      "learning_rate": 1.7542513807361037e-05,
+      "loss": 0.2956,
+      "step": 2474
+    },
+    {
+      "epoch": 0.4952,
+      "learning_rate": 1.755167411873159e-05,
+      "loss": 0.3368,
+      "step": 2476
+    },
+    {
+      "epoch": 0.4956,
+      "learning_rate": 1.7560819707727027e-05,
+      "loss": 0.1103,
+      "step": 2478
+    },
+    {
+      "epoch": 0.496,
+      "learning_rate": 1.7569950556517563e-05,
+      "loss": 0.4512,
+      "step": 2480
+    },
+    {
+      "epoch": 0.4964,
+      "learning_rate": 1.757906664730213e-05,
+      "loss": 0.1489,
+      "step": 2482
+    },
+    {
+      "epoch": 0.4968,
+      "learning_rate": 1.758816796230845e-05,
+      "loss": 0.3424,
+      "step": 2484
+    },
+    {
+      "epoch": 0.4972,
+      "learning_rate": 1.759725448379304e-05,
+      "loss": 0.2624,
+      "step": 2486
+    },
+    {
+      "epoch": 0.4976,
+      "learning_rate": 1.7606326194041278e-05,
+      "loss": 0.1538,
+      "step": 2488
+    },
+    {
+      "epoch": 0.498,
+      "learning_rate": 1.7615383075367363e-05,
+      "loss": 0.2462,
+      "step": 2490
+    },
+    {
+      "epoch": 0.4984,
+      "learning_rate": 1.762442511011447e-05,
+      "loss": 0.5561,
+      "step": 2492
+    },
+    {
+      "epoch": 0.4988,
+      "learning_rate": 1.763345228065469e-05,
+      "loss": 0.1346,
+      "step": 2494
+    },
+    {
+      "epoch": 0.4992,
+      "learning_rate": 1.7642464569389083e-05,
+      "loss": 0.3682,
+      "step": 2496
+    },
+    {
+      "epoch": 0.4996,
+      "learning_rate": 1.7651461958747745e-05,
+      "loss": 0.1838,
+      "step": 2498
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.766044443118977e-05,
+      "loss": 0.3603,
+      "step": 2500
+    },
+    {
+      "epoch": 0.5004,
+      "learning_rate": 1.766941196920342e-05,
+      "loss": 0.3738,
+      "step": 2502
+    },
+    {
+      "epoch": 0.5008,
+      "learning_rate": 1.767836455530598e-05,
+      "loss": 0.2771,
+      "step": 2504
+    },
+    {
+      "epoch": 0.5012,
+      "learning_rate": 1.7687302172043933e-05,
+      "loss": 0.4026,
+      "step": 2506
+    },
+    {
+      "epoch": 0.5016,
+      "learning_rate": 1.7696224801992947e-05,
+      "loss": 0.0629,
+      "step": 2508
+    },
+    {
+      "epoch": 0.502,
+      "learning_rate": 1.7705132427757885e-05,
+      "loss": 0.2518,
+      "step": 2510
+    },
+    {
+      "epoch": 0.5024,
+      "learning_rate": 1.77140250319729e-05,
+      "loss": 0.2931,
+      "step": 2512
+    },
+    {
+      "epoch": 0.5028,
+      "learning_rate": 1.7722902597301385e-05,
+      "loss": 0.4337,
+      "step": 2514
+    },
+    {
+      "epoch": 0.5032,
+      "learning_rate": 1.7731765106436073e-05,
+      "loss": 0.4581,
+      "step": 2516
+    },
+    {
+      "epoch": 0.5036,
+      "learning_rate": 1.774061254209905e-05,
+      "loss": 0.5083,
+      "step": 2518
+    },
+    {
+      "epoch": 0.504,
+      "learning_rate": 1.7749444887041793e-05,
+      "loss": 0.3948,
+      "step": 2520
+    },
+    {
+      "epoch": 0.5044,
+      "learning_rate": 1.7758262124045192e-05,
+      "loss": 0.2454,
+      "step": 2522
+    },
+    {
+      "epoch": 0.5048,
+      "learning_rate": 1.776706423591959e-05,
+      "loss": 0.331,
+      "step": 2524
+    },
+    {
+      "epoch": 0.5052,
+      "learning_rate": 1.7775851205504816e-05,
+      "loss": 0.2174,
+      "step": 2526
+    },
+    {
+      "epoch": 0.5056,
+      "learning_rate": 1.778462301567023e-05,
+      "loss": 0.1906,
+      "step": 2528
+    },
+    {
+      "epoch": 0.506,
+      "learning_rate": 1.7793379649314736e-05,
+      "loss": 0.2221,
+      "step": 2530
+    },
+    {
+      "epoch": 0.5064,
+      "learning_rate": 1.7802121089366832e-05,
+      "loss": 0.3887,
+      "step": 2532
+    },
+    {
+      "epoch": 0.5068,
+      "learning_rate": 1.7810847318784635e-05,
+      "loss": 0.2998,
+      "step": 2534
+    },
+    {
+      "epoch": 0.5072,
+      "learning_rate": 1.7819558320555895e-05,
+      "loss": 0.3406,
+      "step": 2536
+    },
+    {
+      "epoch": 0.5076,
+      "learning_rate": 1.7828254077698103e-05,
+      "loss": 1.002,
+      "step": 2538
+    },
+    {
+      "epoch": 0.508,
+      "learning_rate": 1.7836934573258392e-05,
+      "loss": 0.1692,
+      "step": 2540
+    },
+    {
+      "epoch": 0.5084,
+      "learning_rate": 1.7845599790313735e-05,
+      "loss": 1.3339,
+      "step": 2542
+    },
+    {
+      "epoch": 0.5088,
+      "learning_rate": 1.785424971197082e-05,
+      "loss": 0.3784,
+      "step": 2544
+    },
+    {
+      "epoch": 0.5092,
+      "learning_rate": 1.786288432136618e-05,
+      "loss": 0.1756,
+      "step": 2546
+    },
+    {
+      "epoch": 0.5096,
+      "learning_rate": 1.7871503601666233e-05,
+      "loss": 0.4221,
+      "step": 2548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.788010753606722e-05,
+      "loss": 0.2934,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5104,
+      "learning_rate": 1.7888696107795343e-05,
+      "loss": 0.2025,
+      "step": 2552
+    },
+    {
+      "epoch": 0.5108,
+      "learning_rate": 1.7897269300106735e-05,
+      "loss": 0.8862,
+      "step": 2554
+    },
+    {
+      "epoch": 0.5112,
+      "learning_rate": 1.790582709628753e-05,
+      "loss": 0.56,
+      "step": 2556
+    },
+    {
+      "epoch": 0.5116,
+      "learning_rate": 1.7914369479653854e-05,
+      "loss": 0.3542,
+      "step": 2558
+    },
+    {
+      "epoch": 0.512,
+      "learning_rate": 1.7922896433551903e-05,
+      "loss": 0.271,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5124,
+      "learning_rate": 1.7931407941357945e-05,
+      "loss": 0.2647,
+      "step": 2562
+    },
+    {
+      "epoch": 0.5128,
+      "learning_rate": 1.793990398647835e-05,
+      "loss": 0.4603,
+      "step": 2564
+    },
+    {
+      "epoch": 0.5132,
+      "learning_rate": 1.7948384552349655e-05,
+      "loss": 0.3959,
+      "step": 2566
+    },
+    {
+      "epoch": 0.5136,
+      "learning_rate": 1.795684962243855e-05,
+      "loss": 0.9489,
+      "step": 2568
+    },
+    {
+      "epoch": 0.514,
+      "learning_rate": 1.796529918024196e-05,
+      "loss": 0.6159,
+      "step": 2570
+    },
+    {
+      "epoch": 0.5144,
+      "learning_rate": 1.7973733209287032e-05,
+      "loss": 0.43,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5148,
+      "learning_rate": 1.798215169313121e-05,
+      "loss": 0.3463,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5152,
+      "learning_rate": 1.7990554615362193e-05,
+      "loss": 0.301,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5156,
+      "learning_rate": 1.79989419595981e-05,
+      "loss": 0.1485,
+      "step": 2578
+    },
+    {
+      "epoch": 0.516,
+      "learning_rate": 1.800731370948734e-05,
+      "loss": 0.2955,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5164,
+      "learning_rate": 1.8015669848708757e-05,
+      "loss": 0.3198,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5168,
+      "learning_rate": 1.802401036097167e-05,
+      "loss": 0.3733,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5172,
+      "learning_rate": 1.803233523001577e-05,
+      "loss": 0.2532,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5176,
+      "learning_rate": 1.804064443961135e-05,
+      "loss": 0.6135,
+      "step": 2588
+    },
+    {
+      "epoch": 0.518,
+      "learning_rate": 1.804893797355914e-05,
+      "loss": 0.1644,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5184,
+      "learning_rate": 1.8057215815690494e-05,
+      "loss": 0.2538,
+      "step": 2592
+    },
+    {
+      "epoch": 0.5188,
+      "learning_rate": 1.8065477949867327e-05,
+      "loss": 0.4518,
+      "step": 2594
+    },
+    {
+      "epoch": 0.5192,
+      "learning_rate": 1.8073724359982184e-05,
+      "loss": 0.4101,
+      "step": 2596
+    },
+    {
+      "epoch": 0.5196,
+      "learning_rate": 1.808195502995827e-05,
+      "loss": 0.2644,
+      "step": 2598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.809016994374947e-05,
+      "loss": 0.3485,
+      "step": 2600
+    },
+    {
+      "epoch": 0.5204,
+      "learning_rate": 1.8098369085340397e-05,
+      "loss": 0.3839,
+      "step": 2602
+    },
+    {
+      "epoch": 0.5208,
+      "learning_rate": 1.81065524387464e-05,
+      "loss": 0.3478,
+      "step": 2604
+    },
+    {
+      "epoch": 0.5212,
+      "learning_rate": 1.8114719988013606e-05,
+      "loss": 0.2145,
+      "step": 2606
+    },
+    {
+      "epoch": 0.5216,
+      "learning_rate": 1.8122871717218968e-05,
+      "loss": 0.4506,
+      "step": 2608
+    },
+    {
+      "epoch": 0.522,
+      "learning_rate": 1.813100761047028e-05,
+      "loss": 0.295,
+      "step": 2610
+    },
+    {
+      "epoch": 0.5224,
+      "learning_rate": 1.8139127651906176e-05,
+      "loss": 0.5042,
+      "step": 2612
+    },
+    {
+      "epoch": 0.5228,
+      "learning_rate": 1.8147231825696258e-05,
+      "loss": 0.4108,
+      "step": 2614
+    },
+    {
+      "epoch": 0.5232,
+      "learning_rate": 1.8155320116040976e-05,
+      "loss": 0.7792,
+      "step": 2616
+    },
+    {
+      "epoch": 0.5236,
+      "learning_rate": 1.8163392507171834e-05,
+      "loss": 0.3315,
+      "step": 2618
+    },
+    {
+      "epoch": 0.524,
+      "learning_rate": 1.817144898335129e-05,
+      "loss": 0.5713,
+      "step": 2620
+    },
+    {
+      "epoch": 0.5244,
+      "learning_rate": 1.8179489528872797e-05,
+      "loss": 0.3929,
+      "step": 2622
+    },
+    {
+      "epoch": 0.5248,
+      "learning_rate": 1.818751412806095e-05,
+      "loss": 0.5568,
+      "step": 2624
+    },
+    {
+      "epoch": 0.5252,
+      "learning_rate": 1.819552276527134e-05,
+      "loss": 0.4067,
+      "step": 2626
+    },
+    {
+      "epoch": 0.5256,
+      "learning_rate": 1.8203515424890738e-05,
+      "loss": 0.3545,
+      "step": 2628
+    },
+    {
+      "epoch": 0.526,
+      "learning_rate": 1.821149209133704e-05,
+      "loss": 0.202,
+      "step": 2630
+    },
+    {
+      "epoch": 0.5264,
+      "learning_rate": 1.8219452749059322e-05,
+      "loss": 0.413,
+      "step": 2632
+    },
+    {
+      "epoch": 0.5268,
+      "learning_rate": 1.82273973825379e-05,
+      "loss": 0.1691,
+      "step": 2634
+    },
+    {
+      "epoch": 0.5272,
+      "learning_rate": 1.8235325976284276e-05,
+      "loss": 0.5171,
+      "step": 2636
+    },
+    {
+      "epoch": 0.5276,
+      "learning_rate": 1.8243238514841258e-05,
+      "loss": 0.3644,
+      "step": 2638
+    },
+    {
+      "epoch": 0.528,
+      "learning_rate": 1.8251134982782952e-05,
+      "loss": 0.5052,
+      "step": 2640
+    },
+    {
+      "epoch": 0.5284,
+      "learning_rate": 1.8259015364714786e-05,
+      "loss": 0.295,
+      "step": 2642
+    },
+    {
+      "epoch": 0.5288,
+      "learning_rate": 1.826687964527355e-05,
+      "loss": 0.167,
+      "step": 2644
+    },
+    {
+      "epoch": 0.5292,
+      "learning_rate": 1.8274727809127437e-05,
+      "loss": 0.1897,
+      "step": 2646
+    },
+    {
+      "epoch": 0.5296,
+      "learning_rate": 1.828255984097604e-05,
+      "loss": 0.8242,
+      "step": 2648
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8290375725550413e-05,
+      "loss": 0.4699,
+      "step": 2650
+    },
+    {
+      "epoch": 0.5304,
+      "learning_rate": 1.8298175447613093e-05,
+      "loss": 0.5471,
+      "step": 2652
+    },
+    {
+      "epoch": 0.5308,
+      "learning_rate": 1.8305958991958125e-05,
+      "loss": 0.5586,
+      "step": 2654
+    },
+    {
+      "epoch": 0.5312,
+      "learning_rate": 1.8313726343411092e-05,
+      "loss": 0.3535,
+      "step": 2656
+    },
+    {
+      "epoch": 0.5316,
+      "learning_rate": 1.832147748682912e-05,
+      "loss": 0.3339,
+      "step": 2658
+    },
+    {
+      "epoch": 0.532,
+      "learning_rate": 1.8329212407101e-05,
+      "loss": 0.3805,
+      "step": 2660
+    },
+    {
+      "epoch": 0.5324,
+      "learning_rate": 1.8336931089147065e-05,
+      "loss": 0.614,
+      "step": 2662
+    },
+    {
+      "epoch": 0.5328,
+      "learning_rate": 1.8344633517919394e-05,
+      "loss": 0.3325,
+      "step": 2664
+    },
+    {
+      "epoch": 0.5332,
+      "learning_rate": 1.8352319678401677e-05,
+      "loss": 0.0367,
+      "step": 2666
+    },
+    {
+      "epoch": 0.5336,
+      "learning_rate": 1.8359989555609344e-05,
+      "loss": 0.2787,
+      "step": 2668
+    },
+    {
+      "epoch": 0.534,
+      "learning_rate": 1.836764313458962e-05,
+      "loss": 1.5264,
+      "step": 2670
+    },
+    {
+      "epoch": 0.5344,
+      "learning_rate": 1.8375280400421407e-05,
+      "loss": 0.1218,
+      "step": 2672
+    },
+    {
+      "epoch": 0.5348,
+      "learning_rate": 1.8382901338215515e-05,
+      "loss": 0.5839,
+      "step": 2674
+    },
+    {
+      "epoch": 0.5352,
+      "learning_rate": 1.8390505933114503e-05,
+      "loss": 0.5254,
+      "step": 2676
+    },
+    {
+      "epoch": 0.5356,
+      "learning_rate": 1.839809417029283e-05,
+      "loss": 0.2326,
+      "step": 2678
+    },
+    {
+      "epoch": 0.536,
+      "learning_rate": 1.8405666034956842e-05,
+      "loss": 0.2566,
+      "step": 2680
+    },
+    {
+      "epoch": 0.5364,
+      "learning_rate": 1.8413221512344805e-05,
+      "loss": 1.2278,
+      "step": 2682
+    },
+    {
+      "epoch": 0.5368,
+      "learning_rate": 1.842076058772692e-05,
+      "loss": 0.643,
+      "step": 2684
+    },
+    {
+      "epoch": 0.5372,
+      "learning_rate": 1.8428283246405386e-05,
+      "loss": 1.0574,
+      "step": 2686
+    },
+    {
+      "epoch": 0.5376,
+      "learning_rate": 1.8435789473714384e-05,
+      "loss": 0.5391,
+      "step": 2688
+    },
+    {
+      "epoch": 0.538,
+      "learning_rate": 1.844327925502015e-05,
+      "loss": 0.3007,
+      "step": 2690
+    },
+    {
+      "epoch": 0.5384,
+      "learning_rate": 1.8450752575720964e-05,
+      "loss": 0.2207,
+      "step": 2692
+    },
+    {
+      "epoch": 0.5388,
+      "learning_rate": 1.8458209421247205e-05,
+      "loss": 0.2711,
+      "step": 2694
+    },
+    {
+      "epoch": 0.5392,
+      "learning_rate": 1.8465649777061384e-05,
+      "loss": 0.2487,
+      "step": 2696
+    },
+    {
+      "epoch": 0.5396,
+      "learning_rate": 1.8473073628658116e-05,
+      "loss": 0.469,
+      "step": 2698
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8480480961564266e-05,
+      "loss": 0.4119,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5404,
+      "learning_rate": 1.848787176133881e-05,
+      "loss": 0.3165,
+      "step": 2702
+    },
+    {
+      "epoch": 0.5408,
+      "learning_rate": 1.8495246013573047e-05,
+      "loss": 0.6014,
+      "step": 2704
+    },
+    {
+      "epoch": 0.5412,
+      "learning_rate": 1.850260370389049e-05,
+      "loss": 0.3239,
+      "step": 2706
+    },
+    {
+      "epoch": 0.5416,
+      "learning_rate": 1.850994481794691e-05,
+      "loss": 0.2028,
+      "step": 2708
+    },
+    {
+      "epoch": 0.542,
+      "learning_rate": 1.851726934143048e-05,
+      "loss": 0.5771,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5424,
+      "learning_rate": 1.8524577260061628e-05,
+      "loss": 0.8184,
+      "step": 2712
+    },
+    {
+      "epoch": 0.5428,
+      "learning_rate": 1.8531868559593205e-05,
+      "loss": 0.2131,
+      "step": 2714
+    },
+    {
+      "epoch": 0.5432,
+      "learning_rate": 1.8539143225810453e-05,
+      "loss": 0.2988,
+      "step": 2716
+    },
+    {
+      "epoch": 0.5436,
+      "learning_rate": 1.8546401244531028e-05,
+      "loss": 0.217,
+      "step": 2718
+    },
+    {
+      "epoch": 0.544,
+      "learning_rate": 1.8553642601605066e-05,
+      "loss": 0.4787,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5444,
+      "learning_rate": 1.856086728291516e-05,
+      "loss": 0.4268,
+      "step": 2722
+    },
+    {
+      "epoch": 0.5448,
+      "learning_rate": 1.856807527437643e-05,
+      "loss": 0.6164,
+      "step": 2724
+    },
+    {
+      "epoch": 0.5452,
+      "learning_rate": 1.857526656193652e-05,
+      "loss": 0.5475,
+      "step": 2726
+    },
+    {
+      "epoch": 0.5456,
+      "learning_rate": 1.8582441131575658e-05,
+      "loss": 0.2327,
+      "step": 2728
+    },
+    {
+      "epoch": 0.546,
+      "learning_rate": 1.8589598969306643e-05,
+      "loss": 0.1463,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5464,
+      "learning_rate": 1.859674006117491e-05,
+      "loss": 0.6091,
+      "step": 2732
+    },
+    {
+      "epoch": 0.5468,
+      "learning_rate": 1.860386439325853e-05,
+      "loss": 0.4772,
+      "step": 2734
+    },
+    {
+      "epoch": 0.5472,
+      "learning_rate": 1.8610971951668268e-05,
+      "loss": 0.2985,
+      "step": 2736
+    },
+    {
+      "epoch": 0.5476,
+      "learning_rate": 1.8618062722547544e-05,
+      "loss": 0.6341,
+      "step": 2738
+    },
+    {
+      "epoch": 0.548,
+      "learning_rate": 1.862513669207257e-05,
+      "loss": 0.3775,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5484,
+      "learning_rate": 1.8632193846452274e-05,
+      "loss": 0.657,
+      "step": 2742
+    },
+    {
+      "epoch": 0.5488,
+      "learning_rate": 1.8639234171928348e-05,
+      "loss": 0.784,
+      "step": 2744
+    },
+    {
+      "epoch": 0.5492,
+      "learning_rate": 1.8646257654775354e-05,
+      "loss": 0.2002,
+      "step": 2746
+    },
+    {
+      "epoch": 0.5496,
+      "learning_rate": 1.8653264281300612e-05,
+      "loss": 0.4095,
+      "step": 2748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.204,
+      "step": 2750
+    },
+    {
+      "epoch": 0.5504,
+      "learning_rate": 1.8667226910779767e-05,
+      "loss": 0.2391,
+      "step": 2752
+    },
+    {
+      "epoch": 0.5508,
+      "learning_rate": 1.8674182886512776e-05,
+      "loss": 0.6073,
+      "step": 2754
+    },
+    {
+      "epoch": 0.5512,
+      "learning_rate": 1.8681121951482393e-05,
+      "loss": 0.4849,
+      "step": 2756
+    },
+    {
+      "epoch": 0.5516,
+      "learning_rate": 1.8688044092160554e-05,
+      "loss": 1.1562,
+      "step": 2758
+    },
+    {
+      "epoch": 0.552,
+      "learning_rate": 1.869494929505219e-05,
+      "loss": 0.4646,
+      "step": 2760
+    },
+    {
+      "epoch": 0.5524,
+      "learning_rate": 1.8701837546695256e-05,
+      "loss": 0.1796,
+      "step": 2762
+    },
+    {
+      "epoch": 0.5528,
+      "learning_rate": 1.870870883366075e-05,
+      "loss": 0.5265,
+      "step": 2764
+    },
+    {
+      "epoch": 0.5532,
+      "learning_rate": 1.871556314255275e-05,
+      "loss": 0.3542,
+      "step": 2766
+    },
+    {
+      "epoch": 0.5536,
+      "learning_rate": 1.8722400460008434e-05,
+      "loss": 0.5263,
+      "step": 2768
+    },
+    {
+      "epoch": 0.554,
+      "learning_rate": 1.8729220772698093e-05,
+      "loss": 0.1692,
+      "step": 2770
+    },
+    {
+      "epoch": 0.5544,
+      "learning_rate": 1.8736024067325195e-05,
+      "loss": 0.5318,
+      "step": 2772
+    },
+    {
+      "epoch": 0.5548,
+      "learning_rate": 1.8742810330626335e-05,
+      "loss": 0.3523,
+      "step": 2774
+    },
+    {
+      "epoch": 0.5552,
+      "learning_rate": 1.8749579549371373e-05,
+      "loss": 0.6617,
+      "step": 2776
+    },
+    {
+      "epoch": 0.5556,
+      "learning_rate": 1.8756331710363368e-05,
+      "loss": 0.2415,
+      "step": 2778
+    },
+    {
+      "epoch": 0.556,
+      "learning_rate": 1.876306680043863e-05,
+      "loss": 0.2456,
+      "step": 2780
+    },
+    {
+      "epoch": 0.5564,
+      "learning_rate": 1.876978480646677e-05,
+      "loss": 0.3899,
+      "step": 2782
+    },
+    {
+      "epoch": 0.5568,
+      "learning_rate": 1.8776485715350665e-05,
+      "loss": 0.4971,
+      "step": 2784
+    },
+    {
+      "epoch": 0.5572,
+      "learning_rate": 1.878316951402658e-05,
+      "loss": 0.2779,
+      "step": 2786
+    },
+    {
+      "epoch": 0.5576,
+      "learning_rate": 1.878983618946409e-05,
+      "loss": 0.171,
+      "step": 2788
+    },
+    {
+      "epoch": 0.558,
+      "learning_rate": 1.879648572866617e-05,
+      "loss": 0.4398,
+      "step": 2790
+    },
+    {
+      "epoch": 0.5584,
+      "learning_rate": 1.8803118118669203e-05,
+      "loss": 0.7344,
+      "step": 2792
+    },
+    {
+      "epoch": 0.5588,
+      "learning_rate": 1.8809733346543006e-05,
+      "loss": 0.692,
+      "step": 2794
+    },
+    {
+      "epoch": 0.5592,
+      "learning_rate": 1.881633139939087e-05,
+      "loss": 0.1635,
+      "step": 2796
+    },
+    {
+      "epoch": 0.5596,
+      "learning_rate": 1.8822912264349532e-05,
+      "loss": 0.4421,
+      "step": 2798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.882947592858927e-05,
+      "loss": 0.2291,
+      "step": 2800
+    },
+    {
+      "epoch": 0.5604,
+      "learning_rate": 1.8836022379313877e-05,
+      "loss": 0.1038,
+      "step": 2802
+    },
+    {
+      "epoch": 0.5608,
+      "learning_rate": 1.884255160376072e-05,
+      "loss": 0.4008,
+      "step": 2804
+    },
+    {
+      "epoch": 0.5612,
+      "learning_rate": 1.8849063589200744e-05,
+      "loss": 0.454,
+      "step": 2806
+    },
+    {
+      "epoch": 0.5616,
+      "learning_rate": 1.885555832293849e-05,
+      "loss": 0.1002,
+      "step": 2808
+    },
+    {
+      "epoch": 0.562,
+      "learning_rate": 1.8862035792312145e-05,
+      "loss": 0.3014,
+      "step": 2810
+    },
+    {
+      "epoch": 0.5624,
+      "learning_rate": 1.886849598469356e-05,
+      "loss": 0.237,
+      "step": 2812
+    },
+    {
+      "epoch": 0.5628,
+      "learning_rate": 1.8874938887488246e-05,
+      "loss": 0.2276,
+      "step": 2814
+    },
+    {
+      "epoch": 0.5632,
+      "learning_rate": 1.888136448813544e-05,
+      "loss": 0.3647,
+      "step": 2816
+    },
+    {
+      "epoch": 0.5636,
+      "learning_rate": 1.888777277410812e-05,
+      "loss": 0.4398,
+      "step": 2818
+    },
+    {
+      "epoch": 0.564,
+      "learning_rate": 1.8894163732912972e-05,
+      "loss": 0.2421,
+      "step": 2820
+    },
+    {
+      "epoch": 0.5644,
+      "learning_rate": 1.890053735209053e-05,
+      "loss": 0.0977,
+      "step": 2822
+    },
+    {
+      "epoch": 0.5648,
+      "learning_rate": 1.890689361921506e-05,
+      "loss": 0.509,
+      "step": 2824
+    },
+    {
+      "epoch": 0.5652,
+      "learning_rate": 1.8913232521894737e-05,
+      "loss": 0.7079,
+      "step": 2826
+    },
+    {
+      "epoch": 0.5656,
+      "learning_rate": 1.891955404777151e-05,
+      "loss": 0.3261,
+      "step": 2828
+    },
+    {
+      "epoch": 0.566,
+      "learning_rate": 1.8925858184521248e-05,
+      "loss": 0.28,
+      "step": 2830
+    },
+    {
+      "epoch": 0.5664,
+      "learning_rate": 1.893214491985374e-05,
+      "loss": 0.7739,
+      "step": 2832
+    },
+    {
+      "epoch": 0.5668,
+      "learning_rate": 1.8938414241512634e-05,
+      "loss": 0.5093,
+      "step": 2834
+    },
+    {
+      "epoch": 0.5672,
+      "learning_rate": 1.89446661372756e-05,
+      "loss": 0.4029,
+      "step": 2836
+    },
+    {
+      "epoch": 0.5676,
+      "learning_rate": 1.8950900594954226e-05,
+      "loss": 0.4449,
+      "step": 2838
+    },
+    {
+      "epoch": 0.568,
+      "learning_rate": 1.895711760239413e-05,
+      "loss": 0.3913,
+      "step": 2840
+    },
+    {
+      "epoch": 0.5684,
+      "learning_rate": 1.896331714747493e-05,
+      "loss": 0.3687,
+      "step": 2842
+    },
+    {
+      "epoch": 0.5688,
+      "learning_rate": 1.89694992181103e-05,
+      "loss": 0.4915,
+      "step": 2844
+    },
+    {
+      "epoch": 0.5692,
+      "learning_rate": 1.8975663802247975e-05,
+      "loss": 0.2876,
+      "step": 2846
+    },
+    {
+      "epoch": 0.5696,
+      "learning_rate": 1.8981810887869784e-05,
+      "loss": 0.0464,
+      "step": 2848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.898794046299167e-05,
+      "loss": 0.2611,
+      "step": 2850
+    },
+    {
+      "epoch": 0.5704,
+      "learning_rate": 1.8994052515663708e-05,
+      "loss": 0.5614,
+      "step": 2852
+    },
+    {
+      "epoch": 0.5708,
+      "learning_rate": 1.9000147033970144e-05,
+      "loss": 1.9282,
+      "step": 2854
+    },
+    {
+      "epoch": 0.5712,
+      "learning_rate": 1.90062240060294e-05,
+      "loss": 0.1015,
+      "step": 2856
+    },
+    {
+      "epoch": 0.5716,
+      "learning_rate": 1.901228341999412e-05,
+      "loss": 0.0913,
+      "step": 2858
+    },
+    {
+      "epoch": 0.572,
+      "learning_rate": 1.9018325264051136e-05,
+      "loss": 0.4343,
+      "step": 2860
+    },
+    {
+      "epoch": 0.5724,
+      "learning_rate": 1.9024349526421596e-05,
+      "loss": 0.4359,
+      "step": 2862
+    },
+    {
+      "epoch": 0.5728,
+      "learning_rate": 1.9030356195360868e-05,
+      "loss": 0.3295,
+      "step": 2864
+    },
+    {
+      "epoch": 0.5732,
+      "learning_rate": 1.903634525915866e-05,
+      "loss": 0.1641,
+      "step": 2866
+    },
+    {
+      "epoch": 0.5736,
+      "learning_rate": 1.904231670613899e-05,
+      "loss": 0.184,
+      "step": 2868
+    },
+    {
+      "epoch": 0.574,
+      "learning_rate": 1.904827052466019e-05,
+      "loss": 0.3483,
+      "step": 2870
+    },
+    {
+      "epoch": 0.5744,
+      "learning_rate": 1.905420670311502e-05,
+      "loss": 0.261,
+      "step": 2872
+    },
+    {
+      "epoch": 0.5748,
+      "learning_rate": 1.9060125229930572e-05,
+      "loss": 0.4621,
+      "step": 2874
+    },
+    {
+      "epoch": 0.5752,
+      "learning_rate": 1.906602609356838e-05,
+      "loss": 0.7085,
+      "step": 2876
+    },
+    {
+      "epoch": 0.5756,
+      "learning_rate": 1.907190928252441e-05,
+      "loss": 0.4198,
+      "step": 2878
+    },
+    {
+      "epoch": 0.576,
+      "learning_rate": 1.9077774785329078e-05,
+      "loss": 0.2698,
+      "step": 2880
+    },
+    {
+      "epoch": 0.5764,
+      "learning_rate": 1.908362259054731e-05,
+      "loss": 0.2312,
+      "step": 2882
+    },
+    {
+      "epoch": 0.5768,
+      "learning_rate": 1.9089452686778487e-05,
+      "loss": 0.3692,
+      "step": 2884
+    },
+    {
+      "epoch": 0.5772,
+      "learning_rate": 1.9095265062656542e-05,
+      "loss": 0.342,
+      "step": 2886
+    },
+    {
+      "epoch": 0.5776,
+      "learning_rate": 1.9101059706849957e-05,
+      "loss": 0.4308,
+      "step": 2888
+    },
+    {
+      "epoch": 0.578,
+      "learning_rate": 1.910683660806177e-05,
+      "loss": 0.4425,
+      "step": 2890
+    },
+    {
+      "epoch": 0.5784,
+      "learning_rate": 1.911259575502962e-05,
+      "loss": 0.3312,
+      "step": 2892
+    },
+    {
+      "epoch": 0.5788,
+      "learning_rate": 1.9118337136525754e-05,
+      "loss": 0.5693,
+      "step": 2894
+    },
+    {
+      "epoch": 0.5792,
+      "learning_rate": 1.912406074135706e-05,
+      "loss": 0.208,
+      "step": 2896
+    },
+    {
+      "epoch": 0.5796,
+      "learning_rate": 1.912976655836507e-05,
+      "loss": 0.401,
+      "step": 2898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9135454576426006e-05,
+      "loss": 0.1895,
+      "step": 2900
+    },
+    {
+      "epoch": 0.5804,
+      "learning_rate": 1.9141124784450786e-05,
+      "loss": 0.3374,
+      "step": 2902
+    },
+    {
+      "epoch": 0.5808,
+      "learning_rate": 1.9146777171385053e-05,
+      "loss": 0.2485,
+      "step": 2904
+    },
+    {
+      "epoch": 0.5812,
+      "learning_rate": 1.9152411726209172e-05,
+      "loss": 0.6623,
+      "step": 2906
+    },
+    {
+      "epoch": 0.5816,
+      "learning_rate": 1.9158028437938316e-05,
+      "loss": 0.9272,
+      "step": 2908
+    },
+    {
+      "epoch": 0.582,
+      "learning_rate": 1.916362729562239e-05,
+      "loss": 0.4362,
+      "step": 2910
+    },
+    {
+      "epoch": 0.5824,
+      "learning_rate": 1.9169208288346168e-05,
+      "loss": 0.2934,
+      "step": 2912
+    },
+    {
+      "epoch": 0.5828,
+      "learning_rate": 1.9174771405229187e-05,
+      "loss": 0.233,
+      "step": 2914
+    },
+    {
+      "epoch": 0.5832,
+      "learning_rate": 1.9180316635425876e-05,
+      "loss": 0.4501,
+      "step": 2916
+    },
+    {
+      "epoch": 0.5836,
+      "learning_rate": 1.9185843968125543e-05,
+      "loss": 0.2608,
+      "step": 2918
+    },
+    {
+      "epoch": 0.584,
+      "learning_rate": 1.9191353392552346e-05,
+      "loss": 0.0836,
+      "step": 2920
+    },
+    {
+      "epoch": 0.5844,
+      "learning_rate": 1.919684489796539e-05,
+      "loss": 0.2386,
+      "step": 2922
+    },
+    {
+      "epoch": 0.5848,
+      "learning_rate": 1.9202318473658703e-05,
+      "loss": 0.3728,
+      "step": 2924
+    },
+    {
+      "epoch": 0.5852,
+      "learning_rate": 1.9207774108961273e-05,
+      "loss": 0.2449,
+      "step": 2926
+    },
+    {
+      "epoch": 0.5856,
+      "learning_rate": 1.9213211793237052e-05,
+      "loss": 0.5553,
+      "step": 2928
+    },
+    {
+      "epoch": 0.586,
+      "learning_rate": 1.9218631515885004e-05,
+      "loss": 0.3882,
+      "step": 2930
+    },
+    {
+      "epoch": 0.5864,
+      "learning_rate": 1.92240332663391e-05,
+      "loss": 0.353,
+      "step": 2932
+    },
+    {
+      "epoch": 0.5868,
+      "learning_rate": 1.922941703406835e-05,
+      "loss": 0.5851,
+      "step": 2934
+    },
+    {
+      "epoch": 0.5872,
+      "learning_rate": 1.923478280857682e-05,
+      "loss": 0.4484,
+      "step": 2936
+    },
+    {
+      "epoch": 0.5876,
+      "learning_rate": 1.9240130579403663e-05,
+      "loss": 0.4564,
+      "step": 2938
+    },
+    {
+      "epoch": 0.588,
+      "learning_rate": 1.924546033612313e-05,
+      "loss": 0.3117,
+      "step": 2940
+    },
+    {
+      "epoch": 0.5884,
+      "learning_rate": 1.9250772068344577e-05,
+      "loss": 0.5023,
+      "step": 2942
+    },
+    {
+      "epoch": 0.5888,
+      "learning_rate": 1.9256065765712524e-05,
+      "loss": 0.147,
+      "step": 2944
+    },
+    {
+      "epoch": 0.5892,
+      "learning_rate": 1.9261341417906615e-05,
+      "loss": 0.1917,
+      "step": 2946
+    },
+    {
+      "epoch": 0.5896,
+      "learning_rate": 1.9266599014641724e-05,
+      "loss": 0.45,
+      "step": 2948
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9271838545667876e-05,
+      "loss": 0.1353,
+      "step": 2950
+    },
+    {
+      "epoch": 0.5904,
+      "learning_rate": 1.927706000077034e-05,
+      "loss": 0.4902,
+      "step": 2952
+    },
+    {
+      "epoch": 0.5908,
+      "learning_rate": 1.9282263369769633e-05,
+      "loss": 0.1986,
+      "step": 2954
+    },
+    {
+      "epoch": 0.5912,
+      "learning_rate": 1.9287448642521507e-05,
+      "loss": 0.3804,
+      "step": 2956
+    },
+    {
+      "epoch": 0.5916,
+      "learning_rate": 1.9292615808917024e-05,
+      "loss": 0.5828,
+      "step": 2958
+    },
+    {
+      "epoch": 0.592,
+      "learning_rate": 1.9297764858882516e-05,
+      "loss": 0.4453,
+      "step": 2960
+    },
+    {
+      "epoch": 0.5924,
+      "learning_rate": 1.9302895782379648e-05,
+      "loss": 0.5947,
+      "step": 2962
+    },
+    {
+      "epoch": 0.5928,
+      "learning_rate": 1.9308008569405424e-05,
+      "loss": 0.1968,
+      "step": 2964
+    },
+    {
+      "epoch": 0.5932,
+      "learning_rate": 1.9313103209992205e-05,
+      "loss": 0.6335,
+      "step": 2966
+    },
+    {
+      "epoch": 0.5936,
+      "learning_rate": 1.9318179694207722e-05,
+      "loss": 0.1334,
+      "step": 2968
+    },
+    {
+      "epoch": 0.594,
+      "learning_rate": 1.932323801215512e-05,
+      "loss": 0.2482,
+      "step": 2970
+    },
+    {
+      "epoch": 0.5944,
+      "learning_rate": 1.9328278153972943e-05,
+      "loss": 0.2898,
+      "step": 2972
+    },
+    {
+      "epoch": 0.5948,
+      "learning_rate": 1.933330010983518e-05,
+      "loss": 0.9807,
+      "step": 2974
+    },
+    {
+      "epoch": 0.5952,
+      "learning_rate": 1.9338303869951266e-05,
+      "loss": 0.3055,
+      "step": 2976
+    },
+    {
+      "epoch": 0.5956,
+      "learning_rate": 1.934328942456612e-05,
+      "loss": 0.299,
+      "step": 2978
+    },
+    {
+      "epoch": 0.596,
+      "learning_rate": 1.934825676396015e-05,
+      "loss": 1.6104,
+      "step": 2980
+    },
+    {
+      "epoch": 0.5964,
+      "learning_rate": 1.9353205878449257e-05,
+      "loss": 0.4087,
+      "step": 2982
+    },
+    {
+      "epoch": 0.5968,
+      "learning_rate": 1.935813675838491e-05,
+      "loss": 0.3544,
+      "step": 2984
+    },
+    {
+      "epoch": 0.5972,
+      "learning_rate": 1.9363049394154088e-05,
+      "loss": 0.3972,
+      "step": 2986
+    },
+    {
+      "epoch": 0.5976,
+      "learning_rate": 1.9367943776179375e-05,
+      "loss": 0.8196,
+      "step": 2988
+    },
+    {
+      "epoch": 0.598,
+      "learning_rate": 1.937281989491892e-05,
+      "loss": 0.2637,
+      "step": 2990
+    },
+    {
+      "epoch": 0.5984,
+      "learning_rate": 1.9377677740866457e-05,
+      "loss": 0.5511,
+      "step": 2992
+    },
+    {
+      "epoch": 0.5988,
+      "learning_rate": 1.9382517304551397e-05,
+      "loss": 0.2166,
+      "step": 2994
+    },
+    {
+      "epoch": 0.5992,
+      "learning_rate": 1.9387338576538743e-05,
+      "loss": 0.4095,
+      "step": 2996
+    },
+    {
+      "epoch": 0.5996,
+      "learning_rate": 1.9392141547429183e-05,
+      "loss": 0.469,
+      "step": 2998
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.275,
+      "step": 3000
+    },
+    {
+      "epoch": 0.6004,
+      "learning_rate": 1.9401692548500504e-05,
+      "loss": 0.3073,
+      "step": 3002
+    },
+    {
+      "epoch": 0.6008,
+      "learning_rate": 1.9406440560061214e-05,
+      "loss": 0.5558,
+      "step": 3004
+    },
+    {
+      "epoch": 0.6012,
+      "learning_rate": 1.9411170233284728e-05,
+      "loss": 0.3969,
+      "step": 3006
+    },
+    {
+      "epoch": 0.6016,
+      "learning_rate": 1.9415881558950302e-05,
+      "loss": 0.3593,
+      "step": 3008
+    },
+    {
+      "epoch": 0.602,
+      "learning_rate": 1.942057452787297e-05,
+      "loss": 0.4463,
+      "step": 3010
+    },
+    {
+      "epoch": 0.6024,
+      "learning_rate": 1.942524913090354e-05,
+      "loss": 0.2929,
+      "step": 3012
+    },
+    {
+      "epoch": 0.6028,
+      "learning_rate": 1.9429905358928645e-05,
+      "loss": 0.4766,
+      "step": 3014
+    },
+    {
+      "epoch": 0.6032,
+      "learning_rate": 1.9434543202870723e-05,
+      "loss": 0.0977,
+      "step": 3016
+    },
+    {
+      "epoch": 0.6036,
+      "learning_rate": 1.9439162653688063e-05,
+      "loss": 0.4795,
+      "step": 3018
+    },
+    {
+      "epoch": 0.604,
+      "learning_rate": 1.9443763702374815e-05,
+      "loss": 0.3406,
+      "step": 3020
+    },
+    {
+      "epoch": 0.6044,
+      "learning_rate": 1.944834633996098e-05,
+      "loss": 0.5806,
+      "step": 3022
+    },
+    {
+      "epoch": 0.6048,
+      "learning_rate": 1.9452910557512494e-05,
+      "loss": 0.1126,
+      "step": 3024
+    },
+    {
+      "epoch": 0.6052,
+      "learning_rate": 1.9457456346131172e-05,
+      "loss": 0.1443,
+      "step": 3026
+    },
+    {
+      "epoch": 0.6056,
+      "learning_rate": 1.9461983696954756e-05,
+      "loss": 0.1466,
+      "step": 3028
+    },
+    {
+      "epoch": 0.606,
+      "learning_rate": 1.9466492601156964e-05,
+      "loss": 0.3687,
+      "step": 3030
+    },
+    {
+      "epoch": 0.6064,
+      "learning_rate": 1.947098304994744e-05,
+      "loss": 0.7448,
+      "step": 3032
+    },
+    {
+      "epoch": 0.6068,
+      "learning_rate": 1.947545503457184e-05,
+      "loss": 0.6954,
+      "step": 3034
+    },
+    {
+      "epoch": 0.6072,
+      "learning_rate": 1.9479908546311787e-05,
+      "loss": 0.271,
+      "step": 3036
+    },
+    {
+      "epoch": 0.6076,
+      "learning_rate": 1.9484343576484935e-05,
+      "loss": 0.3586,
+      "step": 3038
+    },
+    {
+      "epoch": 0.608,
+      "learning_rate": 1.9488760116444966e-05,
+      "loss": 0.2513,
+      "step": 3040
+    },
+    {
+      "epoch": 0.6084,
+      "learning_rate": 1.949315815758161e-05,
+      "loss": 0.4403,
+      "step": 3042
+    },
+    {
+      "epoch": 0.6088,
+      "learning_rate": 1.949753769132067e-05,
+      "loss": 0.6838,
+      "step": 3044
+    },
+    {
+      "epoch": 0.6092,
+      "learning_rate": 1.9501898709124008e-05,
+      "loss": 0.2401,
+      "step": 3046
+    },
+    {
+      "epoch": 0.6096,
+      "learning_rate": 1.95062412024896e-05,
+      "loss": 0.2543,
+      "step": 3048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9510565162951534e-05,
+      "loss": 0.9023,
+      "step": 3050
+    },
+    {
+      "epoch": 0.6104,
+      "learning_rate": 1.951487058208003e-05,
+      "loss": 0.4675,
+      "step": 3052
+    },
+    {
+      "epoch": 0.6108,
+      "learning_rate": 1.9519157451481453e-05,
+      "loss": 0.1379,
+      "step": 3054
+    },
+    {
+      "epoch": 0.6112,
+      "learning_rate": 1.952342576279833e-05,
+      "loss": 0.3731,
+      "step": 3056
+    },
+    {
+      "epoch": 0.6116,
+      "learning_rate": 1.9527675507709364e-05,
+      "loss": 0.1114,
+      "step": 3058
+    },
+    {
+      "epoch": 0.612,
+      "learning_rate": 1.953190667792947e-05,
+      "loss": 0.5675,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6124,
+      "learning_rate": 1.953611926520976e-05,
+      "loss": 0.3667,
+      "step": 3062
+    },
+    {
+      "epoch": 0.6128,
+      "learning_rate": 1.9540313261337578e-05,
+      "loss": 0.5543,
+      "step": 3064
+    },
+    {
+      "epoch": 0.6132,
+      "learning_rate": 1.9544488658136522e-05,
+      "loss": 0.1604,
+      "step": 3066
+    },
+    {
+      "epoch": 0.6136,
+      "learning_rate": 1.954864544746643e-05,
+      "loss": 0.4575,
+      "step": 3068
+    },
+    {
+      "epoch": 0.614,
+      "learning_rate": 1.955278362122344e-05,
+      "loss": 0.0978,
+      "step": 3070
+    },
+    {
+      "epoch": 0.6144,
+      "learning_rate": 1.955690317133996e-05,
+      "loss": 0.2581,
+      "step": 3072
+    },
+    {
+      "epoch": 0.6148,
+      "learning_rate": 1.9561004089784726e-05,
+      "loss": 0.5175,
+      "step": 3074
+    },
+    {
+      "epoch": 0.6152,
+      "learning_rate": 1.956508636856278e-05,
+      "loss": 0.2001,
+      "step": 3076
+    },
+    {
+      "epoch": 0.6156,
+      "learning_rate": 1.956914999971551e-05,
+      "loss": 0.2532,
+      "step": 3078
+    },
+    {
+      "epoch": 0.616,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 0.271,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6164,
+      "learning_rate": 1.9577221287492368e-05,
+      "loss": 0.4446,
+      "step": 3082
+    },
+    {
+      "epoch": 0.6168,
+      "learning_rate": 1.95812289283811e-05,
+      "loss": 0.6833,
+      "step": 3084
+    },
+    {
+      "epoch": 0.6172,
+      "learning_rate": 1.958521789017376e-05,
+      "loss": 0.0902,
+      "step": 3086
+    },
+    {
+      "epoch": 0.6176,
+      "learning_rate": 1.958918816509367e-05,
+      "loss": 0.4855,
+      "step": 3088
+    },
+    {
+      "epoch": 0.618,
+      "learning_rate": 1.9593139745400575e-05,
+      "loss": 0.2153,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6184,
+      "learning_rate": 1.9597072623390668e-05,
+      "loss": 0.4365,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6188,
+      "learning_rate": 1.9600986791396597e-05,
+      "loss": 0.6754,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6192,
+      "learning_rate": 1.9604882241787496e-05,
+      "loss": 0.2218,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6196,
+      "learning_rate": 1.9608758966968983e-05,
+      "loss": 0.3194,
+      "step": 3098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9612616959383187e-05,
+      "loss": 0.2613,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6204,
+      "learning_rate": 1.9616456211508752e-05,
+      "loss": 0.2173,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6208,
+      "learning_rate": 1.9620276715860856e-05,
+      "loss": 0.6392,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6212,
+      "learning_rate": 1.962407846499124e-05,
+      "loss": 0.5724,
+      "step": 3106
+    },
+    {
+      "epoch": 0.6216,
+      "learning_rate": 1.9627861451488187e-05,
+      "loss": 0.6815,
+      "step": 3108
+    },
+    {
+      "epoch": 0.622,
+      "learning_rate": 1.9631625667976584e-05,
+      "loss": 0.6591,
+      "step": 3110
+    },
+    {
+      "epoch": 0.6224,
+      "learning_rate": 1.963537110711789e-05,
+      "loss": 0.4591,
+      "step": 3112
+    },
+    {
+      "epoch": 0.6228,
+      "learning_rate": 1.9639097761610174e-05,
+      "loss": 0.6167,
+      "step": 3114
+    },
+    {
+      "epoch": 0.6232,
+      "learning_rate": 1.964280562418815e-05,
+      "loss": 0.2654,
+      "step": 3116
+    },
+    {
+      "epoch": 0.6236,
+      "learning_rate": 1.964649468762313e-05,
+      "loss": 0.2223,
+      "step": 3118
+    },
+    {
+      "epoch": 0.624,
+      "learning_rate": 1.9650164944723116e-05,
+      "loss": 0.5332,
+      "step": 3120
+    },
+    {
+      "epoch": 0.6244,
+      "learning_rate": 1.965381638833274e-05,
+      "loss": 0.3946,
+      "step": 3122
+    },
+    {
+      "epoch": 0.6248,
+      "learning_rate": 1.9657449011333328e-05,
+      "loss": 0.492,
+      "step": 3124
+    },
+    {
+      "epoch": 0.6252,
+      "learning_rate": 1.96610628066429e-05,
+      "loss": 0.2532,
+      "step": 3126
+    },
+    {
+      "epoch": 0.6256,
+      "learning_rate": 1.9664657767216176e-05,
+      "loss": 0.3132,
+      "step": 3128
+    },
+    {
+      "epoch": 0.626,
+      "learning_rate": 1.9668233886044594e-05,
+      "loss": 0.2219,
+      "step": 3130
+    },
+    {
+      "epoch": 0.6264,
+      "learning_rate": 1.967179115615633e-05,
+      "loss": 0.3411,
+      "step": 3132
+    },
+    {
+      "epoch": 0.6268,
+      "learning_rate": 1.96753295706163e-05,
+      "loss": 0.2371,
+      "step": 3134
+    },
+    {
+      "epoch": 0.6272,
+      "learning_rate": 1.967884912252619e-05,
+      "loss": 0.2038,
+      "step": 3136
+    },
+    {
+      "epoch": 0.6276,
+      "learning_rate": 1.9682349805024443e-05,
+      "loss": 0.3467,
+      "step": 3138
+    },
+    {
+      "epoch": 0.628,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 0.6589,
+      "step": 3140
+    },
+    {
+      "epoch": 0.6284,
+      "learning_rate": 1.9689294534523833e-05,
+      "loss": 0.3652,
+      "step": 3142
+    },
+    {
+      "epoch": 0.6288,
+      "learning_rate": 1.969273856798585e-05,
+      "loss": 0.6779,
+      "step": 3144
+    },
+    {
+      "epoch": 0.6292,
+      "learning_rate": 1.969616370495806e-05,
+      "loss": 0.6917,
+      "step": 3146
+    },
+    {
+      "epoch": 0.6296,
+      "learning_rate": 1.9699569938762972e-05,
+      "loss": 0.3718,
+      "step": 3148
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9702957262759964e-05,
+      "loss": 0.5987,
+      "step": 3150
+    },
+    {
+      "epoch": 0.6304,
+      "learning_rate": 1.9706325670345276e-05,
+      "loss": 0.8134,
+      "step": 3152
+    },
+    {
+      "epoch": 0.6308,
+      "learning_rate": 1.9709675154952013e-05,
+      "loss": 0.4146,
+      "step": 3154
+    },
+    {
+      "epoch": 0.6312,
+      "learning_rate": 1.9713005710050203e-05,
+      "loss": 0.4077,
+      "step": 3156
+    },
+    {
+      "epoch": 0.6316,
+      "learning_rate": 1.971631732914674e-05,
+      "loss": 0.1297,
+      "step": 3158
+    },
+    {
+      "epoch": 0.632,
+      "learning_rate": 1.9719610005785466e-05,
+      "loss": 0.2461,
+      "step": 3160
+    },
+    {
+      "epoch": 0.6324,
+      "learning_rate": 1.9722883733547128e-05,
+      "loss": 0.4181,
+      "step": 3162
+    },
+    {
+      "epoch": 0.6328,
+      "learning_rate": 1.9726138506049434e-05,
+      "loss": 0.4337,
+      "step": 3164
+    },
+    {
+      "epoch": 0.6332,
+      "learning_rate": 1.972937431694704e-05,
+      "loss": 0.3711,
+      "step": 3166
+    },
+    {
+      "epoch": 0.6336,
+      "learning_rate": 1.9732591159931564e-05,
+      "loss": 0.0643,
+      "step": 3168
+    },
+    {
+      "epoch": 0.634,
+      "learning_rate": 1.9735789028731603e-05,
+      "loss": 0.3915,
+      "step": 3170
+    },
+    {
+      "epoch": 0.6344,
+      "learning_rate": 1.9738967917112752e-05,
+      "loss": 0.3595,
+      "step": 3172
+    },
+    {
+      "epoch": 0.6348,
+      "learning_rate": 1.9742127818877605e-05,
+      "loss": 0.2648,
+      "step": 3174
+    },
+    {
+      "epoch": 0.6352,
+      "learning_rate": 1.974526872786577e-05,
+      "loss": 0.1521,
+      "step": 3176
+    },
+    {
+      "epoch": 0.6356,
+      "learning_rate": 1.974839063795389e-05,
+      "loss": 0.3582,
+      "step": 3178
+    },
+    {
+      "epoch": 0.636,
+      "learning_rate": 1.9751493543055634e-05,
+      "loss": 0.3462,
+      "step": 3180
+    },
+    {
+      "epoch": 0.6364,
+      "learning_rate": 1.975457743712173e-05,
+      "loss": 0.2345,
+      "step": 3182
+    },
+    {
+      "epoch": 0.6368,
+      "learning_rate": 1.9757642314139977e-05,
+      "loss": 0.4006,
+      "step": 3184
+    },
+    {
+      "epoch": 0.6372,
+      "learning_rate": 1.976068816813523e-05,
+      "loss": 1.0042,
+      "step": 3186
+    },
+    {
+      "epoch": 0.6376,
+      "learning_rate": 1.976371499316945e-05,
+      "loss": 0.4905,
+      "step": 3188
+    },
+    {
+      "epoch": 0.638,
+      "learning_rate": 1.9766722783341675e-05,
+      "loss": 0.6901,
+      "step": 3190
+    },
+    {
+      "epoch": 0.6384,
+      "learning_rate": 1.9769711532788083e-05,
+      "loss": 0.3572,
+      "step": 3192
+    },
+    {
+      "epoch": 0.6388,
+      "learning_rate": 1.9772681235681933e-05,
+      "loss": 0.616,
+      "step": 3194
+    },
+    {
+      "epoch": 0.6392,
+      "learning_rate": 1.9775631886233655e-05,
+      "loss": 0.1736,
+      "step": 3196
+    },
+    {
+      "epoch": 0.6396,
+      "learning_rate": 1.977856347869079e-05,
+      "loss": 0.0096,
+      "step": 3198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9781476007338054e-05,
+      "loss": 0.5169,
+      "step": 3200
+    },
+    {
+      "epoch": 0.6404,
+      "learning_rate": 1.9784369466497333e-05,
+      "loss": 0.4037,
+      "step": 3202
+    },
+    {
+      "epoch": 0.6408,
+      "learning_rate": 1.978724385052766e-05,
+      "loss": 0.6378,
+      "step": 3204
+    },
+    {
+      "epoch": 0.6412,
+      "learning_rate": 1.97900991538253e-05,
+      "loss": 0.2244,
+      "step": 3206
+    },
+    {
+      "epoch": 0.6416,
+      "learning_rate": 1.9792935370823673e-05,
+      "loss": 0.4333,
+      "step": 3208
+    },
+    {
+      "epoch": 0.642,
+      "learning_rate": 1.979575249599344e-05,
+      "loss": 0.4466,
+      "step": 3210
+    },
+    {
+      "epoch": 0.6424,
+      "learning_rate": 1.979855052384247e-05,
+      "loss": 0.907,
+      "step": 3212
+    },
+    {
+      "epoch": 0.6428,
+      "learning_rate": 1.980132944891586e-05,
+      "loss": 0.8047,
+      "step": 3214
+    },
+    {
+      "epoch": 0.6432,
+      "learning_rate": 1.9804089265795956e-05,
+      "loss": 0.2624,
+      "step": 3216
+    },
+    {
+      "epoch": 0.6436,
+      "learning_rate": 1.9806829969102356e-05,
+      "loss": 0.5118,
+      "step": 3218
+    },
+    {
+      "epoch": 0.644,
+      "learning_rate": 1.9809551553491918e-05,
+      "loss": 0.3744,
+      "step": 3220
+    },
+    {
+      "epoch": 0.6444,
+      "learning_rate": 1.981225401365877e-05,
+      "loss": 0.3694,
+      "step": 3222
+    },
+    {
+      "epoch": 0.6448,
+      "learning_rate": 1.981493734433433e-05,
+      "loss": 0.554,
+      "step": 3224
+    },
+    {
+      "epoch": 0.6452,
+      "learning_rate": 1.981760154028731e-05,
+      "loss": 0.1444,
+      "step": 3226
+    },
+    {
+      "epoch": 0.6456,
+      "learning_rate": 1.982024659632372e-05,
+      "loss": 0.1377,
+      "step": 3228
+    },
+    {
+      "epoch": 0.646,
+      "learning_rate": 1.9822872507286887e-05,
+      "loss": 0.2529,
+      "step": 3230
+    },
+    {
+      "epoch": 0.6464,
+      "learning_rate": 1.9825479268057472e-05,
+      "loss": 0.3273,
+      "step": 3232
+    },
+    {
+      "epoch": 0.6468,
+      "learning_rate": 1.9828066873553445e-05,
+      "loss": 0.6496,
+      "step": 3234
+    },
+    {
+      "epoch": 0.6472,
+      "learning_rate": 1.9830635318730155e-05,
+      "loss": 0.3853,
+      "step": 3236
+    },
+    {
+      "epoch": 0.6476,
+      "learning_rate": 1.983318459858028e-05,
+      "loss": 0.2348,
+      "step": 3238
+    },
+    {
+      "epoch": 0.648,
+      "learning_rate": 1.9835714708133858e-05,
+      "loss": 0.3269,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6484,
+      "learning_rate": 1.983822564245833e-05,
+      "loss": 0.6148,
+      "step": 3242
+    },
+    {
+      "epoch": 0.6488,
+      "learning_rate": 1.9840717396658483e-05,
+      "loss": 0.5324,
+      "step": 3244
+    },
+    {
+      "epoch": 0.6492,
+      "learning_rate": 1.9843189965876525e-05,
+      "loss": 0.4096,
+      "step": 3246
+    },
+    {
+      "epoch": 0.6496,
+      "learning_rate": 1.9845643345292055e-05,
+      "loss": 0.1341,
+      "step": 3248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.3241,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6504,
+      "learning_rate": 1.9850492515621038e-05,
+      "loss": 0.1163,
+      "step": 3252
+    },
+    {
+      "epoch": 0.6508,
+      "learning_rate": 1.9852888297080785e-05,
+      "loss": 0.1876,
+      "step": 3254
+    },
+    {
+      "epoch": 0.6512,
+      "learning_rate": 1.985526486983063e-05,
+      "loss": 0.2528,
+      "step": 3256
+    },
+    {
+      "epoch": 0.6516,
+      "learning_rate": 1.9857622229237315e-05,
+      "loss": 0.2096,
+      "step": 3258
+    },
+    {
+      "epoch": 0.652,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 0.6599,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6524,
+      "learning_rate": 1.986227928967551e-05,
+      "loss": 0.3456,
+      "step": 3262
+    },
+    {
+      "epoch": 0.6528,
+      "learning_rate": 1.9864578981627844e-05,
+      "loss": 0.2582,
+      "step": 3264
+    },
+    {
+      "epoch": 0.6532,
+      "learning_rate": 1.986685944207868e-05,
+      "loss": 0.2767,
+      "step": 3266
+    },
+    {
+      "epoch": 0.6536,
+      "learning_rate": 1.9869120666582153e-05,
+      "loss": 0.4581,
+      "step": 3268
+    },
+    {
+      "epoch": 0.654,
+      "learning_rate": 1.9871362650729877e-05,
+      "loss": 0.2086,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6544,
+      "learning_rate": 1.9873585390151003e-05,
+      "loss": 0.695,
+      "step": 3272
+    },
+    {
+      "epoch": 0.6548,
+      "learning_rate": 1.9875788880512183e-05,
+      "loss": 0.4846,
+      "step": 3274
+    },
+    {
+      "epoch": 0.6552,
+      "learning_rate": 1.987797311751759e-05,
+      "loss": 0.2572,
+      "step": 3276
+    },
+    {
+      "epoch": 0.6556,
+      "learning_rate": 1.9880138096908955e-05,
+      "loss": 0.8055,
+      "step": 3278
+    },
+    {
+      "epoch": 0.656,
+      "learning_rate": 1.9882283814465528e-05,
+      "loss": 0.303,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6564,
+      "learning_rate": 1.9884410266004134e-05,
+      "loss": 0.8064,
+      "step": 3282
+    },
+    {
+      "epoch": 0.6568,
+      "learning_rate": 1.988651744737914e-05,
+      "loss": 0.3538,
+      "step": 3284
+    },
+    {
+      "epoch": 0.6572,
+      "learning_rate": 1.9888605354482494e-05,
+      "loss": 0.3561,
+      "step": 3286
+    },
+    {
+      "epoch": 0.6576,
+      "learning_rate": 1.9890673983243704e-05,
+      "loss": 0.2092,
+      "step": 3288
+    },
+    {
+      "epoch": 0.658,
+      "learning_rate": 1.9892723329629885e-05,
+      "loss": 0.4701,
+      "step": 3290
+    },
+    {
+      "epoch": 0.6584,
+      "learning_rate": 1.9894753389645723e-05,
+      "loss": 0.1267,
+      "step": 3292
+    },
+    {
+      "epoch": 0.6588,
+      "learning_rate": 1.989676415933351e-05,
+      "loss": 0.3411,
+      "step": 3294
+    },
+    {
+      "epoch": 0.6592,
+      "learning_rate": 1.9898755634773155e-05,
+      "loss": 0.2245,
+      "step": 3296
+    },
+    {
+      "epoch": 0.6596,
+      "learning_rate": 1.9900727812082174e-05,
+      "loss": 0.0913,
+      "step": 3298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 0.221,
+      "step": 3300
+    },
+    {
+      "epoch": 0.6604,
+      "learning_rate": 1.9904614256966514e-05,
+      "loss": 0.6643,
+      "step": 3302
+    },
+    {
+      "epoch": 0.6608,
+      "learning_rate": 1.9906528516965014e-05,
+      "loss": 0.5714,
+      "step": 3304
+    },
+    {
+      "epoch": 0.6612,
+      "learning_rate": 1.9908423463679246e-05,
+      "loss": 0.5798,
+      "step": 3306
+    },
+    {
+      "epoch": 0.6616,
+      "learning_rate": 1.9910299093414926e-05,
+      "loss": 0.1887,
+      "step": 3308
+    },
+    {
+      "epoch": 0.662,
+      "learning_rate": 1.991215540251542e-05,
+      "loss": 0.2578,
+      "step": 3310
+    },
+    {
+      "epoch": 0.6624,
+      "learning_rate": 1.9913992387361744e-05,
+      "loss": 0.5057,
+      "step": 3312
+    },
+    {
+      "epoch": 0.6628,
+      "learning_rate": 1.9915810044372618e-05,
+      "loss": 0.6017,
+      "step": 3314
+    },
+    {
+      "epoch": 0.6632,
+      "learning_rate": 1.9917608370004414e-05,
+      "loss": 0.5663,
+      "step": 3316
+    },
+    {
+      "epoch": 0.6636,
+      "learning_rate": 1.9919387360751216e-05,
+      "loss": 0.1981,
+      "step": 3318
+    },
+    {
+      "epoch": 0.664,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.2789,
+      "step": 3320
+    },
+    {
+      "epoch": 0.6644,
+      "learning_rate": 1.992288732375458e-05,
+      "loss": 0.7267,
+      "step": 3322
+    },
+    {
+      "epoch": 0.6648,
+      "learning_rate": 1.9924608289187786e-05,
+      "loss": 0.3839,
+      "step": 3324
+    },
+    {
+      "epoch": 0.6652,
+      "learning_rate": 1.992630990608929e-05,
+      "loss": 0.3621,
+      "step": 3326
+    },
+    {
+      "epoch": 0.6656,
+      "learning_rate": 1.9927992171141707e-05,
+      "loss": 0.2789,
+      "step": 3328
+    },
+    {
+      "epoch": 0.666,
+      "learning_rate": 1.992965508106537e-05,
+      "loss": 0.4148,
+      "step": 3330
+    },
+    {
+      "epoch": 0.6664,
+      "learning_rate": 1.9931298632618355e-05,
+      "loss": 0.6315,
+      "step": 3332
+    },
+    {
+      "epoch": 0.6668,
+      "learning_rate": 1.993292282259647e-05,
+      "loss": 0.2781,
+      "step": 3334
+    },
+    {
+      "epoch": 0.6672,
+      "learning_rate": 1.9934527647833276e-05,
+      "loss": 1.7686,
+      "step": 3336
+    },
+    {
+      "epoch": 0.6676,
+      "learning_rate": 1.9936113105200085e-05,
+      "loss": 0.4896,
+      "step": 3338
+    },
+    {
+      "epoch": 0.668,
+      "learning_rate": 1.9937679191605964e-05,
+      "loss": 0.8246,
+      "step": 3340
+    },
+    {
+      "epoch": 0.6684,
+      "learning_rate": 1.9939225903997748e-05,
+      "loss": 0.2005,
+      "step": 3342
+    },
+    {
+      "epoch": 0.6688,
+      "learning_rate": 1.9940753239360047e-05,
+      "loss": 0.2305,
+      "step": 3344
+    },
+    {
+      "epoch": 0.6692,
+      "learning_rate": 1.9942261194715236e-05,
+      "loss": 0.2591,
+      "step": 3346
+    },
+    {
+      "epoch": 0.6696,
+      "learning_rate": 1.994374976712348e-05,
+      "loss": 0.4019,
+      "step": 3348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 0.5286,
+      "step": 3350
+    },
+    {
+      "epoch": 0.6704,
+      "learning_rate": 1.994666875152874e-05,
+      "loss": 0.4193,
+      "step": 3352
+    },
+    {
+      "epoch": 0.6708,
+      "learning_rate": 1.994809915783505e-05,
+      "loss": 0.4511,
+      "step": 3354
+    },
+    {
+      "epoch": 0.6712,
+      "learning_rate": 1.9949510169813003e-05,
+      "loss": 0.2584,
+      "step": 3356
+    },
+    {
+      "epoch": 0.6716,
+      "learning_rate": 1.9950901784711768e-05,
+      "loss": 0.3761,
+      "step": 3358
+    },
+    {
+      "epoch": 0.672,
+      "learning_rate": 1.9952273999818312e-05,
+      "loss": 0.4273,
+      "step": 3360
+    },
+    {
+      "epoch": 0.6724,
+      "learning_rate": 1.995362681245744e-05,
+      "loss": 0.4113,
+      "step": 3362
+    },
+    {
+      "epoch": 0.6728,
+      "learning_rate": 1.995496021999177e-05,
+      "loss": 0.1796,
+      "step": 3364
+    },
+    {
+      "epoch": 0.6732,
+      "learning_rate": 1.995627421982176e-05,
+      "loss": 0.6577,
+      "step": 3366
+    },
+    {
+      "epoch": 0.6736,
+      "learning_rate": 1.9957568809385693e-05,
+      "loss": 0.2723,
+      "step": 3368
+    },
+    {
+      "epoch": 0.674,
+      "learning_rate": 1.9958843986159705e-05,
+      "loss": 0.2148,
+      "step": 3370
+    },
+    {
+      "epoch": 0.6744,
+      "learning_rate": 1.9960099747657774e-05,
+      "loss": 0.1762,
+      "step": 3372
+    },
+    {
+      "epoch": 0.6748,
+      "learning_rate": 1.9961336091431725e-05,
+      "loss": 0.3485,
+      "step": 3374
+    },
+    {
+      "epoch": 0.6752,
+      "learning_rate": 1.996255301507125e-05,
+      "loss": 0.1408,
+      "step": 3376
+    },
+    {
+      "epoch": 0.6756,
+      "learning_rate": 1.9963750516203884e-05,
+      "loss": 0.5658,
+      "step": 3378
+    },
+    {
+      "epoch": 0.676,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 0.3747,
+      "step": 3380
+    },
+    {
+      "epoch": 0.6764,
+      "learning_rate": 1.996608724164801e-05,
+      "loss": 0.8133,
+      "step": 3382
+    },
+    {
+      "epoch": 0.6768,
+      "learning_rate": 1.9967226461403934e-05,
+      "loss": 0.5632,
+      "step": 3384
+    },
+    {
+      "epoch": 0.6772,
+      "learning_rate": 1.9968346249541848e-05,
+      "loss": 0.344,
+      "step": 3386
+    },
+    {
+      "epoch": 0.6776,
+      "learning_rate": 1.996944660387867e-05,
+      "loss": 0.4145,
+      "step": 3388
+    },
+    {
+      "epoch": 0.678,
+      "learning_rate": 1.9970527522269204e-05,
+      "loss": 0.0796,
+      "step": 3390
+    },
+    {
+      "epoch": 0.6784,
+      "learning_rate": 1.997158900260614e-05,
+      "loss": 0.2299,
+      "step": 3392
+    },
+    {
+      "epoch": 0.6788,
+      "learning_rate": 1.997263104282007e-05,
+      "loss": 0.4028,
+      "step": 3394
+    },
+    {
+      "epoch": 0.6792,
+      "learning_rate": 1.9973653640879486e-05,
+      "loss": 0.832,
+      "step": 3396
+    },
+    {
+      "epoch": 0.6796,
+      "learning_rate": 1.9974656794790777e-05,
+      "loss": 0.6258,
+      "step": 3398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9975640502598246e-05,
+      "loss": 0.2126,
+      "step": 3400
+    },
+    {
+      "epoch": 0.6804,
+      "learning_rate": 1.99766047623841e-05,
+      "loss": 0.1655,
+      "step": 3402
+    },
+    {
+      "epoch": 0.6808,
+      "learning_rate": 1.997754957226847e-05,
+      "loss": 0.392,
+      "step": 3404
+    },
+    {
+      "epoch": 0.6812,
+      "learning_rate": 1.9978474930409396e-05,
+      "loss": 0.3815,
+      "step": 3406
+    },
+    {
+      "epoch": 0.6816,
+      "learning_rate": 1.9979380835002846e-05,
+      "loss": 0.4513,
+      "step": 3408
+    },
+    {
+      "epoch": 0.682,
+      "learning_rate": 1.9980267284282718e-05,
+      "loss": 0.684,
+      "step": 3410
+    },
+    {
+      "epoch": 0.6824,
+      "learning_rate": 1.9981134276520828e-05,
+      "loss": 0.7855,
+      "step": 3412
+    },
+    {
+      "epoch": 0.6828,
+      "learning_rate": 1.9981981810026932e-05,
+      "loss": 0.2719,
+      "step": 3414
+    },
+    {
+      "epoch": 0.6832,
+      "learning_rate": 1.998280988314872e-05,
+      "loss": 0.2801,
+      "step": 3416
+    },
+    {
+      "epoch": 0.6836,
+      "learning_rate": 1.9983618494271825e-05,
+      "loss": 0.2722,
+      "step": 3418
+    },
+    {
+      "epoch": 0.684,
+      "learning_rate": 1.998440764181981e-05,
+      "loss": 0.1707,
+      "step": 3420
+    },
+    {
+      "epoch": 0.6844,
+      "learning_rate": 1.99851773242542e-05,
+      "loss": 0.3289,
+      "step": 3422
+    },
+    {
+      "epoch": 0.6848,
+      "learning_rate": 1.9985927540074453e-05,
+      "loss": 0.3892,
+      "step": 3424
+    },
+    {
+      "epoch": 0.6852,
+      "learning_rate": 1.9986658287817992e-05,
+      "loss": 0.2403,
+      "step": 3426
+    },
+    {
+      "epoch": 0.6856,
+      "learning_rate": 1.998736956606018e-05,
+      "loss": 0.3955,
+      "step": 3428
+    },
+    {
+      "epoch": 0.686,
+      "learning_rate": 1.9988061373414342e-05,
+      "loss": 0.1709,
+      "step": 3430
+    },
+    {
+      "epoch": 0.6864,
+      "learning_rate": 1.9988733708531772e-05,
+      "loss": 0.294,
+      "step": 3432
+    },
+    {
+      "epoch": 0.6868,
+      "learning_rate": 1.9989386570101712e-05,
+      "loss": 0.3262,
+      "step": 3434
+    },
+    {
+      "epoch": 0.6872,
+      "learning_rate": 1.9990019956851384e-05,
+      "loss": 0.7574,
+      "step": 3436
+    },
+    {
+      "epoch": 0.6876,
+      "learning_rate": 1.9990633867545956e-05,
+      "loss": 0.2661,
+      "step": 3438
+    },
+    {
+      "epoch": 0.688,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 0.4533,
+      "step": 3440
+    },
+    {
+      "epoch": 0.6884,
+      "learning_rate": 1.9991803256020393e-05,
+      "loss": 1.2708,
+      "step": 3442
+    },
+    {
+      "epoch": 0.6888,
+      "learning_rate": 1.999235873152047e-05,
+      "loss": 0.3983,
+      "step": 3444
+    },
+    {
+      "epoch": 0.6892,
+      "learning_rate": 1.9992894726405894e-05,
+      "loss": 0.2866,
+      "step": 3446
+    },
+    {
+      "epoch": 0.6896,
+      "learning_rate": 1.9993411239631713e-05,
+      "loss": 0.5536,
+      "step": 3448
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.999390827019096e-05,
+      "loss": 0.7414,
+      "step": 3450
+    },
+    {
+      "epoch": 0.6904,
+      "learning_rate": 1.9994385817114644e-05,
+      "loss": 0.6119,
+      "step": 3452
+    },
+    {
+      "epoch": 0.6908,
+      "learning_rate": 1.999484387947177e-05,
+      "loss": 0.2394,
+      "step": 3454
+    },
+    {
+      "epoch": 0.6912,
+      "learning_rate": 1.9995282456369313e-05,
+      "loss": 0.1891,
+      "step": 3456
+    },
+    {
+      "epoch": 0.6916,
+      "learning_rate": 1.9995701546952252e-05,
+      "loss": 0.3597,
+      "step": 3458
+    },
+    {
+      "epoch": 0.692,
+      "learning_rate": 1.9996101150403543e-05,
+      "loss": 0.4461,
+      "step": 3460
+    },
+    {
+      "epoch": 0.6924,
+      "learning_rate": 1.9996481265944146e-05,
+      "loss": 0.2969,
+      "step": 3462
+    },
+    {
+      "epoch": 0.6928,
+      "learning_rate": 1.9996841892833e-05,
+      "loss": 0.3047,
+      "step": 3464
+    },
+    {
+      "epoch": 0.6932,
+      "learning_rate": 1.999718303036705e-05,
+      "loss": 0.6878,
+      "step": 3466
+    },
+    {
+      "epoch": 0.6936,
+      "learning_rate": 1.9997504677881224e-05,
+      "loss": 0.5323,
+      "step": 3468
+    },
+    {
+      "epoch": 0.694,
+      "learning_rate": 1.9997806834748455e-05,
+      "loss": 0.2311,
+      "step": 3470
+    },
+    {
+      "epoch": 0.6944,
+      "learning_rate": 1.999808950037968e-05,
+      "loss": 0.3805,
+      "step": 3472
+    },
+    {
+      "epoch": 0.6948,
+      "learning_rate": 1.9998352674223816e-05,
+      "loss": 0.4133,
+      "step": 3474
+    },
+    {
+      "epoch": 0.6952,
+      "learning_rate": 1.9998596355767805e-05,
+      "loss": 0.3408,
+      "step": 3476
+    },
+    {
+      "epoch": 0.6956,
+      "learning_rate": 1.999882054453657e-05,
+      "loss": 0.2311,
+      "step": 3478
+    },
+    {
+      "epoch": 0.696,
+      "learning_rate": 1.9999025240093045e-05,
+      "loss": 0.0774,
+      "step": 3480
+    },
+    {
+      "epoch": 0.6964,
+      "learning_rate": 1.9999210442038164e-05,
+      "loss": 0.2566,
+      "step": 3482
+    },
+    {
+      "epoch": 0.6968,
+      "learning_rate": 1.9999376150010868e-05,
+      "loss": 0.6279,
+      "step": 3484
+    },
+    {
+      "epoch": 0.6972,
+      "learning_rate": 1.99995223636881e-05,
+      "loss": 0.2741,
+      "step": 3486
+    },
+    {
+      "epoch": 0.6976,
+      "learning_rate": 1.9999649082784807e-05,
+      "loss": 0.4703,
+      "step": 3488
+    },
+    {
+      "epoch": 0.698,
+      "learning_rate": 1.9999756307053947e-05,
+      "loss": 0.6449,
+      "step": 3490
+    },
+    {
+      "epoch": 0.6984,
+      "learning_rate": 1.9999844036286483e-05,
+      "loss": 0.6222,
+      "step": 3492
+    },
+    {
+      "epoch": 0.6988,
+      "learning_rate": 1.9999912270311376e-05,
+      "loss": 0.6077,
+      "step": 3494
+    },
+    {
+      "epoch": 0.6992,
+      "learning_rate": 1.9999961008995607e-05,
+      "loss": 0.6108,
+      "step": 3496
+    },
+    {
+      "epoch": 0.6996,
+      "learning_rate": 1.9999990252244153e-05,
+      "loss": 0.5728,
+      "step": 3498
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2e-05,
+      "loss": 0.3419,
+      "step": 3500
+    },
+    {
+      "epoch": 0.7004,
+      "learning_rate": 1.9999990252244153e-05,
+      "loss": 0.5606,
+      "step": 3502
+    },
+    {
+      "epoch": 0.7008,
+      "learning_rate": 1.9999961008995607e-05,
+      "loss": 0.4986,
+      "step": 3504
+    },
+    {
+      "epoch": 0.7012,
+      "learning_rate": 1.9999912270311376e-05,
+      "loss": 0.2225,
+      "step": 3506
+    },
+    {
+      "epoch": 0.7016,
+      "learning_rate": 1.9999844036286483e-05,
+      "loss": 0.2449,
+      "step": 3508
+    },
+    {
+      "epoch": 0.702,
+      "learning_rate": 1.9999756307053947e-05,
+      "loss": 0.2541,
+      "step": 3510
+    },
+    {
+      "epoch": 0.7024,
+      "learning_rate": 1.9999649082784807e-05,
+      "loss": 0.4902,
+      "step": 3512
+    },
+    {
+      "epoch": 0.7028,
+      "learning_rate": 1.99995223636881e-05,
+      "loss": 0.1641,
+      "step": 3514
+    },
+    {
+      "epoch": 0.7032,
+      "learning_rate": 1.9999376150010868e-05,
+      "loss": 0.3167,
+      "step": 3516
+    },
+    {
+      "epoch": 0.7036,
+      "learning_rate": 1.9999210442038164e-05,
+      "loss": 0.324,
+      "step": 3518
+    },
+    {
+      "epoch": 0.704,
+      "learning_rate": 1.9999025240093045e-05,
+      "loss": 0.2836,
+      "step": 3520
+    },
+    {
+      "epoch": 0.7044,
+      "learning_rate": 1.999882054453657e-05,
+      "loss": 0.4192,
+      "step": 3522
+    },
+    {
+      "epoch": 0.7048,
+      "learning_rate": 1.9998596355767805e-05,
+      "loss": 0.4379,
+      "step": 3524
+    },
+    {
+      "epoch": 0.7052,
+      "learning_rate": 1.9998352674223816e-05,
+      "loss": 0.3336,
+      "step": 3526
+    },
+    {
+      "epoch": 0.7056,
+      "learning_rate": 1.999808950037968e-05,
+      "loss": 0.2089,
+      "step": 3528
+    },
+    {
+      "epoch": 0.706,
+      "learning_rate": 1.9997806834748455e-05,
+      "loss": 0.4131,
+      "step": 3530
+    },
+    {
+      "epoch": 0.7064,
+      "learning_rate": 1.9997504677881224e-05,
+      "loss": 0.1525,
+      "step": 3532
+    },
+    {
+      "epoch": 0.7068,
+      "learning_rate": 1.999718303036705e-05,
+      "loss": 0.2674,
+      "step": 3534
+    },
+    {
+      "epoch": 0.7072,
+      "learning_rate": 1.9996841892833e-05,
+      "loss": 0.6451,
+      "step": 3536
+    },
+    {
+      "epoch": 0.7076,
+      "learning_rate": 1.9996481265944146e-05,
+      "loss": 0.0169,
+      "step": 3538
+    },
+    {
+      "epoch": 0.708,
+      "learning_rate": 1.9996101150403547e-05,
+      "loss": 0.6018,
+      "step": 3540
+    },
+    {
+      "epoch": 0.7084,
+      "learning_rate": 1.9995701546952252e-05,
+      "loss": 0.1616,
+      "step": 3542
+    },
+    {
+      "epoch": 0.7088,
+      "learning_rate": 1.9995282456369313e-05,
+      "loss": 0.947,
+      "step": 3544
+    },
+    {
+      "epoch": 0.7092,
+      "learning_rate": 1.9994843879471766e-05,
+      "loss": 0.3946,
+      "step": 3546
+    },
+    {
+      "epoch": 0.7096,
+      "learning_rate": 1.9994385817114644e-05,
+      "loss": 0.1565,
+      "step": 3548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999390827019096e-05,
+      "loss": 0.2182,
+      "step": 3550
+    },
+    {
+      "epoch": 0.7104,
+      "learning_rate": 1.9993411239631713e-05,
+      "loss": 0.3062,
+      "step": 3552
+    },
+    {
+      "epoch": 0.7108,
+      "learning_rate": 1.9992894726405898e-05,
+      "loss": 0.3799,
+      "step": 3554
+    },
+    {
+      "epoch": 0.7112,
+      "learning_rate": 1.999235873152047e-05,
+      "loss": 0.347,
+      "step": 3556
+    },
+    {
+      "epoch": 0.7116,
+      "learning_rate": 1.9991803256020393e-05,
+      "loss": 0.2475,
+      "step": 3558
+    },
+    {
+      "epoch": 0.712,
+      "learning_rate": 1.9991228300988586e-05,
+      "loss": 0.4497,
+      "step": 3560
+    },
+    {
+      "epoch": 0.7124,
+      "learning_rate": 1.9990633867545956e-05,
+      "loss": 0.2116,
+      "step": 3562
+    },
+    {
+      "epoch": 0.7128,
+      "learning_rate": 1.9990019956851384e-05,
+      "loss": 0.276,
+      "step": 3564
+    },
+    {
+      "epoch": 0.7132,
+      "learning_rate": 1.9989386570101716e-05,
+      "loss": 1.1488,
+      "step": 3566
+    },
+    {
+      "epoch": 0.7136,
+      "learning_rate": 1.9988733708531772e-05,
+      "loss": 0.5636,
+      "step": 3568
+    },
+    {
+      "epoch": 0.714,
+      "learning_rate": 1.9988061373414342e-05,
+      "loss": 0.1948,
+      "step": 3570
+    },
+    {
+      "epoch": 0.7144,
+      "learning_rate": 1.998736956606018e-05,
+      "loss": 0.6853,
+      "step": 3572
+    },
+    {
+      "epoch": 0.7148,
+      "learning_rate": 1.998665828781799e-05,
+      "loss": 1.031,
+      "step": 3574
+    },
+    {
+      "epoch": 0.7152,
+      "learning_rate": 1.9985927540074453e-05,
+      "loss": 0.2748,
+      "step": 3576
+    },
+    {
+      "epoch": 0.7156,
+      "learning_rate": 1.99851773242542e-05,
+      "loss": 0.3538,
+      "step": 3578
+    },
+    {
+      "epoch": 0.716,
+      "learning_rate": 1.9984407641819812e-05,
+      "loss": 0.4645,
+      "step": 3580
+    },
+    {
+      "epoch": 0.7164,
+      "learning_rate": 1.9983618494271825e-05,
+      "loss": 0.4368,
+      "step": 3582
+    },
+    {
+      "epoch": 0.7168,
+      "learning_rate": 1.998280988314872e-05,
+      "loss": 0.9549,
+      "step": 3584
+    },
+    {
+      "epoch": 0.7172,
+      "learning_rate": 1.9981981810026932e-05,
+      "loss": 0.3611,
+      "step": 3586
+    },
+    {
+      "epoch": 0.7176,
+      "learning_rate": 1.9981134276520828e-05,
+      "loss": 0.7267,
+      "step": 3588
+    },
+    {
+      "epoch": 0.718,
+      "learning_rate": 1.9980267284282718e-05,
+      "loss": 0.6691,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7184,
+      "learning_rate": 1.9979380835002846e-05,
+      "loss": 1.1764,
+      "step": 3592
+    },
+    {
+      "epoch": 0.7188,
+      "learning_rate": 1.9978474930409396e-05,
+      "loss": 0.9693,
+      "step": 3594
+    },
+    {
+      "epoch": 0.7192,
+      "learning_rate": 1.9977549572268467e-05,
+      "loss": 0.3634,
+      "step": 3596
+    },
+    {
+      "epoch": 0.7196,
+      "learning_rate": 1.99766047623841e-05,
+      "loss": 0.2632,
+      "step": 3598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9975640502598246e-05,
+      "loss": 0.4736,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7204,
+      "learning_rate": 1.9974656794790777e-05,
+      "loss": 0.2326,
+      "step": 3602
+    },
+    {
+      "epoch": 0.7208,
+      "learning_rate": 1.9973653640879486e-05,
+      "loss": 0.4168,
+      "step": 3604
+    },
+    {
+      "epoch": 0.7212,
+      "learning_rate": 1.9972631042820074e-05,
+      "loss": 0.2944,
+      "step": 3606
+    },
+    {
+      "epoch": 0.7216,
+      "learning_rate": 1.997158900260614e-05,
+      "loss": 0.264,
+      "step": 3608
+    },
+    {
+      "epoch": 0.722,
+      "learning_rate": 1.9970527522269204e-05,
+      "loss": 0.2096,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7224,
+      "learning_rate": 1.9969446603878673e-05,
+      "loss": 0.1529,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7228,
+      "learning_rate": 1.9968346249541848e-05,
+      "loss": 0.37,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7232,
+      "learning_rate": 1.9967226461403934e-05,
+      "loss": 0.3915,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7236,
+      "learning_rate": 1.996608724164801e-05,
+      "loss": 1.1172,
+      "step": 3618
+    },
+    {
+      "epoch": 0.724,
+      "learning_rate": 1.9964928592495046e-05,
+      "loss": 0.1966,
+      "step": 3620
+    },
+    {
+      "epoch": 0.7244,
+      "learning_rate": 1.9963750516203887e-05,
+      "loss": 0.0913,
+      "step": 3622
+    },
+    {
+      "epoch": 0.7248,
+      "learning_rate": 1.996255301507125e-05,
+      "loss": 0.365,
+      "step": 3624
+    },
+    {
+      "epoch": 0.7252,
+      "learning_rate": 1.9961336091431728e-05,
+      "loss": 0.256,
+      "step": 3626
+    },
+    {
+      "epoch": 0.7256,
+      "learning_rate": 1.9960099747657774e-05,
+      "loss": 0.3708,
+      "step": 3628
+    },
+    {
+      "epoch": 0.726,
+      "learning_rate": 1.9958843986159705e-05,
+      "loss": 0.6788,
+      "step": 3630
+    },
+    {
+      "epoch": 0.7264,
+      "learning_rate": 1.9957568809385693e-05,
+      "loss": 0.4347,
+      "step": 3632
+    },
+    {
+      "epoch": 0.7268,
+      "learning_rate": 1.995627421982176e-05,
+      "loss": 0.4676,
+      "step": 3634
+    },
+    {
+      "epoch": 0.7272,
+      "learning_rate": 1.995496021999177e-05,
+      "loss": 0.5483,
+      "step": 3636
+    },
+    {
+      "epoch": 0.7276,
+      "learning_rate": 1.995362681245744e-05,
+      "loss": 0.6501,
+      "step": 3638
+    },
+    {
+      "epoch": 0.728,
+      "learning_rate": 1.9952273999818312e-05,
+      "loss": 0.3121,
+      "step": 3640
+    },
+    {
+      "epoch": 0.7284,
+      "learning_rate": 1.9950901784711768e-05,
+      "loss": 0.6384,
+      "step": 3642
+    },
+    {
+      "epoch": 0.7288,
+      "learning_rate": 1.9949510169813006e-05,
+      "loss": 0.2575,
+      "step": 3644
+    },
+    {
+      "epoch": 0.7292,
+      "learning_rate": 1.994809915783505e-05,
+      "loss": 0.2528,
+      "step": 3646
+    },
+    {
+      "epoch": 0.7296,
+      "learning_rate": 1.9946668751528745e-05,
+      "loss": 0.4282,
+      "step": 3648
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9945218953682736e-05,
+      "loss": 0.1945,
+      "step": 3650
+    },
+    {
+      "epoch": 0.7304,
+      "learning_rate": 1.994374976712348e-05,
+      "loss": 0.5285,
+      "step": 3652
+    },
+    {
+      "epoch": 0.7308,
+      "learning_rate": 1.9942261194715236e-05,
+      "loss": 0.3788,
+      "step": 3654
+    },
+    {
+      "epoch": 0.7312,
+      "learning_rate": 1.9940753239360047e-05,
+      "loss": 0.2462,
+      "step": 3656
+    },
+    {
+      "epoch": 0.7316,
+      "learning_rate": 1.9939225903997748e-05,
+      "loss": 0.4954,
+      "step": 3658
+    },
+    {
+      "epoch": 0.732,
+      "learning_rate": 1.9937679191605964e-05,
+      "loss": 0.8218,
+      "step": 3660
+    },
+    {
+      "epoch": 0.7324,
+      "learning_rate": 1.993611310520009e-05,
+      "loss": 0.2381,
+      "step": 3662
+    },
+    {
+      "epoch": 0.7328,
+      "learning_rate": 1.993452764783328e-05,
+      "loss": 0.1129,
+      "step": 3664
+    },
+    {
+      "epoch": 0.7332,
+      "learning_rate": 1.993292282259647e-05,
+      "loss": 0.3222,
+      "step": 3666
+    },
+    {
+      "epoch": 0.7336,
+      "learning_rate": 1.9931298632618352e-05,
+      "loss": 0.3178,
+      "step": 3668
+    },
+    {
+      "epoch": 0.734,
+      "learning_rate": 1.9929655081065373e-05,
+      "loss": 0.447,
+      "step": 3670
+    },
+    {
+      "epoch": 0.7344,
+      "learning_rate": 1.9927992171141707e-05,
+      "loss": 0.0943,
+      "step": 3672
+    },
+    {
+      "epoch": 0.7348,
+      "learning_rate": 1.992630990608929e-05,
+      "loss": 0.3269,
+      "step": 3674
+    },
+    {
+      "epoch": 0.7352,
+      "learning_rate": 1.9924608289187786e-05,
+      "loss": 0.8699,
+      "step": 3676
+    },
+    {
+      "epoch": 0.7356,
+      "learning_rate": 1.992288732375458e-05,
+      "loss": 0.3081,
+      "step": 3678
+    },
+    {
+      "epoch": 0.736,
+      "learning_rate": 1.9921147013144782e-05,
+      "loss": 0.1423,
+      "step": 3680
+    },
+    {
+      "epoch": 0.7364,
+      "learning_rate": 1.9919387360751216e-05,
+      "loss": 0.5361,
+      "step": 3682
+    },
+    {
+      "epoch": 0.7368,
+      "learning_rate": 1.9917608370004417e-05,
+      "loss": 0.5461,
+      "step": 3684
+    },
+    {
+      "epoch": 0.7372,
+      "learning_rate": 1.991581004437262e-05,
+      "loss": 0.3078,
+      "step": 3686
+    },
+    {
+      "epoch": 0.7376,
+      "learning_rate": 1.9913992387361747e-05,
+      "loss": 0.0921,
+      "step": 3688
+    },
+    {
+      "epoch": 0.738,
+      "learning_rate": 1.991215540251542e-05,
+      "loss": 0.6288,
+      "step": 3690
+    },
+    {
+      "epoch": 0.7384,
+      "learning_rate": 1.9910299093414932e-05,
+      "loss": 0.2791,
+      "step": 3692
+    },
+    {
+      "epoch": 0.7388,
+      "learning_rate": 1.9908423463679246e-05,
+      "loss": 0.2175,
+      "step": 3694
+    },
+    {
+      "epoch": 0.7392,
+      "learning_rate": 1.990652851696501e-05,
+      "loss": 0.5877,
+      "step": 3696
+    },
+    {
+      "epoch": 0.7396,
+      "learning_rate": 1.9904614256966517e-05,
+      "loss": 0.1377,
+      "step": 3698
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9902680687415704e-05,
+      "loss": 0.4203,
+      "step": 3700
+    },
+    {
+      "epoch": 0.7404,
+      "learning_rate": 1.9900727812082177e-05,
+      "loss": 0.1686,
+      "step": 3702
+    },
+    {
+      "epoch": 0.7408,
+      "learning_rate": 1.9898755634773155e-05,
+      "loss": 0.205,
+      "step": 3704
+    },
+    {
+      "epoch": 0.7412,
+      "learning_rate": 1.9896764159333518e-05,
+      "loss": 0.6709,
+      "step": 3706
+    },
+    {
+      "epoch": 0.7416,
+      "learning_rate": 1.9894753389645723e-05,
+      "loss": 0.433,
+      "step": 3708
+    },
+    {
+      "epoch": 0.742,
+      "learning_rate": 1.9892723329629885e-05,
+      "loss": 0.2659,
+      "step": 3710
+    },
+    {
+      "epoch": 0.7424,
+      "learning_rate": 1.9890673983243704e-05,
+      "loss": 0.2952,
+      "step": 3712
+    },
+    {
+      "epoch": 0.7428,
+      "learning_rate": 1.9888605354482494e-05,
+      "loss": 0.2775,
+      "step": 3714
+    },
+    {
+      "epoch": 0.7432,
+      "learning_rate": 1.9886517447379143e-05,
+      "loss": 0.4384,
+      "step": 3716
+    },
+    {
+      "epoch": 0.7436,
+      "learning_rate": 1.9884410266004134e-05,
+      "loss": 0.2011,
+      "step": 3718
+    },
+    {
+      "epoch": 0.744,
+      "learning_rate": 1.988228381446553e-05,
+      "loss": 0.2599,
+      "step": 3720
+    },
+    {
+      "epoch": 0.7444,
+      "learning_rate": 1.9880138096908955e-05,
+      "loss": 0.2791,
+      "step": 3722
+    },
+    {
+      "epoch": 0.7448,
+      "learning_rate": 1.987797311751759e-05,
+      "loss": 0.403,
+      "step": 3724
+    },
+    {
+      "epoch": 0.7452,
+      "learning_rate": 1.9875788880512183e-05,
+      "loss": 0.6661,
+      "step": 3726
+    },
+    {
+      "epoch": 0.7456,
+      "learning_rate": 1.9873585390151007e-05,
+      "loss": 0.5803,
+      "step": 3728
+    },
+    {
+      "epoch": 0.746,
+      "learning_rate": 1.987136265072988e-05,
+      "loss": 0.2354,
+      "step": 3730
+    },
+    {
+      "epoch": 0.7464,
+      "learning_rate": 1.9869120666582153e-05,
+      "loss": 0.5197,
+      "step": 3732
+    },
+    {
+      "epoch": 0.7468,
+      "learning_rate": 1.9866859442078685e-05,
+      "loss": 0.3423,
+      "step": 3734
+    },
+    {
+      "epoch": 0.7472,
+      "learning_rate": 1.9864578981627844e-05,
+      "loss": 0.1575,
+      "step": 3736
+    },
+    {
+      "epoch": 0.7476,
+      "learning_rate": 1.986227928967551e-05,
+      "loss": 0.5374,
+      "step": 3738
+    },
+    {
+      "epoch": 0.748,
+      "learning_rate": 1.985996037070505e-05,
+      "loss": 0.19,
+      "step": 3740
+    },
+    {
+      "epoch": 0.7484,
+      "learning_rate": 1.985762222923732e-05,
+      "loss": 0.4784,
+      "step": 3742
+    },
+    {
+      "epoch": 0.7488,
+      "learning_rate": 1.985526486983063e-05,
+      "loss": 0.5871,
+      "step": 3744
+    },
+    {
+      "epoch": 0.7492,
+      "learning_rate": 1.985288829708079e-05,
+      "loss": 0.4987,
+      "step": 3746
+    },
+    {
+      "epoch": 0.7496,
+      "learning_rate": 1.9850492515621038e-05,
+      "loss": 0.4662,
+      "step": 3748
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.5098,
+      "step": 3750
+    },
+    {
+      "epoch": 0.7504,
+      "learning_rate": 1.9845643345292055e-05,
+      "loss": 0.6671,
+      "step": 3752
+    },
+    {
+      "epoch": 0.7508,
+      "learning_rate": 1.9843189965876525e-05,
+      "loss": 0.1353,
+      "step": 3754
+    },
+    {
+      "epoch": 0.7512,
+      "learning_rate": 1.9840717396658483e-05,
+      "loss": 0.3356,
+      "step": 3756
+    },
+    {
+      "epoch": 0.7516,
+      "learning_rate": 1.983822564245833e-05,
+      "loss": 0.4787,
+      "step": 3758
+    },
+    {
+      "epoch": 0.752,
+      "learning_rate": 1.983571470813386e-05,
+      "loss": 0.6368,
+      "step": 3760
+    },
+    {
+      "epoch": 0.7524,
+      "learning_rate": 1.9833184598580276e-05,
+      "loss": 0.7173,
+      "step": 3762
+    },
+    {
+      "epoch": 0.7528,
+      "learning_rate": 1.983063531873016e-05,
+      "loss": 0.3104,
+      "step": 3764
+    },
+    {
+      "epoch": 0.7532,
+      "learning_rate": 1.982806687355345e-05,
+      "loss": 0.1855,
+      "step": 3766
+    },
+    {
+      "epoch": 0.7536,
+      "learning_rate": 1.982547926805747e-05,
+      "loss": 0.5287,
+      "step": 3768
+    },
+    {
+      "epoch": 0.754,
+      "learning_rate": 1.982287250728689e-05,
+      "loss": 0.1261,
+      "step": 3770
+    },
+    {
+      "epoch": 0.7544,
+      "learning_rate": 1.9820246596323724e-05,
+      "loss": 0.474,
+      "step": 3772
+    },
+    {
+      "epoch": 0.7548,
+      "learning_rate": 1.981760154028731e-05,
+      "loss": 0.6094,
+      "step": 3774
+    },
+    {
+      "epoch": 0.7552,
+      "learning_rate": 1.981493734433433e-05,
+      "loss": 0.3466,
+      "step": 3776
+    },
+    {
+      "epoch": 0.7556,
+      "learning_rate": 1.9812254013658773e-05,
+      "loss": 0.3143,
+      "step": 3778
+    },
+    {
+      "epoch": 0.756,
+      "learning_rate": 1.9809551553491918e-05,
+      "loss": 0.7478,
+      "step": 3780
+    },
+    {
+      "epoch": 0.7564,
+      "learning_rate": 1.9806829969102356e-05,
+      "loss": 0.4192,
+      "step": 3782
+    },
+    {
+      "epoch": 0.7568,
+      "learning_rate": 1.9804089265795963e-05,
+      "loss": 0.3517,
+      "step": 3784
+    },
+    {
+      "epoch": 0.7572,
+      "learning_rate": 1.9801329448915863e-05,
+      "loss": 0.4283,
+      "step": 3786
+    },
+    {
+      "epoch": 0.7576,
+      "learning_rate": 1.979855052384247e-05,
+      "loss": 0.5023,
+      "step": 3788
+    },
+    {
+      "epoch": 0.758,
+      "learning_rate": 1.979575249599344e-05,
+      "loss": 0.6817,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7584,
+      "learning_rate": 1.979293537082368e-05,
+      "loss": 0.5462,
+      "step": 3792
+    },
+    {
+      "epoch": 0.7588,
+      "learning_rate": 1.9790099153825303e-05,
+      "loss": 0.6389,
+      "step": 3794
+    },
+    {
+      "epoch": 0.7592,
+      "learning_rate": 1.9787243850527663e-05,
+      "loss": 0.4307,
+      "step": 3796
+    },
+    {
+      "epoch": 0.7596,
+      "learning_rate": 1.978436946649733e-05,
+      "loss": 0.5304,
+      "step": 3798
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.978147600733806e-05,
+      "loss": 0.4771,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7604,
+      "learning_rate": 1.9778563478690793e-05,
+      "loss": 0.4063,
+      "step": 3802
+    },
+    {
+      "epoch": 0.7608,
+      "learning_rate": 1.977563188623365e-05,
+      "loss": 0.2103,
+      "step": 3804
+    },
+    {
+      "epoch": 0.7612,
+      "learning_rate": 1.977268123568194e-05,
+      "loss": 0.4816,
+      "step": 3806
+    },
+    {
+      "epoch": 0.7616,
+      "learning_rate": 1.9769711532788086e-05,
+      "loss": 0.4608,
+      "step": 3808
+    },
+    {
+      "epoch": 0.762,
+      "learning_rate": 1.9766722783341682e-05,
+      "loss": 0.7588,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7624,
+      "learning_rate": 1.9763714993169448e-05,
+      "loss": 0.3087,
+      "step": 3812
+    },
+    {
+      "epoch": 0.7628,
+      "learning_rate": 1.9760688168135236e-05,
+      "loss": 0.5229,
+      "step": 3814
+    },
+    {
+      "epoch": 0.7632,
+      "learning_rate": 1.9757642314139977e-05,
+      "loss": 0.869,
+      "step": 3816
+    },
+    {
+      "epoch": 0.7636,
+      "learning_rate": 1.9754577437121733e-05,
+      "loss": 0.693,
+      "step": 3818
+    },
+    {
+      "epoch": 0.764,
+      "learning_rate": 1.9751493543055638e-05,
+      "loss": 0.2707,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7644,
+      "learning_rate": 1.974839063795389e-05,
+      "loss": 0.2564,
+      "step": 3822
+    },
+    {
+      "epoch": 0.7648,
+      "learning_rate": 1.9745268727865774e-05,
+      "loss": 0.2223,
+      "step": 3824
+    },
+    {
+      "epoch": 0.7652,
+      "learning_rate": 1.97421278188776e-05,
+      "loss": 0.6009,
+      "step": 3826
+    },
+    {
+      "epoch": 0.7656,
+      "learning_rate": 1.973896791711276e-05,
+      "loss": 0.3943,
+      "step": 3828
+    },
+    {
+      "epoch": 0.766,
+      "learning_rate": 1.9735789028731607e-05,
+      "loss": 0.2591,
+      "step": 3830
+    },
+    {
+      "epoch": 0.7664,
+      "learning_rate": 1.9732591159931567e-05,
+      "loss": 0.7535,
+      "step": 3832
+    },
+    {
+      "epoch": 0.7668,
+      "learning_rate": 1.9729374316947037e-05,
+      "loss": 0.3732,
+      "step": 3834
+    },
+    {
+      "epoch": 0.7672,
+      "learning_rate": 1.972613850604944e-05,
+      "loss": 0.2653,
+      "step": 3836
+    },
+    {
+      "epoch": 0.7676,
+      "learning_rate": 1.972288373354713e-05,
+      "loss": 0.337,
+      "step": 3838
+    },
+    {
+      "epoch": 0.768,
+      "learning_rate": 1.9719610005785463e-05,
+      "loss": 0.515,
+      "step": 3840
+    },
+    {
+      "epoch": 0.7684,
+      "learning_rate": 1.9716317329146743e-05,
+      "loss": 0.3201,
+      "step": 3842
+    },
+    {
+      "epoch": 0.7688,
+      "learning_rate": 1.9713005710050206e-05,
+      "loss": 0.4186,
+      "step": 3844
+    },
+    {
+      "epoch": 0.7692,
+      "learning_rate": 1.9709675154952017e-05,
+      "loss": 0.4841,
+      "step": 3846
+    },
+    {
+      "epoch": 0.7696,
+      "learning_rate": 1.9706325670345276e-05,
+      "loss": 0.2801,
+      "step": 3848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.970295726275997e-05,
+      "loss": 0.1944,
+      "step": 3850
+    },
+    {
+      "epoch": 0.7704,
+      "learning_rate": 1.9699569938762975e-05,
+      "loss": 0.2437,
+      "step": 3852
+    },
+    {
+      "epoch": 0.7708,
+      "learning_rate": 1.969616370495806e-05,
+      "loss": 0.3189,
+      "step": 3854
+    },
+    {
+      "epoch": 0.7712,
+      "learning_rate": 1.969273856798586e-05,
+      "loss": 0.4711,
+      "step": 3856
+    },
+    {
+      "epoch": 0.7716,
+      "learning_rate": 1.9689294534523836e-05,
+      "loss": 0.4434,
+      "step": 3858
+    },
+    {
+      "epoch": 0.772,
+      "learning_rate": 1.9685831611286312e-05,
+      "loss": 0.0725,
+      "step": 3860
+    },
+    {
+      "epoch": 0.7724,
+      "learning_rate": 1.9682349805024447e-05,
+      "loss": 0.0958,
+      "step": 3862
+    },
+    {
+      "epoch": 0.7728,
+      "learning_rate": 1.9678849122526195e-05,
+      "loss": 0.7279,
+      "step": 3864
+    },
+    {
+      "epoch": 0.7732,
+      "learning_rate": 1.9675329570616302e-05,
+      "loss": 0.1513,
+      "step": 3866
+    },
+    {
+      "epoch": 0.7736,
+      "learning_rate": 1.967179115615633e-05,
+      "loss": 0.8891,
+      "step": 3868
+    },
+    {
+      "epoch": 0.774,
+      "learning_rate": 1.966823388604459e-05,
+      "loss": 0.4914,
+      "step": 3870
+    },
+    {
+      "epoch": 0.7744,
+      "learning_rate": 1.966465776721618e-05,
+      "loss": 1.0113,
+      "step": 3872
+    },
+    {
+      "epoch": 0.7748,
+      "learning_rate": 1.9661062806642906e-05,
+      "loss": 0.1897,
+      "step": 3874
+    },
+    {
+      "epoch": 0.7752,
+      "learning_rate": 1.9657449011333328e-05,
+      "loss": 0.8487,
+      "step": 3876
+    },
+    {
+      "epoch": 0.7756,
+      "learning_rate": 1.9653816388332743e-05,
+      "loss": 0.3197,
+      "step": 3878
+    },
+    {
+      "epoch": 0.776,
+      "learning_rate": 1.965016494472312e-05,
+      "loss": 0.3202,
+      "step": 3880
+    },
+    {
+      "epoch": 0.7764,
+      "learning_rate": 1.964649468762313e-05,
+      "loss": 0.6519,
+      "step": 3882
+    },
+    {
+      "epoch": 0.7768,
+      "learning_rate": 1.964280562418815e-05,
+      "loss": 0.462,
+      "step": 3884
+    },
+    {
+      "epoch": 0.7772,
+      "learning_rate": 1.963909776161018e-05,
+      "loss": 0.2097,
+      "step": 3886
+    },
+    {
+      "epoch": 0.7776,
+      "learning_rate": 1.963537110711789e-05,
+      "loss": 0.1438,
+      "step": 3888
+    },
+    {
+      "epoch": 0.778,
+      "learning_rate": 1.9631625667976584e-05,
+      "loss": 0.2072,
+      "step": 3890
+    },
+    {
+      "epoch": 0.7784,
+      "learning_rate": 1.9627861451488194e-05,
+      "loss": 0.4208,
+      "step": 3892
+    },
+    {
+      "epoch": 0.7788,
+      "learning_rate": 1.9624078464991246e-05,
+      "loss": 0.2932,
+      "step": 3894
+    },
+    {
+      "epoch": 0.7792,
+      "learning_rate": 1.962027671586086e-05,
+      "loss": 0.3947,
+      "step": 3896
+    },
+    {
+      "epoch": 0.7796,
+      "learning_rate": 1.9616456211508756e-05,
+      "loss": 0.6846,
+      "step": 3898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9612616959383194e-05,
+      "loss": 0.2149,
+      "step": 3900
+    },
+    {
+      "epoch": 0.7804,
+      "learning_rate": 1.9608758966968987e-05,
+      "loss": 0.3504,
+      "step": 3902
+    },
+    {
+      "epoch": 0.7808,
+      "learning_rate": 1.96048822417875e-05,
+      "loss": 0.8159,
+      "step": 3904
+    },
+    {
+      "epoch": 0.7812,
+      "learning_rate": 1.9600986791396597e-05,
+      "loss": 0.5325,
+      "step": 3906
+    },
+    {
+      "epoch": 0.7816,
+      "learning_rate": 1.9597072623390668e-05,
+      "loss": 1.0138,
+      "step": 3908
+    },
+    {
+      "epoch": 0.782,
+      "learning_rate": 1.9593139745400578e-05,
+      "loss": 0.1415,
+      "step": 3910
+    },
+    {
+      "epoch": 0.7824,
+      "learning_rate": 1.9589188165093666e-05,
+      "loss": 0.315,
+      "step": 3912
+    },
+    {
+      "epoch": 0.7828,
+      "learning_rate": 1.9585217890173765e-05,
+      "loss": 0.5427,
+      "step": 3914
+    },
+    {
+      "epoch": 0.7832,
+      "learning_rate": 1.95812289283811e-05,
+      "loss": 0.4662,
+      "step": 3916
+    },
+    {
+      "epoch": 0.7836,
+      "learning_rate": 1.957722128749237e-05,
+      "loss": 0.1832,
+      "step": 3918
+    },
+    {
+      "epoch": 0.784,
+      "learning_rate": 1.9573194975320672e-05,
+      "loss": 0.4523,
+      "step": 3920
+    },
+    {
+      "epoch": 0.7844,
+      "learning_rate": 1.9569149999715518e-05,
+      "loss": 0.268,
+      "step": 3922
+    },
+    {
+      "epoch": 0.7848,
+      "learning_rate": 1.9565086368562784e-05,
+      "loss": 0.6059,
+      "step": 3924
+    },
+    {
+      "epoch": 0.7852,
+      "learning_rate": 1.9561004089784722e-05,
+      "loss": 0.2079,
+      "step": 3926
+    },
+    {
+      "epoch": 0.7856,
+      "learning_rate": 1.9556903171339966e-05,
+      "loss": 0.2928,
+      "step": 3928
+    },
+    {
+      "epoch": 0.786,
+      "learning_rate": 1.955278362122344e-05,
+      "loss": 0.4607,
+      "step": 3930
+    },
+    {
+      "epoch": 0.7864,
+      "learning_rate": 1.954864544746643e-05,
+      "loss": 0.4708,
+      "step": 3932
+    },
+    {
+      "epoch": 0.7868,
+      "learning_rate": 1.954448865813652e-05,
+      "loss": 0.1758,
+      "step": 3934
+    },
+    {
+      "epoch": 0.7872,
+      "learning_rate": 1.9540313261337585e-05,
+      "loss": 0.2164,
+      "step": 3936
+    },
+    {
+      "epoch": 0.7876,
+      "learning_rate": 1.9536119265209763e-05,
+      "loss": 0.3027,
+      "step": 3938
+    },
+    {
+      "epoch": 0.788,
+      "learning_rate": 1.9531906677929472e-05,
+      "loss": 0.5666,
+      "step": 3940
+    },
+    {
+      "epoch": 0.7884,
+      "learning_rate": 1.9527675507709364e-05,
+      "loss": 0.3472,
+      "step": 3942
+    },
+    {
+      "epoch": 0.7888,
+      "learning_rate": 1.9523425762798335e-05,
+      "loss": 0.1398,
+      "step": 3944
+    },
+    {
+      "epoch": 0.7892,
+      "learning_rate": 1.9519157451481456e-05,
+      "loss": 0.2788,
+      "step": 3946
+    },
+    {
+      "epoch": 0.7896,
+      "learning_rate": 1.9514870582080035e-05,
+      "loss": 0.1883,
+      "step": 3948
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9510565162951545e-05,
+      "loss": 0.265,
+      "step": 3950
+    },
+    {
+      "epoch": 0.7904,
+      "learning_rate": 1.95062412024896e-05,
+      "loss": 0.439,
+      "step": 3952
+    },
+    {
+      "epoch": 0.7908,
+      "learning_rate": 1.950189870912401e-05,
+      "loss": 0.1476,
+      "step": 3954
+    },
+    {
+      "epoch": 0.7912,
+      "learning_rate": 1.9497537691320667e-05,
+      "loss": 0.1598,
+      "step": 3956
+    },
+    {
+      "epoch": 0.7916,
+      "learning_rate": 1.9493158157581617e-05,
+      "loss": 0.245,
+      "step": 3958
+    },
+    {
+      "epoch": 0.792,
+      "learning_rate": 1.948876011644497e-05,
+      "loss": 0.5239,
+      "step": 3960
+    },
+    {
+      "epoch": 0.7924,
+      "learning_rate": 1.948434357648493e-05,
+      "loss": 0.1886,
+      "step": 3962
+    },
+    {
+      "epoch": 0.7928,
+      "learning_rate": 1.9479908546311787e-05,
+      "loss": 0.3543,
+      "step": 3964
+    },
+    {
+      "epoch": 0.7932,
+      "learning_rate": 1.9475455034571843e-05,
+      "loss": 0.1824,
+      "step": 3966
+    },
+    {
+      "epoch": 0.7936,
+      "learning_rate": 1.9470983049947443e-05,
+      "loss": 1.2475,
+      "step": 3968
+    },
+    {
+      "epoch": 0.794,
+      "learning_rate": 1.9466492601156964e-05,
+      "loss": 0.6929,
+      "step": 3970
+    },
+    {
+      "epoch": 0.7944,
+      "learning_rate": 1.9461983696954767e-05,
+      "loss": 0.1836,
+      "step": 3972
+    },
+    {
+      "epoch": 0.7948,
+      "learning_rate": 1.9457456346131175e-05,
+      "loss": 0.5604,
+      "step": 3974
+    },
+    {
+      "epoch": 0.7952,
+      "learning_rate": 1.9452910557512494e-05,
+      "loss": 0.6166,
+      "step": 3976
+    },
+    {
+      "epoch": 0.7956,
+      "learning_rate": 1.9448346339960984e-05,
+      "loss": 0.1246,
+      "step": 3978
+    },
+    {
+      "epoch": 0.796,
+      "learning_rate": 1.9443763702374818e-05,
+      "loss": 0.362,
+      "step": 3980
+    },
+    {
+      "epoch": 0.7964,
+      "learning_rate": 1.9439162653688066e-05,
+      "loss": 0.5638,
+      "step": 3982
+    },
+    {
+      "epoch": 0.7968,
+      "learning_rate": 1.9434543202870726e-05,
+      "loss": 0.4297,
+      "step": 3984
+    },
+    {
+      "epoch": 0.7972,
+      "learning_rate": 1.9429905358928655e-05,
+      "loss": 0.7007,
+      "step": 3986
+    },
+    {
+      "epoch": 0.7976,
+      "learning_rate": 1.9425249130903544e-05,
+      "loss": 1.2002,
+      "step": 3988
+    },
+    {
+      "epoch": 0.798,
+      "learning_rate": 1.942057452787297e-05,
+      "loss": 0.7052,
+      "step": 3990
+    },
+    {
+      "epoch": 0.7984,
+      "learning_rate": 1.94158815589503e-05,
+      "loss": 0.4422,
+      "step": 3992
+    },
+    {
+      "epoch": 0.7988,
+      "learning_rate": 1.941117023328473e-05,
+      "loss": 0.4989,
+      "step": 3994
+    },
+    {
+      "epoch": 0.7992,
+      "learning_rate": 1.940644056006122e-05,
+      "loss": 0.1465,
+      "step": 3996
+    },
+    {
+      "epoch": 0.7996,
+      "learning_rate": 1.94016925485005e-05,
+      "loss": 0.2389,
+      "step": 3998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.939692620785909e-05,
+      "loss": 0.2036,
+      "step": 4000
+    },
+    {
+      "epoch": 0.8004,
+      "learning_rate": 1.939214154742919e-05,
+      "loss": 0.1124,
+      "step": 4002
+    },
+    {
+      "epoch": 0.8008,
+      "learning_rate": 1.9387338576538746e-05,
+      "loss": 0.3612,
+      "step": 4004
+    },
+    {
+      "epoch": 0.8012,
+      "learning_rate": 1.9382517304551393e-05,
+      "loss": 0.7357,
+      "step": 4006
+    },
+    {
+      "epoch": 0.8016,
+      "learning_rate": 1.9377677740866464e-05,
+      "loss": 0.1286,
+      "step": 4008
+    },
+    {
+      "epoch": 0.802,
+      "learning_rate": 1.9372819894918922e-05,
+      "loss": 1.0977,
+      "step": 4010
+    },
+    {
+      "epoch": 0.8024,
+      "learning_rate": 1.936794377617938e-05,
+      "loss": 0.2182,
+      "step": 4012
+    },
+    {
+      "epoch": 0.8028,
+      "learning_rate": 1.9363049394154102e-05,
+      "loss": 0.1257,
+      "step": 4014
+    },
+    {
+      "epoch": 0.8032,
+      "learning_rate": 1.9358136758384917e-05,
+      "loss": 0.3986,
+      "step": 4016
+    },
+    {
+      "epoch": 0.8036,
+      "learning_rate": 1.935320587844926e-05,
+      "loss": 0.642,
+      "step": 4018
+    },
+    {
+      "epoch": 0.804,
+      "learning_rate": 1.9348256763960146e-05,
+      "loss": 0.2709,
+      "step": 4020
+    },
+    {
+      "epoch": 0.8044,
+      "learning_rate": 1.934328942456613e-05,
+      "loss": 0.1656,
+      "step": 4022
+    },
+    {
+      "epoch": 0.8048,
+      "learning_rate": 1.9338303869951273e-05,
+      "loss": 0.3589,
+      "step": 4024
+    },
+    {
+      "epoch": 0.8052,
+      "learning_rate": 1.9333300109835186e-05,
+      "loss": 0.5024,
+      "step": 4026
+    },
+    {
+      "epoch": 0.8056,
+      "learning_rate": 1.9328278153972943e-05,
+      "loss": 0.3645,
+      "step": 4028
+    },
+    {
+      "epoch": 0.806,
+      "learning_rate": 1.9323238012155125e-05,
+      "loss": 0.4799,
+      "step": 4030
+    },
+    {
+      "epoch": 0.8064,
+      "learning_rate": 1.931817969420773e-05,
+      "loss": 0.2426,
+      "step": 4032
+    },
+    {
+      "epoch": 0.8068,
+      "learning_rate": 1.93131032099922e-05,
+      "loss": 0.3426,
+      "step": 4034
+    },
+    {
+      "epoch": 0.8072,
+      "learning_rate": 1.930800856940543e-05,
+      "loss": 0.1273,
+      "step": 4036
+    },
+    {
+      "epoch": 0.8076,
+      "learning_rate": 1.9302895782379648e-05,
+      "loss": 0.432,
+      "step": 4038
+    },
+    {
+      "epoch": 0.808,
+      "learning_rate": 1.929776485888252e-05,
+      "loss": 0.1161,
+      "step": 4040
+    },
+    {
+      "epoch": 0.8084,
+      "learning_rate": 1.9292615808917024e-05,
+      "loss": 0.416,
+      "step": 4042
+    },
+    {
+      "epoch": 0.8088,
+      "learning_rate": 1.9287448642521517e-05,
+      "loss": 0.2371,
+      "step": 4044
+    },
+    {
+      "epoch": 0.8092,
+      "learning_rate": 1.9282263369769637e-05,
+      "loss": 0.516,
+      "step": 4046
+    },
+    {
+      "epoch": 0.8096,
+      "learning_rate": 1.9277060000770342e-05,
+      "loss": 0.6714,
+      "step": 4048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.927183854566788e-05,
+      "loss": 0.2919,
+      "step": 4050
+    },
+    {
+      "epoch": 0.8104,
+      "learning_rate": 1.9266599014641727e-05,
+      "loss": 0.066,
+      "step": 4052
+    },
+    {
+      "epoch": 0.8108,
+      "learning_rate": 1.9261341417906622e-05,
+      "loss": 0.4084,
+      "step": 4054
+    },
+    {
+      "epoch": 0.8112,
+      "learning_rate": 1.925606576571252e-05,
+      "loss": 0.4662,
+      "step": 4056
+    },
+    {
+      "epoch": 0.8116,
+      "learning_rate": 1.925077206834459e-05,
+      "loss": 0.268,
+      "step": 4058
+    },
+    {
+      "epoch": 0.812,
+      "learning_rate": 1.9245460336123136e-05,
+      "loss": 0.1871,
+      "step": 4060
+    },
+    {
+      "epoch": 0.8124,
+      "learning_rate": 1.924013057940367e-05,
+      "loss": 0.3816,
+      "step": 4062
+    },
+    {
+      "epoch": 0.8128,
+      "learning_rate": 1.923478280857682e-05,
+      "loss": 0.3537,
+      "step": 4064
+    },
+    {
+      "epoch": 0.8132,
+      "learning_rate": 1.922941703406836e-05,
+      "loss": 0.4385,
+      "step": 4066
+    },
+    {
+      "epoch": 0.8136,
+      "learning_rate": 1.9224033266339103e-05,
+      "loss": 0.1177,
+      "step": 4068
+    },
+    {
+      "epoch": 0.814,
+      "learning_rate": 1.9218631515885007e-05,
+      "loss": 0.3492,
+      "step": 4070
+    },
+    {
+      "epoch": 0.8144,
+      "learning_rate": 1.9213211793237066e-05,
+      "loss": 0.2725,
+      "step": 4072
+    },
+    {
+      "epoch": 0.8148,
+      "learning_rate": 1.9207774108961276e-05,
+      "loss": 0.1472,
+      "step": 4074
+    },
+    {
+      "epoch": 0.8152,
+      "learning_rate": 1.9202318473658707e-05,
+      "loss": 0.1998,
+      "step": 4076
+    },
+    {
+      "epoch": 0.8156,
+      "learning_rate": 1.9196844897965387e-05,
+      "loss": 0.2372,
+      "step": 4078
+    },
+    {
+      "epoch": 0.816,
+      "learning_rate": 1.919135339255235e-05,
+      "loss": 0.194,
+      "step": 4080
+    },
+    {
+      "epoch": 0.8164,
+      "learning_rate": 1.9185843968125546e-05,
+      "loss": 0.2637,
+      "step": 4082
+    },
+    {
+      "epoch": 0.8168,
+      "learning_rate": 1.918031663542588e-05,
+      "loss": 0.7861,
+      "step": 4084
+    },
+    {
+      "epoch": 0.8172,
+      "learning_rate": 1.917477140522919e-05,
+      "loss": 0.5257,
+      "step": 4086
+    },
+    {
+      "epoch": 0.8176,
+      "learning_rate": 1.916920828834617e-05,
+      "loss": 0.3554,
+      "step": 4088
+    },
+    {
+      "epoch": 0.818,
+      "learning_rate": 1.9163627295622394e-05,
+      "loss": 0.1994,
+      "step": 4090
+    },
+    {
+      "epoch": 0.8184,
+      "learning_rate": 1.9158028437938313e-05,
+      "loss": 0.6129,
+      "step": 4092
+    },
+    {
+      "epoch": 0.8188,
+      "learning_rate": 1.9152411726209183e-05,
+      "loss": 0.5995,
+      "step": 4094
+    },
+    {
+      "epoch": 0.8192,
+      "learning_rate": 1.9146777171385057e-05,
+      "loss": 0.3436,
+      "step": 4096
+    },
+    {
+      "epoch": 0.8196,
+      "learning_rate": 1.914112478445079e-05,
+      "loss": 0.6085,
+      "step": 4098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.913545457642601e-05,
+      "loss": 0.2151,
+      "step": 4100
+    },
+    {
+      "epoch": 0.8204,
+      "learning_rate": 1.9129766558365082e-05,
+      "loss": 0.368,
+      "step": 4102
+    },
+    {
+      "epoch": 0.8208,
+      "learning_rate": 1.9124060741357065e-05,
+      "loss": 0.3586,
+      "step": 4104
+    },
+    {
+      "epoch": 0.8212,
+      "learning_rate": 1.911833713652576e-05,
+      "loss": 0.7349,
+      "step": 4106
+    },
+    {
+      "epoch": 0.8216,
+      "learning_rate": 1.911259575502963e-05,
+      "loss": 0.6216,
+      "step": 4108
+    },
+    {
+      "epoch": 0.822,
+      "learning_rate": 1.9106836608061775e-05,
+      "loss": 0.7198,
+      "step": 4110
+    },
+    {
+      "epoch": 0.8224,
+      "learning_rate": 1.910105970684996e-05,
+      "loss": 0.5756,
+      "step": 4112
+    },
+    {
+      "epoch": 0.8228,
+      "learning_rate": 1.909526506265654e-05,
+      "loss": 0.3778,
+      "step": 4114
+    },
+    {
+      "epoch": 0.8232,
+      "learning_rate": 1.908945268677849e-05,
+      "loss": 0.2778,
+      "step": 4116
+    },
+    {
+      "epoch": 0.8236,
+      "learning_rate": 1.9083622590547313e-05,
+      "loss": 0.2248,
+      "step": 4118
+    },
+    {
+      "epoch": 0.824,
+      "learning_rate": 1.9077774785329085e-05,
+      "loss": 0.6537,
+      "step": 4120
+    },
+    {
+      "epoch": 0.8244,
+      "learning_rate": 1.9071909282524422e-05,
+      "loss": 0.5248,
+      "step": 4122
+    },
+    {
+      "epoch": 0.8248,
+      "learning_rate": 1.9066026093568383e-05,
+      "loss": 0.4543,
+      "step": 4124
+    },
+    {
+      "epoch": 0.8252,
+      "learning_rate": 1.9060125229930576e-05,
+      "loss": 0.2578,
+      "step": 4126
+    },
+    {
+      "epoch": 0.8256,
+      "learning_rate": 1.9054206703115013e-05,
+      "loss": 0.5293,
+      "step": 4128
+    },
+    {
+      "epoch": 0.826,
+      "learning_rate": 1.9048270524660203e-05,
+      "loss": 0.1624,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8264,
+      "learning_rate": 1.9042316706138994e-05,
+      "loss": 0.2403,
+      "step": 4132
+    },
+    {
+      "epoch": 0.8268,
+      "learning_rate": 1.9036345259158664e-05,
+      "loss": 0.188,
+      "step": 4134
+    },
+    {
+      "epoch": 0.8272,
+      "learning_rate": 1.903035619536087e-05,
+      "loss": 0.3277,
+      "step": 4136
+    },
+    {
+      "epoch": 0.8276,
+      "learning_rate": 1.9024349526421603e-05,
+      "loss": 0.3697,
+      "step": 4138
+    },
+    {
+      "epoch": 0.828,
+      "learning_rate": 1.901832526405114e-05,
+      "loss": 0.6074,
+      "step": 4140
+    },
+    {
+      "epoch": 0.8284,
+      "learning_rate": 1.9012283419994112e-05,
+      "loss": 0.5585,
+      "step": 4142
+    },
+    {
+      "epoch": 0.8288,
+      "learning_rate": 1.9006224006029414e-05,
+      "loss": 0.2183,
+      "step": 4144
+    },
+    {
+      "epoch": 0.8292,
+      "learning_rate": 1.9000147033970148e-05,
+      "loss": 0.1998,
+      "step": 4146
+    },
+    {
+      "epoch": 0.8296,
+      "learning_rate": 1.899405251566371e-05,
+      "loss": 0.4712,
+      "step": 4148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8987940462991666e-05,
+      "loss": 0.7746,
+      "step": 4150
+    },
+    {
+      "epoch": 0.8304,
+      "learning_rate": 1.8981810887869797e-05,
+      "loss": 0.7434,
+      "step": 4152
+    },
+    {
+      "epoch": 0.8308,
+      "learning_rate": 1.8975663802247978e-05,
+      "loss": 0.3145,
+      "step": 4154
+    },
+    {
+      "epoch": 0.8312,
+      "learning_rate": 1.8969499218110302e-05,
+      "loss": 0.3599,
+      "step": 4156
+    },
+    {
+      "epoch": 0.8316,
+      "learning_rate": 1.8963317147474943e-05,
+      "loss": 0.3584,
+      "step": 4158
+    },
+    {
+      "epoch": 0.832,
+      "learning_rate": 1.8957117602394133e-05,
+      "loss": 0.3001,
+      "step": 4160
+    },
+    {
+      "epoch": 0.8324,
+      "learning_rate": 1.8950900594954233e-05,
+      "loss": 0.4026,
+      "step": 4162
+    },
+    {
+      "epoch": 0.8328,
+      "learning_rate": 1.8944666137275596e-05,
+      "loss": 0.4066,
+      "step": 4164
+    },
+    {
+      "epoch": 0.8332,
+      "learning_rate": 1.8938414241512644e-05,
+      "loss": 0.4274,
+      "step": 4166
+    },
+    {
+      "epoch": 0.8336,
+      "learning_rate": 1.8932144919853744e-05,
+      "loss": 0.1486,
+      "step": 4168
+    },
+    {
+      "epoch": 0.834,
+      "learning_rate": 1.892585818452125e-05,
+      "loss": 0.1694,
+      "step": 4170
+    },
+    {
+      "epoch": 0.8344,
+      "learning_rate": 1.8919554047771508e-05,
+      "loss": 0.4472,
+      "step": 4172
+    },
+    {
+      "epoch": 0.8348,
+      "learning_rate": 1.891323252189474e-05,
+      "loss": 0.4597,
+      "step": 4174
+    },
+    {
+      "epoch": 0.8352,
+      "learning_rate": 1.890689361921507e-05,
+      "loss": 0.2891,
+      "step": 4176
+    },
+    {
+      "epoch": 0.8356,
+      "learning_rate": 1.8900537352090523e-05,
+      "loss": 0.131,
+      "step": 4178
+    },
+    {
+      "epoch": 0.836,
+      "learning_rate": 1.8894163732912986e-05,
+      "loss": 0.7281,
+      "step": 4180
+    },
+    {
+      "epoch": 0.8364,
+      "learning_rate": 1.8887772774108122e-05,
+      "loss": 0.2622,
+      "step": 4182
+    },
+    {
+      "epoch": 0.8368,
+      "learning_rate": 1.8881364488135445e-05,
+      "loss": 0.7192,
+      "step": 4184
+    },
+    {
+      "epoch": 0.8372,
+      "learning_rate": 1.887493888748825e-05,
+      "loss": 0.3687,
+      "step": 4186
+    },
+    {
+      "epoch": 0.8376,
+      "learning_rate": 1.886849598469357e-05,
+      "loss": 0.5268,
+      "step": 4188
+    },
+    {
+      "epoch": 0.838,
+      "learning_rate": 1.886203579231215e-05,
+      "loss": 0.3431,
+      "step": 4190
+    },
+    {
+      "epoch": 0.8384,
+      "learning_rate": 1.8855558322938492e-05,
+      "loss": 1.0116,
+      "step": 4192
+    },
+    {
+      "epoch": 0.8388,
+      "learning_rate": 1.8849063589200754e-05,
+      "loss": 0.5597,
+      "step": 4194
+    },
+    {
+      "epoch": 0.8392,
+      "learning_rate": 1.8842551603760725e-05,
+      "loss": 0.5993,
+      "step": 4196
+    },
+    {
+      "epoch": 0.8396,
+      "learning_rate": 1.8836022379313884e-05,
+      "loss": 0.3999,
+      "step": 4198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8829475928589265e-05,
+      "loss": 0.2387,
+      "step": 4200
+    },
+    {
+      "epoch": 0.8404,
+      "learning_rate": 1.882291226434954e-05,
+      "loss": 0.4045,
+      "step": 4202
+    },
+    {
+      "epoch": 0.8408,
+      "learning_rate": 1.8816331399390874e-05,
+      "loss": 0.3815,
+      "step": 4204
+    },
+    {
+      "epoch": 0.8412,
+      "learning_rate": 1.880973334654301e-05,
+      "loss": 0.2008,
+      "step": 4206
+    },
+    {
+      "epoch": 0.8416,
+      "learning_rate": 1.88031181186692e-05,
+      "loss": 0.2996,
+      "step": 4208
+    },
+    {
+      "epoch": 0.842,
+      "learning_rate": 1.8796485728666172e-05,
+      "loss": 0.3003,
+      "step": 4210
+    },
+    {
+      "epoch": 0.8424,
+      "learning_rate": 1.8789836189464092e-05,
+      "loss": 0.0456,
+      "step": 4212
+    },
+    {
+      "epoch": 0.8428,
+      "learning_rate": 1.8783169514026574e-05,
+      "loss": 0.3398,
+      "step": 4214
+    },
+    {
+      "epoch": 0.8432,
+      "learning_rate": 1.877648571535068e-05,
+      "loss": 0.3103,
+      "step": 4216
+    },
+    {
+      "epoch": 0.8436,
+      "learning_rate": 1.8769784806466775e-05,
+      "loss": 0.5355,
+      "step": 4218
+    },
+    {
+      "epoch": 0.844,
+      "learning_rate": 1.8763066800438638e-05,
+      "loss": 0.577,
+      "step": 4220
+    },
+    {
+      "epoch": 0.8444,
+      "learning_rate": 1.8756331710363375e-05,
+      "loss": 0.0266,
+      "step": 4222
+    },
+    {
+      "epoch": 0.8448,
+      "learning_rate": 1.8749579549371387e-05,
+      "loss": 0.7541,
+      "step": 4224
+    },
+    {
+      "epoch": 0.8452,
+      "learning_rate": 1.8742810330626338e-05,
+      "loss": 0.3709,
+      "step": 4226
+    },
+    {
+      "epoch": 0.8456,
+      "learning_rate": 1.8736024067325188e-05,
+      "loss": 0.0728,
+      "step": 4228
+    },
+    {
+      "epoch": 0.846,
+      "learning_rate": 1.8729220772698106e-05,
+      "loss": 0.1955,
+      "step": 4230
+    },
+    {
+      "epoch": 0.8464,
+      "learning_rate": 1.8722400460008437e-05,
+      "loss": 0.1702,
+      "step": 4232
+    },
+    {
+      "epoch": 0.8468,
+      "learning_rate": 1.8715563142552758e-05,
+      "loss": 0.2511,
+      "step": 4234
+    },
+    {
+      "epoch": 0.8472,
+      "learning_rate": 1.8708708833660748e-05,
+      "loss": 0.5291,
+      "step": 4236
+    },
+    {
+      "epoch": 0.8476,
+      "learning_rate": 1.870183754669526e-05,
+      "loss": 0.5172,
+      "step": 4238
+    },
+    {
+      "epoch": 0.848,
+      "learning_rate": 1.8694949295052198e-05,
+      "loss": 0.3539,
+      "step": 4240
+    },
+    {
+      "epoch": 0.8484,
+      "learning_rate": 1.8688044092160558e-05,
+      "loss": 0.2288,
+      "step": 4242
+    },
+    {
+      "epoch": 0.8488,
+      "learning_rate": 1.868112195148239e-05,
+      "loss": 0.1428,
+      "step": 4244
+    },
+    {
+      "epoch": 0.8492,
+      "learning_rate": 1.867418288651278e-05,
+      "loss": 0.259,
+      "step": 4246
+    },
+    {
+      "epoch": 0.8496,
+      "learning_rate": 1.866722691077977e-05,
+      "loss": 0.7427,
+      "step": 4248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8660254037844384e-05,
+      "loss": 0.2486,
+      "step": 4250
+    },
+    {
+      "epoch": 0.8504,
+      "learning_rate": 1.8653264281300626e-05,
+      "loss": 0.3036,
+      "step": 4252
+    },
+    {
+      "epoch": 0.8508,
+      "learning_rate": 1.8646257654775357e-05,
+      "loss": 0.0994,
+      "step": 4254
+    },
+    {
+      "epoch": 0.8512,
+      "learning_rate": 1.8639234171928355e-05,
+      "loss": 0.3369,
+      "step": 4256
+    },
+    {
+      "epoch": 0.8516,
+      "learning_rate": 1.8632193846452267e-05,
+      "loss": 0.3736,
+      "step": 4258
+    },
+    {
+      "epoch": 0.852,
+      "learning_rate": 1.8625136692072587e-05,
+      "loss": 0.2933,
+      "step": 4260
+    },
+    {
+      "epoch": 0.8524,
+      "learning_rate": 1.861806272254755e-05,
+      "loss": 0.195,
+      "step": 4262
+    },
+    {
+      "epoch": 0.8528,
+      "learning_rate": 1.8610971951668265e-05,
+      "loss": 0.1631,
+      "step": 4264
+    },
+    {
+      "epoch": 0.8532,
+      "learning_rate": 1.8603864393258547e-05,
+      "loss": 0.2164,
+      "step": 4266
+    },
+    {
+      "epoch": 0.8536,
+      "learning_rate": 1.8596740061174912e-05,
+      "loss": 0.3692,
+      "step": 4268
+    },
+    {
+      "epoch": 0.854,
+      "learning_rate": 1.8589598969306646e-05,
+      "loss": 0.6011,
+      "step": 4270
+    },
+    {
+      "epoch": 0.8544,
+      "learning_rate": 1.858244113157566e-05,
+      "loss": 0.166,
+      "step": 4272
+    },
+    {
+      "epoch": 0.8548,
+      "learning_rate": 1.8575266561936533e-05,
+      "loss": 0.2227,
+      "step": 4274
+    },
+    {
+      "epoch": 0.8552,
+      "learning_rate": 1.8568075274376432e-05,
+      "loss": 0.2524,
+      "step": 4276
+    },
+    {
+      "epoch": 0.8556,
+      "learning_rate": 1.8560867282915164e-05,
+      "loss": 0.3329,
+      "step": 4278
+    },
+    {
+      "epoch": 0.856,
+      "learning_rate": 1.8553642601605083e-05,
+      "loss": 0.7938,
+      "step": 4280
+    },
+    {
+      "epoch": 0.8564,
+      "learning_rate": 1.8546401244531034e-05,
+      "loss": 0.1912,
+      "step": 4282
+    },
+    {
+      "epoch": 0.8568,
+      "learning_rate": 1.8539143225810457e-05,
+      "loss": 0.1686,
+      "step": 4284
+    },
+    {
+      "epoch": 0.8572,
+      "learning_rate": 1.85318685595932e-05,
+      "loss": 0.2463,
+      "step": 4286
+    },
+    {
+      "epoch": 0.8576,
+      "learning_rate": 1.852457726006163e-05,
+      "loss": 0.2364,
+      "step": 4288
+    },
+    {
+      "epoch": 0.858,
+      "learning_rate": 1.8517269341430485e-05,
+      "loss": 0.3476,
+      "step": 4290
+    },
+    {
+      "epoch": 0.8584,
+      "learning_rate": 1.8509944817946917e-05,
+      "loss": 0.1867,
+      "step": 4292
+    },
+    {
+      "epoch": 0.8588,
+      "learning_rate": 1.8502603703890484e-05,
+      "loss": 0.1037,
+      "step": 4294
+    },
+    {
+      "epoch": 0.8592,
+      "learning_rate": 1.8495246013573064e-05,
+      "loss": 0.4552,
+      "step": 4296
+    },
+    {
+      "epoch": 0.8596,
+      "learning_rate": 1.8487871761338817e-05,
+      "loss": 0.2835,
+      "step": 4298
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.848048096156426e-05,
+      "loss": 0.8144,
+      "step": 4300
+    },
+    {
+      "epoch": 0.8604,
+      "learning_rate": 1.847307362865813e-05,
+      "loss": 0.464,
+      "step": 4302
+    },
+    {
+      "epoch": 0.8608,
+      "learning_rate": 1.8465649777061387e-05,
+      "loss": 0.4534,
+      "step": 4304
+    },
+    {
+      "epoch": 0.8612,
+      "learning_rate": 1.8458209421247208e-05,
+      "loss": 0.3222,
+      "step": 4306
+    },
+    {
+      "epoch": 0.8616,
+      "learning_rate": 1.8450752575720967e-05,
+      "loss": 1.3721,
+      "step": 4308
+    },
+    {
+      "epoch": 0.862,
+      "learning_rate": 1.8443279255020163e-05,
+      "loss": 0.5144,
+      "step": 4310
+    },
+    {
+      "epoch": 0.8624,
+      "learning_rate": 1.843578947371439e-05,
+      "loss": 0.2906,
+      "step": 4312
+    },
+    {
+      "epoch": 0.8628,
+      "learning_rate": 1.842828324640539e-05,
+      "loss": 0.2659,
+      "step": 4314
+    },
+    {
+      "epoch": 0.8632,
+      "learning_rate": 1.8420760587726935e-05,
+      "loss": 0.1603,
+      "step": 4316
+    },
+    {
+      "epoch": 0.8636,
+      "learning_rate": 1.8413221512344808e-05,
+      "loss": 0.2822,
+      "step": 4318
+    },
+    {
+      "epoch": 0.864,
+      "learning_rate": 1.8405666034956846e-05,
+      "loss": 0.7518,
+      "step": 4320
+    },
+    {
+      "epoch": 0.8644,
+      "learning_rate": 1.8398094170292826e-05,
+      "loss": 0.254,
+      "step": 4322
+    },
+    {
+      "epoch": 0.8648,
+      "learning_rate": 1.8390505933114507e-05,
+      "loss": 0.7464,
+      "step": 4324
+    },
+    {
+      "epoch": 0.8652,
+      "learning_rate": 1.838290133821552e-05,
+      "loss": 0.2612,
+      "step": 4326
+    },
+    {
+      "epoch": 0.8656,
+      "learning_rate": 1.8375280400421414e-05,
+      "loss": 0.3864,
+      "step": 4328
+    },
+    {
+      "epoch": 0.866,
+      "learning_rate": 1.8367643134589613e-05,
+      "loss": 0.842,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8664,
+      "learning_rate": 1.8359989555609365e-05,
+      "loss": 0.3539,
+      "step": 4332
+    },
+    {
+      "epoch": 0.8668,
+      "learning_rate": 1.835231967840168e-05,
+      "loss": 0.1371,
+      "step": 4334
+    },
+    {
+      "epoch": 0.8672,
+      "learning_rate": 1.834463351791939e-05,
+      "loss": 0.2737,
+      "step": 4336
+    },
+    {
+      "epoch": 0.8676,
+      "learning_rate": 1.8336931089147082e-05,
+      "loss": 0.3905,
+      "step": 4338
+    },
+    {
+      "epoch": 0.868,
+      "learning_rate": 1.8329212407101006e-05,
+      "loss": 0.2898,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8684,
+      "learning_rate": 1.8321477486829128e-05,
+      "loss": 0.1239,
+      "step": 4342
+    },
+    {
+      "epoch": 0.8688,
+      "learning_rate": 1.8313726343411085e-05,
+      "loss": 0.3536,
+      "step": 4344
+    },
+    {
+      "epoch": 0.8692,
+      "learning_rate": 1.8305958991958135e-05,
+      "loss": 0.1583,
+      "step": 4346
+    },
+    {
+      "epoch": 0.8696,
+      "learning_rate": 1.82981754476131e-05,
+      "loss": 0.3235,
+      "step": 4348
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8290375725550417e-05,
+      "loss": 0.3145,
+      "step": 4350
+    },
+    {
+      "epoch": 0.8704,
+      "learning_rate": 1.8282559840976053e-05,
+      "loss": 0.2681,
+      "step": 4352
+    },
+    {
+      "epoch": 0.8708,
+      "learning_rate": 1.827472780912744e-05,
+      "loss": 0.3063,
+      "step": 4354
+    },
+    {
+      "epoch": 0.8712,
+      "learning_rate": 1.8266879645273557e-05,
+      "loss": 0.4603,
+      "step": 4356
+    },
+    {
+      "epoch": 0.8716,
+      "learning_rate": 1.825901536471478e-05,
+      "loss": 0.2483,
+      "step": 4358
+    },
+    {
+      "epoch": 0.872,
+      "learning_rate": 1.8251134982782966e-05,
+      "loss": 0.2126,
+      "step": 4360
+    },
+    {
+      "epoch": 0.8724,
+      "learning_rate": 1.824323851484126e-05,
+      "loss": 0.3628,
+      "step": 4362
+    },
+    {
+      "epoch": 0.8728,
+      "learning_rate": 1.823532597628428e-05,
+      "loss": 0.2192,
+      "step": 4364
+    },
+    {
+      "epoch": 0.8732,
+      "learning_rate": 1.8227397382537893e-05,
+      "loss": 0.2212,
+      "step": 4366
+    },
+    {
+      "epoch": 0.8736,
+      "learning_rate": 1.8219452749059336e-05,
+      "loss": 0.3073,
+      "step": 4368
+    },
+    {
+      "epoch": 0.874,
+      "learning_rate": 1.8211492091337048e-05,
+      "loss": 0.5134,
+      "step": 4370
+    },
+    {
+      "epoch": 0.8744,
+      "learning_rate": 1.8203515424890734e-05,
+      "loss": 0.308,
+      "step": 4372
+    },
+    {
+      "epoch": 0.8748,
+      "learning_rate": 1.8195522765271346e-05,
+      "loss": 0.5017,
+      "step": 4374
+    },
+    {
+      "epoch": 0.8752,
+      "learning_rate": 1.8187514128060956e-05,
+      "loss": 0.3202,
+      "step": 4376
+    },
+    {
+      "epoch": 0.8756,
+      "learning_rate": 1.8179489528872804e-05,
+      "loss": 0.216,
+      "step": 4378
+    },
+    {
+      "epoch": 0.876,
+      "learning_rate": 1.8171448983351284e-05,
+      "loss": 0.1125,
+      "step": 4380
+    },
+    {
+      "epoch": 0.8764,
+      "learning_rate": 1.816339250717185e-05,
+      "loss": 0.3441,
+      "step": 4382
+    },
+    {
+      "epoch": 0.8768,
+      "learning_rate": 1.8155320116040983e-05,
+      "loss": 1.0369,
+      "step": 4384
+    },
+    {
+      "epoch": 0.8772,
+      "learning_rate": 1.814723182569625e-05,
+      "loss": 0.142,
+      "step": 4386
+    },
+    {
+      "epoch": 0.8776,
+      "learning_rate": 1.8139127651906193e-05,
+      "loss": 0.0946,
+      "step": 4388
+    },
+    {
+      "epoch": 0.878,
+      "learning_rate": 1.813100761047029e-05,
+      "loss": 0.128,
+      "step": 4390
+    },
+    {
+      "epoch": 0.8784,
+      "learning_rate": 1.8122871717218974e-05,
+      "loss": 0.1914,
+      "step": 4392
+    },
+    {
+      "epoch": 0.8788,
+      "learning_rate": 1.8114719988013612e-05,
+      "loss": 0.2878,
+      "step": 4394
+    },
+    {
+      "epoch": 0.8792,
+      "learning_rate": 1.8106552438746413e-05,
+      "loss": 0.6396,
+      "step": 4396
+    },
+    {
+      "epoch": 0.8796,
+      "learning_rate": 1.8098369085340404e-05,
+      "loss": 0.0462,
+      "step": 4398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 0.3171,
+      "step": 4400
+    },
+    {
+      "epoch": 0.8804,
+      "learning_rate": 1.8081955029958265e-05,
+      "loss": 0.6735,
+      "step": 4402
+    },
+    {
+      "epoch": 0.8808,
+      "learning_rate": 1.807372435998219e-05,
+      "loss": 0.1648,
+      "step": 4404
+    },
+    {
+      "epoch": 0.8812,
+      "learning_rate": 1.806547794986733e-05,
+      "loss": 0.3631,
+      "step": 4406
+    },
+    {
+      "epoch": 0.8816,
+      "learning_rate": 1.8057215815690487e-05,
+      "loss": 0.4089,
+      "step": 4408
+    },
+    {
+      "epoch": 0.882,
+      "learning_rate": 1.8048937973559148e-05,
+      "loss": 0.1935,
+      "step": 4410
+    },
+    {
+      "epoch": 0.8824,
+      "learning_rate": 1.8040644439611355e-05,
+      "loss": 0.3471,
+      "step": 4412
+    },
+    {
+      "epoch": 0.8828,
+      "learning_rate": 1.8032335230015777e-05,
+      "loss": 0.2551,
+      "step": 4414
+    },
+    {
+      "epoch": 0.8832,
+      "learning_rate": 1.8024010360971665e-05,
+      "loss": 0.284,
+      "step": 4416
+    },
+    {
+      "epoch": 0.8836,
+      "learning_rate": 1.8015669848708774e-05,
+      "loss": 0.3208,
+      "step": 4418
+    },
+    {
+      "epoch": 0.884,
+      "learning_rate": 1.8007313709487345e-05,
+      "loss": 0.4543,
+      "step": 4420
+    },
+    {
+      "epoch": 0.8844,
+      "learning_rate": 1.7998941959598093e-05,
+      "loss": 0.1541,
+      "step": 4422
+    },
+    {
+      "epoch": 0.8848,
+      "learning_rate": 1.7990554615362207e-05,
+      "loss": 0.7097,
+      "step": 4424
+    },
+    {
+      "epoch": 0.8852,
+      "learning_rate": 1.7982151693131213e-05,
+      "loss": 0.4324,
+      "step": 4426
+    },
+    {
+      "epoch": 0.8856,
+      "learning_rate": 1.7973733209287036e-05,
+      "loss": 0.7662,
+      "step": 4428
+    },
+    {
+      "epoch": 0.886,
+      "learning_rate": 1.7965299180241963e-05,
+      "loss": 0.4728,
+      "step": 4430
+    },
+    {
+      "epoch": 0.8864,
+      "learning_rate": 1.7956849622438568e-05,
+      "loss": 0.5772,
+      "step": 4432
+    },
+    {
+      "epoch": 0.8868,
+      "learning_rate": 1.794838455234966e-05,
+      "loss": 0.4076,
+      "step": 4434
+    },
+    {
+      "epoch": 0.8872,
+      "learning_rate": 1.7939903986478357e-05,
+      "loss": 0.5115,
+      "step": 4436
+    },
+    {
+      "epoch": 0.8876,
+      "learning_rate": 1.7931407941357938e-05,
+      "loss": 0.2346,
+      "step": 4438
+    },
+    {
+      "epoch": 0.888,
+      "learning_rate": 1.7922896433551913e-05,
+      "loss": 0.2064,
+      "step": 4440
+    },
+    {
+      "epoch": 0.8884,
+      "learning_rate": 1.7914369479653864e-05,
+      "loss": 0.3454,
+      "step": 4442
+    },
+    {
+      "epoch": 0.8888,
+      "learning_rate": 1.7905827096287525e-05,
+      "loss": 0.1642,
+      "step": 4444
+    },
+    {
+      "epoch": 0.8892,
+      "learning_rate": 1.7897269300106752e-05,
+      "loss": 0.5338,
+      "step": 4446
+    },
+    {
+      "epoch": 0.8896,
+      "learning_rate": 1.7888696107795347e-05,
+      "loss": 1.0457,
+      "step": 4448
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.7880107536067228e-05,
+      "loss": 0.111,
+      "step": 4450
+    },
+    {
+      "epoch": 0.8904,
+      "learning_rate": 1.787150360166623e-05,
+      "loss": 0.0891,
+      "step": 4452
+    },
+    {
+      "epoch": 0.8908,
+      "learning_rate": 1.78628843213662e-05,
+      "loss": 0.2385,
+      "step": 4454
+    },
+    {
+      "epoch": 0.8912,
+      "learning_rate": 1.7854249711970826e-05,
+      "loss": 0.0239,
+      "step": 4456
+    },
+    {
+      "epoch": 0.8916,
+      "learning_rate": 1.7845599790313732e-05,
+      "loss": 0.3398,
+      "step": 4458
+    },
+    {
+      "epoch": 0.892,
+      "learning_rate": 1.783693457325841e-05,
+      "loss": 0.2394,
+      "step": 4460
+    },
+    {
+      "epoch": 0.8924,
+      "learning_rate": 1.782825407769811e-05,
+      "loss": 0.2798,
+      "step": 4462
+    },
+    {
+      "epoch": 0.8928,
+      "learning_rate": 1.7819558320555902e-05,
+      "loss": 0.6028,
+      "step": 4464
+    },
+    {
+      "epoch": 0.8932,
+      "learning_rate": 1.7810847318784632e-05,
+      "loss": 0.3387,
+      "step": 4466
+    },
+    {
+      "epoch": 0.8936,
+      "learning_rate": 1.780212108936685e-05,
+      "loss": 0.0928,
+      "step": 4468
+    },
+    {
+      "epoch": 0.894,
+      "learning_rate": 1.7793379649314743e-05,
+      "loss": 0.2647,
+      "step": 4470
+    },
+    {
+      "epoch": 0.8944,
+      "learning_rate": 1.7784623015670237e-05,
+      "loss": 0.2654,
+      "step": 4472
+    },
+    {
+      "epoch": 0.8948,
+      "learning_rate": 1.777585120550481e-05,
+      "loss": 0.6053,
+      "step": 4474
+    },
+    {
+      "epoch": 0.8952,
+      "learning_rate": 1.7767064235919594e-05,
+      "loss": 0.1552,
+      "step": 4476
+    },
+    {
+      "epoch": 0.8956,
+      "learning_rate": 1.77582621240452e-05,
+      "loss": 0.2883,
+      "step": 4478
+    },
+    {
+      "epoch": 0.896,
+      "learning_rate": 1.77494448870418e-05,
+      "loss": 0.4466,
+      "step": 4480
+    },
+    {
+      "epoch": 0.8964,
+      "learning_rate": 1.774061254209907e-05,
+      "loss": 0.5614,
+      "step": 4482
+    },
+    {
+      "epoch": 0.8968,
+      "learning_rate": 1.773176510643608e-05,
+      "loss": 0.7328,
+      "step": 4484
+    },
+    {
+      "epoch": 0.8972,
+      "learning_rate": 1.7722902597301388e-05,
+      "loss": 0.5033,
+      "step": 4486
+    },
+    {
+      "epoch": 0.8976,
+      "learning_rate": 1.7714025031972894e-05,
+      "loss": 0.4507,
+      "step": 4488
+    },
+    {
+      "epoch": 0.898,
+      "learning_rate": 1.77051324277579e-05,
+      "loss": 0.1814,
+      "step": 4490
+    },
+    {
+      "epoch": 0.8984,
+      "learning_rate": 1.769622480199295e-05,
+      "loss": 0.4668,
+      "step": 4492
+    },
+    {
+      "epoch": 0.8988,
+      "learning_rate": 1.7687302172043926e-05,
+      "loss": 0.4958,
+      "step": 4494
+    },
+    {
+      "epoch": 0.8992,
+      "learning_rate": 1.7678364555305982e-05,
+      "loss": 0.4973,
+      "step": 4496
+    },
+    {
+      "epoch": 0.8996,
+      "learning_rate": 1.7669411969203424e-05,
+      "loss": 0.1237,
+      "step": 4498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.7660444431189777e-05,
+      "loss": 0.3068,
+      "step": 4500
+    },
+    {
+      "epoch": 0.9004,
+      "learning_rate": 1.765146195874774e-05,
+      "loss": 0.5162,
+      "step": 4502
+    },
+    {
+      "epoch": 0.9008,
+      "learning_rate": 1.76424645693891e-05,
+      "loss": 0.7352,
+      "step": 4504
+    },
+    {
+      "epoch": 0.9012,
+      "learning_rate": 1.7633452280654696e-05,
+      "loss": 0.294,
+      "step": 4506
+    },
+    {
+      "epoch": 0.9016,
+      "learning_rate": 1.762442511011448e-05,
+      "loss": 0.1496,
+      "step": 4508
+    },
+    {
+      "epoch": 0.902,
+      "learning_rate": 1.761538307536738e-05,
+      "loss": 0.2762,
+      "step": 4510
+    },
+    {
+      "epoch": 0.9024,
+      "learning_rate": 1.7606326194041285e-05,
+      "loss": 0.1972,
+      "step": 4512
+    },
+    {
+      "epoch": 0.9028,
+      "learning_rate": 1.759725448379305e-05,
+      "loss": 0.577,
+      "step": 4514
+    },
+    {
+      "epoch": 0.9032,
+      "learning_rate": 1.7588167962308458e-05,
+      "loss": 0.1347,
+      "step": 4516
+    },
+    {
+      "epoch": 0.9036,
+      "learning_rate": 1.7579066647302147e-05,
+      "loss": 0.3436,
+      "step": 4518
+    },
+    {
+      "epoch": 0.904,
+      "learning_rate": 1.756995055651757e-05,
+      "loss": 0.1125,
+      "step": 4520
+    },
+    {
+      "epoch": 0.9044,
+      "learning_rate": 1.7560819707727037e-05,
+      "loss": 0.1553,
+      "step": 4522
+    },
+    {
+      "epoch": 0.9048,
+      "learning_rate": 1.7551674118731585e-05,
+      "loss": 0.4901,
+      "step": 4524
+    },
+    {
+      "epoch": 0.9052,
+      "learning_rate": 1.7542513807361044e-05,
+      "loss": 0.66,
+      "step": 4526
+    },
+    {
+      "epoch": 0.9056,
+      "learning_rate": 1.7533338791473875e-05,
+      "loss": 0.2167,
+      "step": 4528
+    },
+    {
+      "epoch": 0.906,
+      "learning_rate": 1.7524149088957238e-05,
+      "loss": 0.191,
+      "step": 4530
+    },
+    {
+      "epoch": 0.9064,
+      "learning_rate": 1.751494471772697e-05,
+      "loss": 0.1084,
+      "step": 4532
+    },
+    {
+      "epoch": 0.9068,
+      "learning_rate": 1.750572569572742e-05,
+      "loss": 0.2735,
+      "step": 4534
+    },
+    {
+      "epoch": 0.9072,
+      "learning_rate": 1.7496492040931548e-05,
+      "loss": 0.3394,
+      "step": 4536
+    },
+    {
+      "epoch": 0.9076,
+      "learning_rate": 1.7487243771340862e-05,
+      "loss": 0.56,
+      "step": 4538
+    },
+    {
+      "epoch": 0.908,
+      "learning_rate": 1.747798090498533e-05,
+      "loss": 0.3243,
+      "step": 4540
+    },
+    {
+      "epoch": 0.9084,
+      "learning_rate": 1.7468703459923365e-05,
+      "loss": 0.5008,
+      "step": 4542
+    },
+    {
+      "epoch": 0.9088,
+      "learning_rate": 1.745941145424182e-05,
+      "loss": 0.392,
+      "step": 4544
+    },
+    {
+      "epoch": 0.9092,
+      "learning_rate": 1.7450104906055973e-05,
+      "loss": 0.2317,
+      "step": 4546
+    },
+    {
+      "epoch": 0.9096,
+      "learning_rate": 1.744078383350938e-05,
+      "loss": 0.5204,
+      "step": 4548
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.7431448254773943e-05,
+      "loss": 0.5559,
+      "step": 4550
+    },
+    {
+      "epoch": 0.9104,
+      "learning_rate": 1.7422098188049885e-05,
+      "loss": 0.4116,
+      "step": 4552
+    },
+    {
+      "epoch": 0.9108,
+      "learning_rate": 1.7412733651565624e-05,
+      "loss": 0.2857,
+      "step": 4554
+    },
+    {
+      "epoch": 0.9112,
+      "learning_rate": 1.7403354663577782e-05,
+      "loss": 0.3613,
+      "step": 4556
+    },
+    {
+      "epoch": 0.9116,
+      "learning_rate": 1.739396124237121e-05,
+      "loss": 0.4433,
+      "step": 4558
+    },
+    {
+      "epoch": 0.912,
+      "learning_rate": 1.738455340625883e-05,
+      "loss": 0.2962,
+      "step": 4560
+    },
+    {
+      "epoch": 0.9124,
+      "learning_rate": 1.7375131173581744e-05,
+      "loss": 0.5073,
+      "step": 4562
+    },
+    {
+      "epoch": 0.9128,
+      "learning_rate": 1.7365694562709038e-05,
+      "loss": 0.5834,
+      "step": 4564
+    },
+    {
+      "epoch": 0.9132,
+      "learning_rate": 1.7356243592037865e-05,
+      "loss": 0.419,
+      "step": 4566
+    },
+    {
+      "epoch": 0.9136,
+      "learning_rate": 1.7346778279993433e-05,
+      "loss": 0.1875,
+      "step": 4568
+    },
+    {
+      "epoch": 0.914,
+      "learning_rate": 1.733729864502877e-05,
+      "loss": 0.5513,
+      "step": 4570
+    },
+    {
+      "epoch": 0.9144,
+      "learning_rate": 1.7327804705624962e-05,
+      "loss": 0.306,
+      "step": 4572
+    },
+    {
+      "epoch": 0.9148,
+      "learning_rate": 1.731829648029091e-05,
+      "loss": 0.2081,
+      "step": 4574
+    },
+    {
+      "epoch": 0.9152,
+      "learning_rate": 1.730877398756341e-05,
+      "loss": 0.6129,
+      "step": 4576
+    },
+    {
+      "epoch": 0.9156,
+      "learning_rate": 1.7299237246007025e-05,
+      "loss": 0.1026,
+      "step": 4578
+    },
+    {
+      "epoch": 0.916,
+      "learning_rate": 1.7289686274214113e-05,
+      "loss": 0.2232,
+      "step": 4580
+    },
+    {
+      "epoch": 0.9164,
+      "learning_rate": 1.7280121090804824e-05,
+      "loss": 0.2233,
+      "step": 4582
+    },
+    {
+      "epoch": 0.9168,
+      "learning_rate": 1.727054171442693e-05,
+      "loss": 0.4527,
+      "step": 4584
+    },
+    {
+      "epoch": 0.9172,
+      "learning_rate": 1.7260948163755918e-05,
+      "loss": 0.4862,
+      "step": 4586
+    },
+    {
+      "epoch": 0.9176,
+      "learning_rate": 1.7251340457494934e-05,
+      "loss": 0.5739,
+      "step": 4588
+    },
+    {
+      "epoch": 0.918,
+      "learning_rate": 1.7241718614374688e-05,
+      "loss": 0.2922,
+      "step": 4590
+    },
+    {
+      "epoch": 0.9184,
+      "learning_rate": 1.7232082653153422e-05,
+      "loss": 0.4795,
+      "step": 4592
+    },
+    {
+      "epoch": 0.9188,
+      "learning_rate": 1.722243259261697e-05,
+      "loss": 0.2121,
+      "step": 4594
+    },
+    {
+      "epoch": 0.9192,
+      "learning_rate": 1.7212768451578595e-05,
+      "loss": 0.4462,
+      "step": 4596
+    },
+    {
+      "epoch": 0.9196,
+      "learning_rate": 1.7203090248879084e-05,
+      "loss": 0.5281,
+      "step": 4598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7193398003386517e-05,
+      "loss": 0.445,
+      "step": 4600
+    },
+    {
+      "epoch": 0.9204,
+      "learning_rate": 1.7183691733996463e-05,
+      "loss": 0.094,
+      "step": 4602
+    },
+    {
+      "epoch": 0.9208,
+      "learning_rate": 1.7173971459631803e-05,
+      "loss": 0.2429,
+      "step": 4604
+    },
+    {
+      "epoch": 0.9212,
+      "learning_rate": 1.7164237199242663e-05,
+      "loss": 0.266,
+      "step": 4606
+    },
+    {
+      "epoch": 0.9216,
+      "learning_rate": 1.7154488971806525e-05,
+      "loss": 0.5417,
+      "step": 4608
+    },
+    {
+      "epoch": 0.922,
+      "learning_rate": 1.7144726796328027e-05,
+      "loss": 0.0933,
+      "step": 4610
+    },
+    {
+      "epoch": 0.9224,
+      "learning_rate": 1.713495069183907e-05,
+      "loss": 0.7588,
+      "step": 4612
+    },
+    {
+      "epoch": 0.9228,
+      "learning_rate": 1.7125160677398632e-05,
+      "loss": 0.6549,
+      "step": 4614
+    },
+    {
+      "epoch": 0.9232,
+      "learning_rate": 1.7115356772092847e-05,
+      "loss": 0.4359,
+      "step": 4616
+    },
+    {
+      "epoch": 0.9236,
+      "learning_rate": 1.710553899503497e-05,
+      "loss": 0.2479,
+      "step": 4618
+    },
+    {
+      "epoch": 0.924,
+      "learning_rate": 1.709570736536522e-05,
+      "loss": 0.1795,
+      "step": 4620
+    },
+    {
+      "epoch": 0.9244,
+      "learning_rate": 1.708586190225086e-05,
+      "loss": 0.1223,
+      "step": 4622
+    },
+    {
+      "epoch": 0.9248,
+      "learning_rate": 1.7076002624886152e-05,
+      "loss": 0.397,
+      "step": 4624
+    },
+    {
+      "epoch": 0.9252,
+      "learning_rate": 1.7066129552492258e-05,
+      "loss": 0.3042,
+      "step": 4626
+    },
+    {
+      "epoch": 0.9256,
+      "learning_rate": 1.705624270431722e-05,
+      "loss": 0.9217,
+      "step": 4628
+    },
+    {
+      "epoch": 0.926,
+      "learning_rate": 1.7046342099635945e-05,
+      "loss": 0.1602,
+      "step": 4630
+    },
+    {
+      "epoch": 0.9264,
+      "learning_rate": 1.70364277577502e-05,
+      "loss": 0.646,
+      "step": 4632
+    },
+    {
+      "epoch": 0.9268,
+      "learning_rate": 1.702649969798851e-05,
+      "loss": 0.9233,
+      "step": 4634
+    },
+    {
+      "epoch": 0.9272,
+      "learning_rate": 1.7016557939706078e-05,
+      "loss": 0.3412,
+      "step": 4636
+    },
+    {
+      "epoch": 0.9276,
+      "learning_rate": 1.700660250228492e-05,
+      "loss": 0.8491,
+      "step": 4638
+    },
+    {
+      "epoch": 0.928,
+      "learning_rate": 1.6996633405133673e-05,
+      "loss": 0.4218,
+      "step": 4640
+    },
+    {
+      "epoch": 0.9284,
+      "learning_rate": 1.6986650667687556e-05,
+      "loss": 0.0877,
+      "step": 4642
+    },
+    {
+      "epoch": 0.9288,
+      "learning_rate": 1.6976654309408468e-05,
+      "loss": 0.2737,
+      "step": 4644
+    },
+    {
+      "epoch": 0.9292,
+      "learning_rate": 1.69666443497848e-05,
+      "loss": 0.4594,
+      "step": 4646
+    },
+    {
+      "epoch": 0.9296,
+      "learning_rate": 1.6956620808331515e-05,
+      "loss": 0.2208,
+      "step": 4648
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.694658370458998e-05,
+      "loss": 0.2182,
+      "step": 4650
+    },
+    {
+      "epoch": 0.9304,
+      "learning_rate": 1.6936533058128042e-05,
+      "loss": 0.2622,
+      "step": 4652
+    },
+    {
+      "epoch": 0.9308,
+      "learning_rate": 1.692646888854001e-05,
+      "loss": 0.6582,
+      "step": 4654
+    },
+    {
+      "epoch": 0.9312,
+      "learning_rate": 1.691639121544641e-05,
+      "loss": 0.4253,
+      "step": 4656
+    },
+    {
+      "epoch": 0.9316,
+      "learning_rate": 1.690630005849424e-05,
+      "loss": 0.5316,
+      "step": 4658
+    },
+    {
+      "epoch": 0.932,
+      "learning_rate": 1.6896195437356696e-05,
+      "loss": 0.0417,
+      "step": 4660
+    },
+    {
+      "epoch": 0.9324,
+      "learning_rate": 1.6886077371733295e-05,
+      "loss": 0.1662,
+      "step": 4662
+    },
+    {
+      "epoch": 0.9328,
+      "learning_rate": 1.6875945881349686e-05,
+      "loss": 0.1988,
+      "step": 4664
+    },
+    {
+      "epoch": 0.9332,
+      "learning_rate": 1.6865800985957725e-05,
+      "loss": 0.1458,
+      "step": 4666
+    },
+    {
+      "epoch": 0.9336,
+      "learning_rate": 1.6855642705335435e-05,
+      "loss": 0.394,
+      "step": 4668
+    },
+    {
+      "epoch": 0.934,
+      "learning_rate": 1.68454710592869e-05,
+      "loss": 0.1401,
+      "step": 4670
+    },
+    {
+      "epoch": 0.9344,
+      "learning_rate": 1.6835286067642228e-05,
+      "loss": 0.5404,
+      "step": 4672
+    },
+    {
+      "epoch": 0.9348,
+      "learning_rate": 1.6825087750257617e-05,
+      "loss": 0.2642,
+      "step": 4674
+    },
+    {
+      "epoch": 0.9352,
+      "learning_rate": 1.681487612701521e-05,
+      "loss": 0.172,
+      "step": 4676
+    },
+    {
+      "epoch": 0.9356,
+      "learning_rate": 1.6804651217823055e-05,
+      "loss": 0.2817,
+      "step": 4678
+    },
+    {
+      "epoch": 0.936,
+      "learning_rate": 1.6794413042615168e-05,
+      "loss": 0.354,
+      "step": 4680
+    },
+    {
+      "epoch": 0.9364,
+      "learning_rate": 1.6784161621351374e-05,
+      "loss": 0.2172,
+      "step": 4682
+    },
+    {
+      "epoch": 0.9368,
+      "learning_rate": 1.677389697401739e-05,
+      "loss": 0.3583,
+      "step": 4684
+    },
+    {
+      "epoch": 0.9372,
+      "learning_rate": 1.67636191206246e-05,
+      "loss": 0.4258,
+      "step": 4686
+    },
+    {
+      "epoch": 0.9376,
+      "learning_rate": 1.675332808121025e-05,
+      "loss": 0.4329,
+      "step": 4688
+    },
+    {
+      "epoch": 0.938,
+      "learning_rate": 1.6743023875837253e-05,
+      "loss": 0.3078,
+      "step": 4690
+    },
+    {
+      "epoch": 0.9384,
+      "learning_rate": 1.6732706524594145e-05,
+      "loss": 0.1831,
+      "step": 4692
+    },
+    {
+      "epoch": 0.9388,
+      "learning_rate": 1.672237604759517e-05,
+      "loss": 0.4118,
+      "step": 4694
+    },
+    {
+      "epoch": 0.9392,
+      "learning_rate": 1.671203246498009e-05,
+      "loss": 0.2657,
+      "step": 4696
+    },
+    {
+      "epoch": 0.9396,
+      "learning_rate": 1.670167579691429e-05,
+      "loss": 0.5089,
+      "step": 4698
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6691306063588593e-05,
+      "loss": 0.305,
+      "step": 4700
+    },
+    {
+      "epoch": 0.9404,
+      "learning_rate": 1.668092328521931e-05,
+      "loss": 0.4701,
+      "step": 4702
+    },
+    {
+      "epoch": 0.9408,
+      "learning_rate": 1.6670527482048242e-05,
+      "loss": 0.3058,
+      "step": 4704
+    },
+    {
+      "epoch": 0.9412,
+      "learning_rate": 1.6660118674342525e-05,
+      "loss": 0.0792,
+      "step": 4706
+    },
+    {
+      "epoch": 0.9416,
+      "learning_rate": 1.6649696882394635e-05,
+      "loss": 0.18,
+      "step": 4708
+    },
+    {
+      "epoch": 0.942,
+      "learning_rate": 1.6639262126522414e-05,
+      "loss": 0.1372,
+      "step": 4710
+    },
+    {
+      "epoch": 0.9424,
+      "learning_rate": 1.6628814427068968e-05,
+      "loss": 0.1889,
+      "step": 4712
+    },
+    {
+      "epoch": 0.9428,
+      "learning_rate": 1.661835380440258e-05,
+      "loss": 1.3578,
+      "step": 4714
+    },
+    {
+      "epoch": 0.9432,
+      "learning_rate": 1.6607880278916778e-05,
+      "loss": 0.213,
+      "step": 4716
+    },
+    {
+      "epoch": 0.9436,
+      "learning_rate": 1.6597393871030264e-05,
+      "loss": 0.2541,
+      "step": 4718
+    },
+    {
+      "epoch": 0.944,
+      "learning_rate": 1.6586894601186824e-05,
+      "loss": 0.2216,
+      "step": 4720
+    },
+    {
+      "epoch": 0.9444,
+      "learning_rate": 1.6576382489855278e-05,
+      "loss": 0.528,
+      "step": 4722
+    },
+    {
+      "epoch": 0.9448,
+      "learning_rate": 1.656585755752957e-05,
+      "loss": 0.3694,
+      "step": 4724
+    },
+    {
+      "epoch": 0.9452,
+      "learning_rate": 1.655531982472859e-05,
+      "loss": 0.7978,
+      "step": 4726
+    },
+    {
+      "epoch": 0.9456,
+      "learning_rate": 1.6544769311996153e-05,
+      "loss": 0.7516,
+      "step": 4728
+    },
+    {
+      "epoch": 0.946,
+      "learning_rate": 1.653420603990106e-05,
+      "loss": 0.1636,
+      "step": 4730
+    },
+    {
+      "epoch": 0.9464,
+      "learning_rate": 1.6523630029036924e-05,
+      "loss": 0.1532,
+      "step": 4732
+    },
+    {
+      "epoch": 0.9468,
+      "learning_rate": 1.651304130002226e-05,
+      "loss": 0.6308,
+      "step": 4734
+    },
+    {
+      "epoch": 0.9472,
+      "learning_rate": 1.6502439873500294e-05,
+      "loss": 0.7698,
+      "step": 4736
+    },
+    {
+      "epoch": 0.9476,
+      "learning_rate": 1.6491825770139058e-05,
+      "loss": 0.5311,
+      "step": 4738
+    },
+    {
+      "epoch": 0.948,
+      "learning_rate": 1.6481199010631305e-05,
+      "loss": 0.9658,
+      "step": 4740
+    },
+    {
+      "epoch": 0.9484,
+      "learning_rate": 1.6470559615694455e-05,
+      "loss": 0.3795,
+      "step": 4742
+    },
+    {
+      "epoch": 0.9488,
+      "learning_rate": 1.645990760607052e-05,
+      "loss": 0.1441,
+      "step": 4744
+    },
+    {
+      "epoch": 0.9492,
+      "learning_rate": 1.644924300252614e-05,
+      "loss": 0.1127,
+      "step": 4746
+    },
+    {
+      "epoch": 0.9496,
+      "learning_rate": 1.643856582585255e-05,
+      "loss": 0.1803,
+      "step": 4748
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6427876096865407e-05,
+      "loss": 0.519,
+      "step": 4750
+    },
+    {
+      "epoch": 0.9504,
+      "learning_rate": 1.641717383640488e-05,
+      "loss": 0.4095,
+      "step": 4752
+    },
+    {
+      "epoch": 0.9508,
+      "learning_rate": 1.6406459065335616e-05,
+      "loss": 0.4066,
+      "step": 4754
+    },
+    {
+      "epoch": 0.9512,
+      "learning_rate": 1.6395731804546596e-05,
+      "loss": 0.3499,
+      "step": 4756
+    },
+    {
+      "epoch": 0.9516,
+      "learning_rate": 1.6384992074951128e-05,
+      "loss": 0.3735,
+      "step": 4758
+    },
+    {
+      "epoch": 0.952,
+      "learning_rate": 1.63742398974869e-05,
+      "loss": 0.2905,
+      "step": 4760
+    },
+    {
+      "epoch": 0.9524,
+      "learning_rate": 1.6363475293115838e-05,
+      "loss": 0.3969,
+      "step": 4762
+    },
+    {
+      "epoch": 0.9528,
+      "learning_rate": 1.6352698282824045e-05,
+      "loss": 0.9632,
+      "step": 4764
+    },
+    {
+      "epoch": 0.9532,
+      "learning_rate": 1.63419088876219e-05,
+      "loss": 0.2666,
+      "step": 4766
+    },
+    {
+      "epoch": 0.9536,
+      "learning_rate": 1.633110712854385e-05,
+      "loss": 0.802,
+      "step": 4768
+    },
+    {
+      "epoch": 0.954,
+      "learning_rate": 1.6320293026648515e-05,
+      "loss": 0.2691,
+      "step": 4770
+    },
+    {
+      "epoch": 0.9544,
+      "learning_rate": 1.6309466603018504e-05,
+      "loss": 0.5391,
+      "step": 4772
+    },
+    {
+      "epoch": 0.9548,
+      "learning_rate": 1.6298627878760495e-05,
+      "loss": 0.214,
+      "step": 4774
+    },
+    {
+      "epoch": 0.9552,
+      "learning_rate": 1.6287776875005148e-05,
+      "loss": 0.2097,
+      "step": 4776
+    },
+    {
+      "epoch": 0.9556,
+      "learning_rate": 1.6276913612907015e-05,
+      "loss": 0.4982,
+      "step": 4778
+    },
+    {
+      "epoch": 0.956,
+      "learning_rate": 1.6266038113644612e-05,
+      "loss": 0.5392,
+      "step": 4780
+    },
+    {
+      "epoch": 0.9564,
+      "learning_rate": 1.6255150398420266e-05,
+      "loss": 0.4207,
+      "step": 4782
+    },
+    {
+      "epoch": 0.9568,
+      "learning_rate": 1.624425048846017e-05,
+      "loss": 0.4933,
+      "step": 4784
+    },
+    {
+      "epoch": 0.9572,
+      "learning_rate": 1.623333840501421e-05,
+      "loss": 0.2727,
+      "step": 4786
+    },
+    {
+      "epoch": 0.9576,
+      "learning_rate": 1.6222414169356063e-05,
+      "loss": 0.3189,
+      "step": 4788
+    },
+    {
+      "epoch": 0.958,
+      "learning_rate": 1.6211477802783102e-05,
+      "loss": 0.0928,
+      "step": 4790
+    },
+    {
+      "epoch": 0.9584,
+      "learning_rate": 1.6200529326616343e-05,
+      "loss": 0.2187,
+      "step": 4792
+    },
+    {
+      "epoch": 0.9588,
+      "learning_rate": 1.618956876220035e-05,
+      "loss": 0.2812,
+      "step": 4794
+    },
+    {
+      "epoch": 0.9592,
+      "learning_rate": 1.6178596130903345e-05,
+      "loss": 0.369,
+      "step": 4796
+    },
+    {
+      "epoch": 0.9596,
+      "learning_rate": 1.616761145411704e-05,
+      "loss": 0.457,
+      "step": 4798
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6156614753256587e-05,
+      "loss": 0.377,
+      "step": 4800
+    },
+    {
+      "epoch": 0.9604,
+      "learning_rate": 1.6145606049760648e-05,
+      "loss": 0.4434,
+      "step": 4802
+    },
+    {
+      "epoch": 0.9608,
+      "learning_rate": 1.613458536509123e-05,
+      "loss": 0.8189,
+      "step": 4804
+    },
+    {
+      "epoch": 0.9612,
+      "learning_rate": 1.612355272073378e-05,
+      "loss": 0.5349,
+      "step": 4806
+    },
+    {
+      "epoch": 0.9616,
+      "learning_rate": 1.6112508138196922e-05,
+      "loss": 0.2653,
+      "step": 4808
+    },
+    {
+      "epoch": 0.962,
+      "learning_rate": 1.610145163901268e-05,
+      "loss": 0.6323,
+      "step": 4810
+    },
+    {
+      "epoch": 0.9624,
+      "learning_rate": 1.6090383244736277e-05,
+      "loss": 0.0429,
+      "step": 4812
+    },
+    {
+      "epoch": 0.9628,
+      "learning_rate": 1.6079302976946062e-05,
+      "loss": 0.3353,
+      "step": 4814
+    },
+    {
+      "epoch": 0.9632,
+      "learning_rate": 1.606821085724363e-05,
+      "loss": 0.1851,
+      "step": 4816
+    },
+    {
+      "epoch": 0.9636,
+      "learning_rate": 1.6057106907253607e-05,
+      "loss": 0.2295,
+      "step": 4818
+    },
+    {
+      "epoch": 0.964,
+      "learning_rate": 1.6045991148623756e-05,
+      "loss": 0.3672,
+      "step": 4820
+    },
+    {
+      "epoch": 0.9644,
+      "learning_rate": 1.6034863603024775e-05,
+      "loss": 0.1451,
+      "step": 4822
+    },
+    {
+      "epoch": 0.9648,
+      "learning_rate": 1.602372429215038e-05,
+      "loss": 0.1974,
+      "step": 4824
+    },
+    {
+      "epoch": 0.9652,
+      "learning_rate": 1.6012573237717265e-05,
+      "loss": 0.3073,
+      "step": 4826
+    },
+    {
+      "epoch": 0.9656,
+      "learning_rate": 1.600141046146497e-05,
+      "loss": 0.2618,
+      "step": 4828
+    },
+    {
+      "epoch": 0.966,
+      "learning_rate": 1.5990235985155856e-05,
+      "loss": 0.2805,
+      "step": 4830
+    },
+    {
+      "epoch": 0.9664,
+      "learning_rate": 1.597904983057519e-05,
+      "loss": 0.2173,
+      "step": 4832
+    },
+    {
+      "epoch": 0.9668,
+      "learning_rate": 1.5967852019530942e-05,
+      "loss": 0.2199,
+      "step": 4834
+    },
+    {
+      "epoch": 0.9672,
+      "learning_rate": 1.5956642573853794e-05,
+      "loss": 0.3157,
+      "step": 4836
+    },
+    {
+      "epoch": 0.9676,
+      "learning_rate": 1.5945421515397135e-05,
+      "loss": 0.3602,
+      "step": 4838
+    },
+    {
+      "epoch": 0.968,
+      "learning_rate": 1.5934188866037014e-05,
+      "loss": 0.212,
+      "step": 4840
+    },
+    {
+      "epoch": 0.9684,
+      "learning_rate": 1.5922944647672068e-05,
+      "loss": 1.6341,
+      "step": 4842
+    },
+    {
+      "epoch": 0.9688,
+      "learning_rate": 1.591168888222342e-05,
+      "loss": 0.442,
+      "step": 4844
+    },
+    {
+      "epoch": 0.9692,
+      "learning_rate": 1.5900421591634816e-05,
+      "loss": 0.3332,
+      "step": 4846
+    },
+    {
+      "epoch": 0.9696,
+      "learning_rate": 1.5889142797872407e-05,
+      "loss": 0.1613,
+      "step": 4848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5877852522924736e-05,
+      "loss": 0.2154,
+      "step": 4850
+    },
+    {
+      "epoch": 0.9704,
+      "learning_rate": 1.5866550788802818e-05,
+      "loss": 0.3457,
+      "step": 4852
+    },
+    {
+      "epoch": 0.9708,
+      "learning_rate": 1.5855237617539932e-05,
+      "loss": 0.517,
+      "step": 4854
+    },
+    {
+      "epoch": 0.9712,
+      "learning_rate": 1.584391303119173e-05,
+      "loss": 0.6896,
+      "step": 4856
+    },
+    {
+      "epoch": 0.9716,
+      "learning_rate": 1.5832577051836023e-05,
+      "loss": 0.3691,
+      "step": 4858
+    },
+    {
+      "epoch": 0.972,
+      "learning_rate": 1.582122970157289e-05,
+      "loss": 0.3319,
+      "step": 4860
+    },
+    {
+      "epoch": 0.9724,
+      "learning_rate": 1.5809871002524592e-05,
+      "loss": 0.2666,
+      "step": 4862
+    },
+    {
+      "epoch": 0.9728,
+      "learning_rate": 1.5798500976835503e-05,
+      "loss": 0.1984,
+      "step": 4864
+    },
+    {
+      "epoch": 0.9732,
+      "learning_rate": 1.5787119646672032e-05,
+      "loss": 0.4403,
+      "step": 4866
+    },
+    {
+      "epoch": 0.9736,
+      "learning_rate": 1.577572703422267e-05,
+      "loss": 0.2345,
+      "step": 4868
+    },
+    {
+      "epoch": 0.974,
+      "learning_rate": 1.5764323161697946e-05,
+      "loss": 0.494,
+      "step": 4870
+    },
+    {
+      "epoch": 0.9744,
+      "learning_rate": 1.575290805133024e-05,
+      "loss": 0.1961,
+      "step": 4872
+    },
+    {
+      "epoch": 0.9748,
+      "learning_rate": 1.5741481725373896e-05,
+      "loss": 0.4632,
+      "step": 4874
+    },
+    {
+      "epoch": 0.9752,
+      "learning_rate": 1.5730044206105156e-05,
+      "loss": 0.353,
+      "step": 4876
+    },
+    {
+      "epoch": 0.9756,
+      "learning_rate": 1.571859551582204e-05,
+      "loss": 0.3184,
+      "step": 4878
+    },
+    {
+      "epoch": 0.976,
+      "learning_rate": 1.570713567684432e-05,
+      "loss": 1.1186,
+      "step": 4880
+    },
+    {
+      "epoch": 0.9764,
+      "learning_rate": 1.5695664711513575e-05,
+      "loss": 0.2795,
+      "step": 4882
+    },
+    {
+      "epoch": 0.9768,
+      "learning_rate": 1.5684182642193047e-05,
+      "loss": 0.9398,
+      "step": 4884
+    },
+    {
+      "epoch": 0.9772,
+      "learning_rate": 1.567268949126757e-05,
+      "loss": 0.7083,
+      "step": 4886
+    },
+    {
+      "epoch": 0.9776,
+      "learning_rate": 1.566118528114367e-05,
+      "loss": 0.2595,
+      "step": 4888
+    },
+    {
+      "epoch": 0.978,
+      "learning_rate": 1.5649670034249372e-05,
+      "loss": 0.1587,
+      "step": 4890
+    },
+    {
+      "epoch": 0.9784,
+      "learning_rate": 1.563814377303429e-05,
+      "loss": 0.4812,
+      "step": 4892
+    },
+    {
+      "epoch": 0.9788,
+      "learning_rate": 1.5626606519969373e-05,
+      "loss": 0.5026,
+      "step": 4894
+    },
+    {
+      "epoch": 0.9792,
+      "learning_rate": 1.561505829754715e-05,
+      "loss": 0.1459,
+      "step": 4896
+    },
+    {
+      "epoch": 0.9796,
+      "learning_rate": 1.5603499128281437e-05,
+      "loss": 0.226,
+      "step": 4898
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5591929034707475e-05,
+      "loss": 0.2713,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9804,
+      "learning_rate": 1.558034803938171e-05,
+      "loss": 0.6236,
+      "step": 4902
+    },
+    {
+      "epoch": 0.9808,
+      "learning_rate": 1.5568756164881874e-05,
+      "loss": 0.4798,
+      "step": 4904
+    },
+    {
+      "epoch": 0.9812,
+      "learning_rate": 1.5557153433806974e-05,
+      "loss": 1.7871,
+      "step": 4906
+    },
+    {
+      "epoch": 0.9816,
+      "learning_rate": 1.5545539868777085e-05,
+      "loss": 0.3544,
+      "step": 4908
+    },
+    {
+      "epoch": 0.982,
+      "learning_rate": 1.5533915492433437e-05,
+      "loss": 0.297,
+      "step": 4910
+    },
+    {
+      "epoch": 0.9824,
+      "learning_rate": 1.5522280327438384e-05,
+      "loss": 0.6249,
+      "step": 4912
+    },
+    {
+      "epoch": 0.9828,
+      "learning_rate": 1.5510634396475275e-05,
+      "loss": 0.2954,
+      "step": 4914
+    },
+    {
+      "epoch": 0.9832,
+      "learning_rate": 1.5498977722248398e-05,
+      "loss": 1.0988,
+      "step": 4916
+    },
+    {
+      "epoch": 0.9836,
+      "learning_rate": 1.5487310327483084e-05,
+      "loss": 0.4274,
+      "step": 4918
+    },
+    {
+      "epoch": 0.984,
+      "learning_rate": 1.547563223492552e-05,
+      "loss": 1.0239,
+      "step": 4920
+    },
+    {
+      "epoch": 0.9844,
+      "learning_rate": 1.5463943467342708e-05,
+      "loss": 0.3867,
+      "step": 4922
+    },
+    {
+      "epoch": 0.9848,
+      "learning_rate": 1.5452244047522504e-05,
+      "loss": 0.1976,
+      "step": 4924
+    },
+    {
+      "epoch": 0.9852,
+      "learning_rate": 1.5440533998273552e-05,
+      "loss": 0.103,
+      "step": 4926
+    },
+    {
+      "epoch": 0.9856,
+      "learning_rate": 1.5428813342425194e-05,
+      "loss": 0.4862,
+      "step": 4928
+    },
+    {
+      "epoch": 0.986,
+      "learning_rate": 1.5417082102827407e-05,
+      "loss": 0.3815,
+      "step": 4930
+    },
+    {
+      "epoch": 0.9864,
+      "learning_rate": 1.5405340302350876e-05,
+      "loss": 0.156,
+      "step": 4932
+    },
+    {
+      "epoch": 0.9868,
+      "learning_rate": 1.5393587963886827e-05,
+      "loss": 0.3448,
+      "step": 4934
+    },
+    {
+      "epoch": 0.9872,
+      "learning_rate": 1.538182511034708e-05,
+      "loss": 0.3498,
+      "step": 4936
+    },
+    {
+      "epoch": 0.9876,
+      "learning_rate": 1.5370051764663875e-05,
+      "loss": 0.6057,
+      "step": 4938
+    },
+    {
+      "epoch": 0.988,
+      "learning_rate": 1.535826794978996e-05,
+      "loss": 0.1878,
+      "step": 4940
+    },
+    {
+      "epoch": 0.9884,
+      "learning_rate": 1.534647368869852e-05,
+      "loss": 0.267,
+      "step": 4942
+    },
+    {
+      "epoch": 0.9888,
+      "learning_rate": 1.5334669004383036e-05,
+      "loss": 0.0995,
+      "step": 4944
+    },
+    {
+      "epoch": 0.9892,
+      "learning_rate": 1.5322853919857337e-05,
+      "loss": 0.2764,
+      "step": 4946
+    },
+    {
+      "epoch": 0.9896,
+      "learning_rate": 1.5311028458155564e-05,
+      "loss": 0.2402,
+      "step": 4948
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5299192642332063e-05,
+      "loss": 0.5147,
+      "step": 4950
+    },
+    {
+      "epoch": 0.9904,
+      "learning_rate": 1.528734649546133e-05,
+      "loss": 0.0648,
+      "step": 4952
+    },
+    {
+      "epoch": 0.9908,
+      "learning_rate": 1.5275490040638038e-05,
+      "loss": 0.2504,
+      "step": 4954
+    },
+    {
+      "epoch": 0.9912,
+      "learning_rate": 1.5263623300976997e-05,
+      "loss": 0.7926,
+      "step": 4956
+    },
+    {
+      "epoch": 0.9916,
+      "learning_rate": 1.5251746299612973e-05,
+      "loss": 0.1338,
+      "step": 4958
+    },
+    {
+      "epoch": 0.992,
+      "learning_rate": 1.5239859059700792e-05,
+      "loss": 0.3954,
+      "step": 4960
+    },
+    {
+      "epoch": 0.9924,
+      "learning_rate": 1.522796160441527e-05,
+      "loss": 0.2655,
+      "step": 4962
+    },
+    {
+      "epoch": 0.9928,
+      "learning_rate": 1.5216053956951096e-05,
+      "loss": 0.3116,
+      "step": 4964
+    },
+    {
+      "epoch": 0.9932,
+      "learning_rate": 1.5204136140522799e-05,
+      "loss": 0.1699,
+      "step": 4966
+    },
+    {
+      "epoch": 0.9936,
+      "learning_rate": 1.5192208178364819e-05,
+      "loss": 0.2399,
+      "step": 4968
+    },
+    {
+      "epoch": 0.994,
+      "learning_rate": 1.5180270093731291e-05,
+      "loss": 0.3266,
+      "step": 4970
+    },
+    {
+      "epoch": 0.9944,
+      "learning_rate": 1.5168321909896176e-05,
+      "loss": 0.253,
+      "step": 4972
+    },
+    {
+      "epoch": 0.9948,
+      "learning_rate": 1.5156363650153017e-05,
+      "loss": 0.2107,
+      "step": 4974
+    },
+    {
+      "epoch": 0.9952,
+      "learning_rate": 1.5144395337815057e-05,
+      "loss": 0.1531,
+      "step": 4976
+    },
+    {
+      "epoch": 0.9956,
+      "learning_rate": 1.5132416996215178e-05,
+      "loss": 0.6295,
+      "step": 4978
+    },
+    {
+      "epoch": 0.996,
+      "learning_rate": 1.5120428648705722e-05,
+      "loss": 0.4572,
+      "step": 4980
+    },
+    {
+      "epoch": 0.9964,
+      "learning_rate": 1.5108430318658607e-05,
+      "loss": 0.261,
+      "step": 4982
+    },
+    {
+      "epoch": 0.9968,
+      "learning_rate": 1.5096422029465171e-05,
+      "loss": 0.6105,
+      "step": 4984
+    },
+    {
+      "epoch": 0.9972,
+      "learning_rate": 1.5084403804536236e-05,
+      "loss": 0.1747,
+      "step": 4986
+    },
+    {
+      "epoch": 0.9976,
+      "learning_rate": 1.5072375667301904e-05,
+      "loss": 0.7721,
+      "step": 4988
+    },
+    {
+      "epoch": 0.998,
+      "learning_rate": 1.5060337641211636e-05,
+      "loss": 0.159,
+      "step": 4990
+    },
+    {
+      "epoch": 0.9984,
+      "learning_rate": 1.5048289749734231e-05,
+      "loss": 0.2571,
+      "step": 4992
+    },
+    {
+      "epoch": 0.9988,
+      "learning_rate": 1.5036232016357622e-05,
+      "loss": 0.5908,
+      "step": 4994
+    },
+    {
+      "epoch": 0.9992,
+      "learning_rate": 1.502416446458898e-05,
+      "loss": 0.344,
+      "step": 4996
+    },
+    {
+      "epoch": 0.9996,
+      "learning_rate": 1.5012087117954641e-05,
+      "loss": 0.4671,
+      "step": 4998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5000000000000014e-05,
+      "loss": 0.1265,
+      "step": 5000
+    },
+    {
+      "epoch": 1.0,
+      "step": 5000,
+      "total_flos": 0,
+      "train_loss": 0.4200633437473327,
+      "train_runtime": 22460.8343,
+      "train_samples_per_second": 3.562,
+      "train_steps_per_second": 0.223
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 5000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {},
+  "total_flos": 0,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}
diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round0.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round0.pth
new file mode 100644
index 0000000000000000000000000000000000000000..12551f8335d5e0528896ce3ebbff583f9b5d4efb
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round0.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:64591c8820f0c4df71367fc4c404b79094ab0e1a2cfe4d84a8aba5d404efbcb9
+size 639793378
diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round1.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round1.pth
new file mode 100644
index 0000000000000000000000000000000000000000..d554d9dd62ff5b1becacdbd1d7bc54f71496299a
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round1.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:62131a6ec5a0420454e2118ed3bab89a132aee41060161c579deb6e79b4d4064
+size 639793378
diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round2.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round2.pth
new file mode 100644
index 0000000000000000000000000000000000000000..725185786c7b940f005fb9789b0f2192123b57ba
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round2.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:19e6be21445b4f8d8e78d3d46c5ce24ec9aa4a578c46de0fb697350f1b5d4a5e
+size 639793378
diff --git a/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round3.pth b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round3.pth
new file mode 100644
index 0000000000000000000000000000000000000000..b4d898b66c864b78f54bfe2f6bffa3df5327e43e
--- /dev/null
+++ b/client_states_v9_NEURIPS_DISJOINT_Memonly_LORA_llava_lr2e-5_bs1_gradacc32_iter0_5_divbs_scenario18_new_10000_random0_0625_seed1/server_model_round3.pth
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5643f57073b3eaf4b4755917e0b453cd04ff3e747957929d121a97d672d3c922
+size 639793378