Namespace(batch_norm=False, batch_size=128, data_dir='~/.mxnet/datasets/imagenet', dtype='float16', input_size=224, label_smoothing=True, last_gamma=False, log_interval=50, logging_file='mobilenet0.5.log', lr=0.4, lr_decay=0.1, lr_decay_epoch='40,60', lr_decay_period=0, lr_mode='cosine', mixup=False, mixup_alpha=0.2, mixup_off_epoch=0, mode='hybrid', model='mobilenet0.5', momentum=0.9, no_wd=True, num_epochs=120, num_gpus=8, num_workers=60, rec_train='/media/ramdisk/rec/train.rec', rec_train_idx='/media/ramdisk/rec/train.idx', rec_val='/media/ramdisk/rec/val.rec', rec_val_idx='/media/ramdisk/rec/val.idx', resume_epoch=0, resume_params='', resume_states='', save_dir='params_mobilenet0.5', save_frequency=10, use_pretrained=False, use_rec=True, use_se=False, warmup_epochs=5, warmup_lr=0.0, wd=0.0001)
Epoch[0] Batch [49]	Speed: 2586.953591 samples/sec	accuracy=0.000957	lr=0.003133
Epoch[0] Batch [99]	Speed: 5870.650790 samples/sec	accuracy=0.001104	lr=0.006331
Epoch[0] Batch [149]	Speed: 5378.944579 samples/sec	accuracy=0.001172	lr=0.009528
Epoch[0] Batch [199]	Speed: 5799.247893 samples/sec	accuracy=0.001152	lr=0.012726
Epoch[0] Batch [249]	Speed: 4671.377252 samples/sec	accuracy=0.001164	lr=0.015923
Epoch[0] Batch [299]	Speed: 4749.100087 samples/sec	accuracy=0.001217	lr=0.019121
Epoch[0] Batch [349]	Speed: 5483.518927 samples/sec	accuracy=0.001214	lr=0.022318
Epoch[0] Batch [399]	Speed: 4636.200859 samples/sec	accuracy=0.001216	lr=0.025516
Epoch[0] Batch [449]	Speed: 5365.456714 samples/sec	accuracy=0.001263	lr=0.028713
Epoch[0] Batch [499]	Speed: 4702.576775 samples/sec	accuracy=0.001326	lr=0.031910
Epoch[0] Batch [549]	Speed: 4958.785421 samples/sec	accuracy=0.001452	lr=0.035108
Epoch[0] Batch [599]	Speed: 5049.956616 samples/sec	accuracy=0.001610	lr=0.038305
Epoch[0] Batch [649]	Speed: 4902.495859 samples/sec	accuracy=0.001812	lr=0.041503
Epoch[0] Batch [699]	Speed: 5313.848681 samples/sec	accuracy=0.002003	lr=0.044700
Epoch[0] Batch [749]	Speed: 5532.507582 samples/sec	accuracy=0.002207	lr=0.047898
Epoch[0] Batch [799]	Speed: 4809.909901 samples/sec	accuracy=0.002500	lr=0.051095
Epoch[0] Batch [849]	Speed: 5084.118832 samples/sec	accuracy=0.002779	lr=0.054293
Epoch[0] Batch [899]	Speed: 4851.782722 samples/sec	accuracy=0.003104	lr=0.057490
Epoch[0] Batch [949]	Speed: 4915.087015 samples/sec	accuracy=0.003535	lr=0.060687
Epoch[0] Batch [999]	Speed: 5735.205222 samples/sec	accuracy=0.004018	lr=0.063885
Epoch[0] Batch [1049]	Speed: 4835.256650 samples/sec	accuracy=0.004540	lr=0.067082
Epoch[0] Batch [1099]	Speed: 5249.702306 samples/sec	accuracy=0.005169	lr=0.070280
Epoch[0] Batch [1149]	Speed: 4336.730181 samples/sec	accuracy=0.005834	lr=0.073477
Epoch[0] Batch [1199]	Speed: 3663.012572 samples/sec	accuracy=0.006576	lr=0.076675
Epoch[0] Batch [1249]	Speed: 5042.237124 samples/sec	accuracy=0.007380	lr=0.079872
[Epoch 0] training: accuracy=0.007411
[Epoch 0] speed: 4803 samples/sec	time cost: 278.026651
[Epoch 0] validation: err-top1=0.971122 err-top5=0.901108
Epoch[1] Batch [49]	Speed: 4082.785815 samples/sec	accuracy=0.027988	lr=0.083133
Epoch[1] Batch [99]	Speed: 5013.185064 samples/sec	accuracy=0.029365	lr=0.086331
Epoch[1] Batch [149]	Speed: 5400.855281 samples/sec	accuracy=0.030677	lr=0.089528
Epoch[1] Batch [199]	Speed: 5401.060935 samples/sec	accuracy=0.032549	lr=0.092726
Epoch[1] Batch [249]	Speed: 4664.240334 samples/sec	accuracy=0.034199	lr=0.095923
Epoch[1] Batch [299]	Speed: 4724.580487 samples/sec	accuracy=0.035697	lr=0.099121
Epoch[1] Batch [349]	Speed: 4348.617054 samples/sec	accuracy=0.037341	lr=0.102318
Epoch[1] Batch [399]	Speed: 4792.897272 samples/sec	accuracy=0.039346	lr=0.105516
Epoch[1] Batch [449]	Speed: 5302.311455 samples/sec	accuracy=0.041172	lr=0.108713
Epoch[1] Batch [499]	Speed: 5396.519273 samples/sec	accuracy=0.042965	lr=0.111910
Epoch[1] Batch [549]	Speed: 5863.771997 samples/sec	accuracy=0.045075	lr=0.115108
Epoch[1] Batch [599]	Speed: 4883.759827 samples/sec	accuracy=0.047088	lr=0.118305
Epoch[1] Batch [649]	Speed: 5702.845087 samples/sec	accuracy=0.049096	lr=0.121503
Epoch[1] Batch [699]	Speed: 5805.623823 samples/sec	accuracy=0.051283	lr=0.124700
Epoch[1] Batch [749]	Speed: 5794.626636 samples/sec	accuracy=0.053434	lr=0.127898
Epoch[1] Batch [799]	Speed: 5379.857260 samples/sec	accuracy=0.055602	lr=0.131095
Epoch[1] Batch [849]	Speed: 5623.037563 samples/sec	accuracy=0.057924	lr=0.134293
Epoch[1] Batch [899]	Speed: 6018.998059 samples/sec	accuracy=0.060178	lr=0.137490
Epoch[1] Batch [949]	Speed: 5697.003353 samples/sec	accuracy=0.062485	lr=0.140687
Epoch[1] Batch [999]	Speed: 5662.380710 samples/sec	accuracy=0.064682	lr=0.143885
Epoch[1] Batch [1049]	Speed: 5420.729558 samples/sec	accuracy=0.067097	lr=0.147082
Epoch[1] Batch [1099]	Speed: 5783.807489 samples/sec	accuracy=0.069322	lr=0.150280
Epoch[1] Batch [1149]	Speed: 5245.028848 samples/sec	accuracy=0.071588	lr=0.153477
Epoch[1] Batch [1199]	Speed: 5485.722180 samples/sec	accuracy=0.073889	lr=0.156675
Epoch[1] Batch [1249]	Speed: 6131.500073 samples/sec	accuracy=0.076076	lr=0.159872
[Epoch 1] training: accuracy=0.076120
[Epoch 1] speed: 5284 samples/sec	time cost: 252.009381
[Epoch 1] validation: err-top1=0.870675 err-top5=0.685108
Epoch[2] Batch [49]	Speed: 5220.417421 samples/sec	accuracy=0.136797	lr=0.163133
Epoch[2] Batch [99]	Speed: 5234.826980 samples/sec	accuracy=0.137393	lr=0.166331
Epoch[2] Batch [149]	Speed: 5729.048817 samples/sec	accuracy=0.139466	lr=0.169528
Epoch[2] Batch [199]	Speed: 5377.360486 samples/sec	accuracy=0.141733	lr=0.172726
Epoch[2] Batch [249]	Speed: 5450.894297 samples/sec	accuracy=0.143527	lr=0.175923
Epoch[2] Batch [299]	Speed: 5561.817510 samples/sec	accuracy=0.145234	lr=0.179121
Epoch[2] Batch [349]	Speed: 4944.591597 samples/sec	accuracy=0.147589	lr=0.182318
Epoch[2] Batch [399]	Speed: 5654.581879 samples/sec	accuracy=0.149578	lr=0.185516
Epoch[2] Batch [449]	Speed: 4669.043294 samples/sec	accuracy=0.151458	lr=0.188713
Epoch[2] Batch [499]	Speed: 5445.219629 samples/sec	accuracy=0.153412	lr=0.191910
Epoch[2] Batch [549]	Speed: 5931.603459 samples/sec	accuracy=0.155357	lr=0.195108
Epoch[2] Batch [599]	Speed: 4346.686255 samples/sec	accuracy=0.157749	lr=0.198305
Epoch[2] Batch [649]	Speed: 4882.790197 samples/sec	accuracy=0.159742	lr=0.201503
Epoch[2] Batch [699]	Speed: 5817.778497 samples/sec	accuracy=0.161503	lr=0.204700
Epoch[2] Batch [749]	Speed: 5422.462807 samples/sec	accuracy=0.163385	lr=0.207898
Epoch[2] Batch [799]	Speed: 5449.564720 samples/sec	accuracy=0.164983	lr=0.211095
Epoch[2] Batch [849]	Speed: 5516.201736 samples/sec	accuracy=0.166654	lr=0.214293
Epoch[2] Batch [899]	Speed: 5437.833225 samples/sec	accuracy=0.168553	lr=0.217490
Epoch[2] Batch [949]	Speed: 5547.549065 samples/sec	accuracy=0.170241	lr=0.220687
Epoch[2] Batch [999]	Speed: 5404.259691 samples/sec	accuracy=0.171983	lr=0.223885
Epoch[2] Batch [1049]	Speed: 5667.267726 samples/sec	accuracy=0.173445	lr=0.227082
Epoch[2] Batch [1099]	Speed: 5641.721027 samples/sec	accuracy=0.175018	lr=0.230280
Epoch[2] Batch [1149]	Speed: 4664.638092 samples/sec	accuracy=0.176762	lr=0.233477
Epoch[2] Batch [1199]	Speed: 5004.325070 samples/sec	accuracy=0.178413	lr=0.236675
Epoch[2] Batch [1249]	Speed: 5405.018004 samples/sec	accuracy=0.180113	lr=0.239872
[Epoch 2] training: accuracy=0.180143
[Epoch 2] speed: 5299 samples/sec	time cost: 252.019205
[Epoch 2] validation: err-top1=0.772301 err-top5=0.538445
Epoch[3] Batch [49]	Speed: 4972.517567 samples/sec	accuracy=0.228516	lr=0.243133
Epoch[3] Batch [99]	Speed: 5347.823063 samples/sec	accuracy=0.227227	lr=0.246331
Epoch[3] Batch [149]	Speed: 5450.539154 samples/sec	accuracy=0.228796	lr=0.249528
Epoch[3] Batch [199]	Speed: 5556.273063 samples/sec	accuracy=0.230293	lr=0.252726
Epoch[3] Batch [249]	Speed: 5168.494935 samples/sec	accuracy=0.230289	lr=0.255923
Epoch[3] Batch [299]	Speed: 5245.498651 samples/sec	accuracy=0.231230	lr=0.259121
Epoch[3] Batch [349]	Speed: 5913.183203 samples/sec	accuracy=0.232224	lr=0.262318
Epoch[3] Batch [399]	Speed: 5658.540856 samples/sec	accuracy=0.233135	lr=0.265516
Epoch[3] Batch [449]	Speed: 5428.116737 samples/sec	accuracy=0.233908	lr=0.268713
Epoch[3] Batch [499]	Speed: 5103.340153 samples/sec	accuracy=0.235051	lr=0.271910
Epoch[3] Batch [549]	Speed: 5435.783151 samples/sec	accuracy=0.235882	lr=0.275108
Epoch[3] Batch [599]	Speed: 5497.617109 samples/sec	accuracy=0.236722	lr=0.278305
Epoch[3] Batch [649]	Speed: 4896.863177 samples/sec	accuracy=0.237988	lr=0.281503
Epoch[3] Batch [699]	Speed: 4877.936734 samples/sec	accuracy=0.239035	lr=0.284700
Epoch[3] Batch [749]	Speed: 4887.073377 samples/sec	accuracy=0.240047	lr=0.287898
Epoch[3] Batch [799]	Speed: 5658.141295 samples/sec	accuracy=0.240933	lr=0.291095
Epoch[3] Batch [849]	Speed: 4877.435413 samples/sec	accuracy=0.241867	lr=0.294293
Epoch[3] Batch [899]	Speed: 5368.664694 samples/sec	accuracy=0.242809	lr=0.297490
Epoch[3] Batch [949]	Speed: 5201.035284 samples/sec	accuracy=0.243780	lr=0.300687
Epoch[3] Batch [999]	Speed: 4898.873251 samples/sec	accuracy=0.244792	lr=0.303885
Epoch[3] Batch [1049]	Speed: 5550.256483 samples/sec	accuracy=0.245677	lr=0.307082
Epoch[3] Batch [1099]	Speed: 4768.415234 samples/sec	accuracy=0.246437	lr=0.310280
Epoch[3] Batch [1149]	Speed: 4563.883158 samples/sec	accuracy=0.247287	lr=0.313477
Epoch[3] Batch [1199]	Speed: 5003.969064 samples/sec	accuracy=0.248095	lr=0.316675
Epoch[3] Batch [1249]	Speed: 5577.537130 samples/sec	accuracy=0.248898	lr=0.319872
[Epoch 3] training: accuracy=0.248914
[Epoch 3] speed: 5205 samples/sec	time cost: 255.222151
[Epoch 3] validation: err-top1=0.720125 err-top5=0.469169
Epoch[4] Batch [49]	Speed: 4972.573065 samples/sec	accuracy=0.275508	lr=0.323133
Epoch[4] Batch [99]	Speed: 5755.471325 samples/sec	accuracy=0.277187	lr=0.326331
Epoch[4] Batch [149]	Speed: 5615.780728 samples/sec	accuracy=0.278021	lr=0.329528
Epoch[4] Batch [199]	Speed: 5411.998342 samples/sec	accuracy=0.277661	lr=0.332726
Epoch[4] Batch [249]	Speed: 5400.663224 samples/sec	accuracy=0.277852	lr=0.335923
Epoch[4] Batch [299]	Speed: 5107.103489 samples/sec	accuracy=0.277689	lr=0.339121
Epoch[4] Batch [349]	Speed: 5689.478655 samples/sec	accuracy=0.278267	lr=0.342318
Epoch[4] Batch [399]	Speed: 5666.003921 samples/sec	accuracy=0.278650	lr=0.345516
Epoch[4] Batch [449]	Speed: 5515.429330 samples/sec	accuracy=0.279325	lr=0.348713
Epoch[4] Batch [499]	Speed: 5264.801549 samples/sec	accuracy=0.279695	lr=0.351910
Epoch[4] Batch [549]	Speed: 5472.029798 samples/sec	accuracy=0.280451	lr=0.355108
Epoch[4] Batch [599]	Speed: 5297.779142 samples/sec	accuracy=0.281123	lr=0.358305
Epoch[4] Batch [649]	Speed: 5055.510869 samples/sec	accuracy=0.281889	lr=0.361503
Epoch[4] Batch [699]	Speed: 5989.237886 samples/sec	accuracy=0.282415	lr=0.364700
Epoch[4] Batch [749]	Speed: 5214.941106 samples/sec	accuracy=0.283022	lr=0.367898
Epoch[4] Batch [799]	Speed: 5543.927442 samples/sec	accuracy=0.283478	lr=0.371095
Epoch[4] Batch [849]	Speed: 5342.395418 samples/sec	accuracy=0.284135	lr=0.374293
Epoch[4] Batch [899]	Speed: 5529.723755 samples/sec	accuracy=0.284629	lr=0.377490
Epoch[4] Batch [949]	Speed: 5353.126970 samples/sec	accuracy=0.285108	lr=0.380687
Epoch[4] Batch [999]	Speed: 5564.989355 samples/sec	accuracy=0.285549	lr=0.383885
Epoch[4] Batch [1049]	Speed: 5434.251216 samples/sec	accuracy=0.286071	lr=0.387082
Epoch[4] Batch [1099]	Speed: 5658.779725 samples/sec	accuracy=0.286596	lr=0.390280
Epoch[4] Batch [1149]	Speed: 5845.840641 samples/sec	accuracy=0.287136	lr=0.393477
Epoch[4] Batch [1199]	Speed: 5569.615020 samples/sec	accuracy=0.287575	lr=0.396675
Epoch[4] Batch [1249]	Speed: 5540.729803 samples/sec	accuracy=0.288095	lr=0.399872
[Epoch 4] training: accuracy=0.288119
[Epoch 4] speed: 5455 samples/sec	time cost: 243.974594
[Epoch 4] validation: err-top1=0.703304 err-top5=0.449478
Epoch[5] Batch [49]	Speed: 5616.409193 samples/sec	accuracy=0.307480	lr=0.400000
Epoch[5] Batch [99]	Speed: 5712.434289 samples/sec	accuracy=0.307715	lr=0.400000
Epoch[5] Batch [149]	Speed: 5429.710698 samples/sec	accuracy=0.307741	lr=0.399999
Epoch[5] Batch [199]	Speed: 5819.645359 samples/sec	accuracy=0.307837	lr=0.399998
Epoch[5] Batch [249]	Speed: 5656.681588 samples/sec	accuracy=0.308195	lr=0.399997
Epoch[5] Batch [299]	Speed: 5323.950403 samples/sec	accuracy=0.308717	lr=0.399996
Epoch[5] Batch [349]	Speed: 4988.627464 samples/sec	accuracy=0.308705	lr=0.399994
Epoch[5] Batch [399]	Speed: 4360.756105 samples/sec	accuracy=0.309063	lr=0.399992
Epoch[5] Batch [449]	Speed: 5408.258634 samples/sec	accuracy=0.309586	lr=0.399990
Epoch[5] Batch [499]	Speed: 5555.696646 samples/sec	accuracy=0.309576	lr=0.399988
Epoch[5] Batch [549]	Speed: 5663.605856 samples/sec	accuracy=0.310321	lr=0.399986
Epoch[5] Batch [599]	Speed: 3511.089582 samples/sec	accuracy=0.310617	lr=0.399983
Epoch[5] Batch [649]	Speed: 4081.442936 samples/sec	accuracy=0.311082	lr=0.399980
Epoch[5] Batch [699]	Speed: 5892.519969 samples/sec	accuracy=0.311550	lr=0.399977
Epoch[5] Batch [749]	Speed: 5223.205882 samples/sec	accuracy=0.312063	lr=0.399973
Epoch[5] Batch [799]	Speed: 3417.156676 samples/sec	accuracy=0.312363	lr=0.399970
Epoch[5] Batch [849]	Speed: 3598.922329 samples/sec	accuracy=0.312935	lr=0.399966
Epoch[5] Batch [899]	Speed: 5286.208975 samples/sec	accuracy=0.313381	lr=0.399961
Epoch[5] Batch [949]	Speed: 4239.875289 samples/sec	accuracy=0.313549	lr=0.399957
Epoch[5] Batch [999]	Speed: 3462.326406 samples/sec	accuracy=0.314180	lr=0.399952
Epoch[5] Batch [1049]	Speed: 3741.559905 samples/sec	accuracy=0.314664	lr=0.399948
Epoch[5] Batch [1099]	Speed: 4128.826457 samples/sec	accuracy=0.315226	lr=0.399942
Epoch[5] Batch [1149]	Speed: 1967.164119 samples/sec	accuracy=0.315510	lr=0.399937
Epoch[5] Batch [1199]	Speed: 4758.904733 samples/sec	accuracy=0.315894	lr=0.399931
Epoch[5] Batch [1249]	Speed: 4613.661699 samples/sec	accuracy=0.316291	lr=0.399926
[Epoch 5] training: accuracy=0.316308
[Epoch 5] speed: 4406 samples/sec	time cost: 299.500018
[Epoch 5] validation: err-top1=0.655741 err-top5=0.396098
Epoch[6] Batch [49]	Speed: 4976.311068 samples/sec	accuracy=0.327715	lr=0.399919
Epoch[6] Batch [99]	Speed: 4941.617021 samples/sec	accuracy=0.331748	lr=0.399913
Epoch[6] Batch [149]	Speed: 5039.572971 samples/sec	accuracy=0.331901	lr=0.399907
Epoch[6] Batch [199]	Speed: 4911.905585 samples/sec	accuracy=0.332666	lr=0.399900
Epoch[6] Batch [249]	Speed: 3807.529636 samples/sec	accuracy=0.333027	lr=0.399893
Epoch[6] Batch [299]	Speed: 5767.492126 samples/sec	accuracy=0.333444	lr=0.399885
Epoch[6] Batch [349]	Speed: 4975.062983 samples/sec	accuracy=0.333652	lr=0.399878
Epoch[6] Batch [399]	Speed: 4509.065765 samples/sec	accuracy=0.333887	lr=0.399870
Epoch[6] Batch [449]	Speed: 4159.383296 samples/sec	accuracy=0.334531	lr=0.399862
Epoch[6] Batch [499]	Speed: 4768.623511 samples/sec	accuracy=0.334498	lr=0.399854
Epoch[6] Batch [549]	Speed: 5038.810630 samples/sec	accuracy=0.334627	lr=0.399846
Epoch[6] Batch [599]	Speed: 4178.709009 samples/sec	accuracy=0.334782	lr=0.399837
Epoch[6] Batch [649]	Speed: 5265.488047 samples/sec	accuracy=0.335003	lr=0.399828
Epoch[6] Batch [699]	Speed: 5105.825241 samples/sec	accuracy=0.335303	lr=0.399819
Epoch[6] Batch [749]	Speed: 3931.818918 samples/sec	accuracy=0.335599	lr=0.399809
Epoch[6] Batch [799]	Speed: 5468.108534 samples/sec	accuracy=0.335908	lr=0.399800
Epoch[6] Batch [849]	Speed: 4413.172624 samples/sec	accuracy=0.336074	lr=0.399790
Epoch[6] Batch [899]	Speed: 4568.468824 samples/sec	accuracy=0.336190	lr=0.399780
Epoch[6] Batch [949]	Speed: 4882.494898 samples/sec	accuracy=0.336554	lr=0.399769
Epoch[6] Batch [999]	Speed: 4184.630475 samples/sec	accuracy=0.336470	lr=0.399759
Epoch[6] Batch [1049]	Speed: 3774.410291 samples/sec	accuracy=0.336632	lr=0.399748
Epoch[6] Batch [1099]	Speed: 4044.989075 samples/sec	accuracy=0.336935	lr=0.399737
Epoch[6] Batch [1149]	Speed: 4788.648940 samples/sec	accuracy=0.337322	lr=0.399725
Epoch[6] Batch [1199]	Speed: 3827.332893 samples/sec	accuracy=0.337525	lr=0.399714
Epoch[6] Batch [1249]	Speed: 4849.091396 samples/sec	accuracy=0.337765	lr=0.399702
[Epoch 6] training: accuracy=0.337770
[Epoch 6] speed: 4580 samples/sec	time cost: 289.244275
[Epoch 6] validation: err-top1=0.705377 err-top5=0.455158
Epoch[7] Batch [49]	Speed: 3168.065118 samples/sec	accuracy=0.350273	lr=0.399690
Epoch[7] Batch [99]	Speed: 5200.553513 samples/sec	accuracy=0.349551	lr=0.399677
Epoch[7] Batch [149]	Speed: 4639.887233 samples/sec	accuracy=0.350163	lr=0.399665
Epoch[7] Batch [199]	Speed: 3792.442564 samples/sec	accuracy=0.350259	lr=0.399652
Epoch[7] Batch [249]	Speed: 4479.065861 samples/sec	accuracy=0.350512	lr=0.399639
Epoch[7] Batch [299]	Speed: 5607.991168 samples/sec	accuracy=0.351289	lr=0.399626
Epoch[7] Batch [349]	Speed: 3835.159445 samples/sec	accuracy=0.350907	lr=0.399613
Epoch[7] Batch [399]	Speed: 5089.041240 samples/sec	accuracy=0.350869	lr=0.399599
Epoch[7] Batch [449]	Speed: 4302.292465 samples/sec	accuracy=0.351048	lr=0.399585
Epoch[7] Batch [499]	Speed: 4502.033746 samples/sec	accuracy=0.351324	lr=0.399571
Epoch[7] Batch [549]	Speed: 4380.596123 samples/sec	accuracy=0.351199	lr=0.399556
Epoch[7] Batch [599]	Speed: 4139.356628 samples/sec	accuracy=0.351060	lr=0.399542
Epoch[7] Batch [649]	Speed: 6031.982102 samples/sec	accuracy=0.351085	lr=0.399527
Epoch[7] Batch [699]	Speed: 4540.777855 samples/sec	accuracy=0.351182	lr=0.399512
Epoch[7] Batch [749]	Speed: 5094.600387 samples/sec	accuracy=0.351193	lr=0.399496
Epoch[7] Batch [799]	Speed: 5247.434481 samples/sec	accuracy=0.351298	lr=0.399481
Epoch[7] Batch [849]	Speed: 4597.619877 samples/sec	accuracy=0.351641	lr=0.399465
Epoch[7] Batch [899]	Speed: 4393.269783 samples/sec	accuracy=0.351809	lr=0.399449
Epoch[7] Batch [949]	Speed: 5206.684517 samples/sec	accuracy=0.351959	lr=0.399432
Epoch[7] Batch [999]	Speed: 5490.806257 samples/sec	accuracy=0.352154	lr=0.399416
Epoch[7] Batch [1049]	Speed: 4969.499647 samples/sec	accuracy=0.352370	lr=0.399399
Epoch[7] Batch [1099]	Speed: 4816.367062 samples/sec	accuracy=0.352470	lr=0.399382
Epoch[7] Batch [1149]	Speed: 5400.142675 samples/sec	accuracy=0.352618	lr=0.399365
Epoch[7] Batch [1199]	Speed: 4709.568198 samples/sec	accuracy=0.352745	lr=0.399347
Epoch[7] Batch [1249]	Speed: 4645.667932 samples/sec	accuracy=0.352952	lr=0.399329
[Epoch 7] training: accuracy=0.352984
[Epoch 7] speed: 4636 samples/sec	time cost: 286.601871
[Epoch 7] validation: err-top1=0.616948 err-top5=0.359375
Epoch[8] Batch [49]	Speed: 4934.223027 samples/sec	accuracy=0.366738	lr=0.399311
Epoch[8] Batch [99]	Speed: 5138.196522 samples/sec	accuracy=0.363389	lr=0.399293
Epoch[8] Batch [149]	Speed: 4855.327762 samples/sec	accuracy=0.363158	lr=0.399274
Epoch[8] Batch [199]	Speed: 5967.508629 samples/sec	accuracy=0.363071	lr=0.399256
Epoch[8] Batch [249]	Speed: 4371.049522 samples/sec	accuracy=0.363781	lr=0.399237
Epoch[8] Batch [299]	Speed: 5306.052991 samples/sec	accuracy=0.363590	lr=0.399218
Epoch[8] Batch [349]	Speed: 5566.796764 samples/sec	accuracy=0.363387	lr=0.399198
Epoch[8] Batch [399]	Speed: 5057.912534 samples/sec	accuracy=0.363032	lr=0.399179
Epoch[8] Batch [449]	Speed: 5032.099292 samples/sec	accuracy=0.363071	lr=0.399159
Epoch[8] Batch [499]	Speed: 5246.469400 samples/sec	accuracy=0.362988	lr=0.399138
Epoch[8] Batch [549]	Speed: 5220.956063 samples/sec	accuracy=0.362830	lr=0.399118
Epoch[8] Batch [599]	Speed: 5050.278339 samples/sec	accuracy=0.362733	lr=0.399098
Epoch[8] Batch [649]	Speed: 5353.530656 samples/sec	accuracy=0.362999	lr=0.399077
Epoch[8] Batch [699]	Speed: 4504.380825 samples/sec	accuracy=0.362953	lr=0.399056
Epoch[8] Batch [749]	Speed: 4646.253117 samples/sec	accuracy=0.363263	lr=0.399034
Epoch[8] Batch [799]	Speed: 5619.884079 samples/sec	accuracy=0.363373	lr=0.399013
Epoch[8] Batch [849]	Speed: 4769.522159 samples/sec	accuracy=0.363155	lr=0.398991
Epoch[8] Batch [899]	Speed: 5438.303360 samples/sec	accuracy=0.363453	lr=0.398969
Epoch[8] Batch [949]	Speed: 5253.920802 samples/sec	accuracy=0.363448	lr=0.398947
Epoch[8] Batch [999]	Speed: 4905.057012 samples/sec	accuracy=0.363354	lr=0.398924
Epoch[8] Batch [1049]	Speed: 5035.310629 samples/sec	accuracy=0.363400	lr=0.398901
Epoch[8] Batch [1099]	Speed: 5309.466229 samples/sec	accuracy=0.363636	lr=0.398878
Epoch[8] Batch [1149]	Speed: 5489.070718 samples/sec	accuracy=0.363669	lr=0.398855
Epoch[8] Batch [1199]	Speed: 5278.921917 samples/sec	accuracy=0.363818	lr=0.398832
Epoch[8] Batch [1249]	Speed: 4634.222206 samples/sec	accuracy=0.364075	lr=0.398808
[Epoch 8] training: accuracy=0.364078
[Epoch 8] speed: 5085 samples/sec	time cost: 261.955386
[Epoch 8] validation: err-top1=0.706254 err-top5=0.467813
Epoch[9] Batch [49]	Speed: 4857.510760 samples/sec	accuracy=0.379453	lr=0.398784
Epoch[9] Batch [99]	Speed: 5036.236314 samples/sec	accuracy=0.376230	lr=0.398760
Epoch[9] Batch [149]	Speed: 4905.022617 samples/sec	accuracy=0.373639	lr=0.398735
Epoch[9] Batch [199]	Speed: 4827.945436 samples/sec	accuracy=0.373491	lr=0.398710
Epoch[9] Batch [249]	Speed: 5210.632216 samples/sec	accuracy=0.373312	lr=0.398686
Epoch[9] Batch [299]	Speed: 4779.995122 samples/sec	accuracy=0.373359	lr=0.398660
Epoch[9] Batch [349]	Speed: 4845.518828 samples/sec	accuracy=0.373524	lr=0.398635
Epoch[9] Batch [399]	Speed: 5513.013347 samples/sec	accuracy=0.373792	lr=0.398610
Epoch[9] Batch [449]	Speed: 5515.494067 samples/sec	accuracy=0.373381	lr=0.398584
Epoch[9] Batch [499]	Speed: 5123.871089 samples/sec	accuracy=0.373199	lr=0.398558
Epoch[9] Batch [549]	Speed: 5096.519797 samples/sec	accuracy=0.373100	lr=0.398531
Epoch[9] Batch [599]	Speed: 4719.530398 samples/sec	accuracy=0.373045	lr=0.398505
Epoch[9] Batch [649]	Speed: 5405.484115 samples/sec	accuracy=0.372800	lr=0.398478
Epoch[9] Batch [699]	Speed: 4618.866412 samples/sec	accuracy=0.372436	lr=0.398451
Epoch[9] Batch [749]	Speed: 5199.439672 samples/sec	accuracy=0.372355	lr=0.398424
Epoch[9] Batch [799]	Speed: 5487.587132 samples/sec	accuracy=0.372271	lr=0.398396
Epoch[9] Batch [849]	Speed: 4723.528710 samples/sec	accuracy=0.372111	lr=0.398369
Epoch[9] Batch [899]	Speed: 4822.116219 samples/sec	accuracy=0.372204	lr=0.398341
Epoch[9] Batch [949]	Speed: 4879.379786 samples/sec	accuracy=0.372199	lr=0.398312
Epoch[9] Batch [999]	Speed: 5199.930176 samples/sec	accuracy=0.372209	lr=0.398284
Epoch[9] Batch [1049]	Speed: 5325.040724 samples/sec	accuracy=0.372079	lr=0.398255
Epoch[9] Batch [1099]	Speed: 5165.639439 samples/sec	accuracy=0.372166	lr=0.398226
Epoch[9] Batch [1149]	Speed: 5842.033286 samples/sec	accuracy=0.372251	lr=0.398197
Epoch[9] Batch [1199]	Speed: 5434.566281 samples/sec	accuracy=0.372402	lr=0.398168
Epoch[9] Batch [1249]	Speed: 4996.032452 samples/sec	accuracy=0.372475	lr=0.398138
[Epoch 9] training: accuracy=0.372480
[Epoch 9] speed: 5074 samples/sec	time cost: 261.339471
[Epoch 9] validation: err-top1=0.617845 err-top5=0.353695
Epoch[10] Batch [49]	Speed: 4317.855905 samples/sec	accuracy=0.379551	lr=0.398108
Epoch[10] Batch [99]	Speed: 5133.617911 samples/sec	accuracy=0.380791	lr=0.398078
Epoch[10] Batch [149]	Speed: 4741.632491 samples/sec	accuracy=0.380137	lr=0.398048
Epoch[10] Batch [199]	Speed: 5152.167018 samples/sec	accuracy=0.379194	lr=0.398017
Epoch[10] Batch [249]	Speed: 5736.274381 samples/sec	accuracy=0.378625	lr=0.397986
Epoch[10] Batch [299]	Speed: 5128.685853 samples/sec	accuracy=0.378701	lr=0.397955
Epoch[10] Batch [349]	Speed: 5877.038254 samples/sec	accuracy=0.378504	lr=0.397924
Epoch[10] Batch [399]	Speed: 5008.732983 samples/sec	accuracy=0.378318	lr=0.397892
Epoch[10] Batch [449]	Speed: 4408.452049 samples/sec	accuracy=0.378197	lr=0.397861
Epoch[10] Batch [499]	Speed: 5495.493311 samples/sec	accuracy=0.378129	lr=0.397829
Epoch[10] Batch [549]	Speed: 5110.057796 samples/sec	accuracy=0.378006	lr=0.397796
Epoch[10] Batch [599]	Speed: 5627.335258 samples/sec	accuracy=0.378322	lr=0.397764
Epoch[10] Batch [649]	Speed: 4717.041986 samples/sec	accuracy=0.378498	lr=0.397731
Epoch[10] Batch [699]	Speed: 4981.842520 samples/sec	accuracy=0.378622	lr=0.397698
Epoch[10] Batch [749]	Speed: 5300.192210 samples/sec	accuracy=0.378897	lr=0.397665
Epoch[10] Batch [799]	Speed: 4931.075209 samples/sec	accuracy=0.378811	lr=0.397632
Epoch[10] Batch [849]	Speed: 5071.937384 samples/sec	accuracy=0.378725	lr=0.397598
Epoch[10] Batch [899]	Speed: 4654.126519 samples/sec	accuracy=0.378634	lr=0.397564
Epoch[10] Batch [949]	Speed: 4471.995812 samples/sec	accuracy=0.378716	lr=0.397530
Epoch[10] Batch [999]	Speed: 5015.351291 samples/sec	accuracy=0.378899	lr=0.397496
Epoch[10] Batch [1049]	Speed: 5727.265431 samples/sec	accuracy=0.378866	lr=0.397461
Epoch[10] Batch [1099]	Speed: 5841.594044 samples/sec	accuracy=0.378842	lr=0.397427
Epoch[10] Batch [1149]	Speed: 5640.271557 samples/sec	accuracy=0.378797	lr=0.397392
Epoch[10] Batch [1199]	Speed: 5019.366248 samples/sec	accuracy=0.378645	lr=0.397356
Epoch[10] Batch [1249]	Speed: 5366.630490 samples/sec	accuracy=0.378851	lr=0.397321
[Epoch 10] training: accuracy=0.378842
[Epoch 10] speed: 5094 samples/sec	time cost: 261.067698
[Epoch 10] validation: err-top1=0.628827 err-top5=0.368742
Epoch[11] Batch [49]	Speed: 5503.151383 samples/sec	accuracy=0.387480	lr=0.397284
Epoch[11] Batch [99]	Speed: 5435.304234 samples/sec	accuracy=0.388369	lr=0.397248
Epoch[11] Batch [149]	Speed: 5942.905849 samples/sec	accuracy=0.386771	lr=0.397212
Epoch[11] Batch [199]	Speed: 5171.072164 samples/sec	accuracy=0.386455	lr=0.397176
Epoch[11] Batch [249]	Speed: 5833.308677 samples/sec	accuracy=0.385691	lr=0.397139
Epoch[11] Batch [299]	Speed: 5612.401267 samples/sec	accuracy=0.385430	lr=0.397102
Epoch[11] Batch [349]	Speed: 5404.718462 samples/sec	accuracy=0.385653	lr=0.397065
Epoch[11] Batch [399]	Speed: 5477.273193 samples/sec	accuracy=0.385264	lr=0.397028
Epoch[11] Batch [449]	Speed: 5851.847288 samples/sec	accuracy=0.385024	lr=0.396990
Epoch[11] Batch [499]	Speed: 5422.574809 samples/sec	accuracy=0.384824	lr=0.396952
Epoch[11] Batch [549]	Speed: 5820.690700 samples/sec	accuracy=0.385103	lr=0.396914
Epoch[11] Batch [599]	Speed: 5689.455291 samples/sec	accuracy=0.385039	lr=0.396876
Epoch[11] Batch [649]	Speed: 5646.410307 samples/sec	accuracy=0.384769	lr=0.396837
Epoch[11] Batch [699]	Speed: 5975.811534 samples/sec	accuracy=0.384788	lr=0.396798
Epoch[11] Batch [749]	Speed: 5253.009999 samples/sec	accuracy=0.384695	lr=0.396759
Epoch[11] Batch [799]	Speed: 5472.188478 samples/sec	accuracy=0.384662	lr=0.396720
Epoch[11] Batch [849]	Speed: 5629.731627 samples/sec	accuracy=0.384601	lr=0.396680
Epoch[11] Batch [899]	Speed: 5339.796126 samples/sec	accuracy=0.384639	lr=0.396641
Epoch[11] Batch [949]	Speed: 4747.703140 samples/sec	accuracy=0.384349	lr=0.396601
Epoch[11] Batch [999]	Speed: 5729.649080 samples/sec	accuracy=0.384466	lr=0.396561
Epoch[11] Batch [1049]	Speed: 5576.997570 samples/sec	accuracy=0.384483	lr=0.396520
Epoch[11] Batch [1099]	Speed: 5170.102975 samples/sec	accuracy=0.384583	lr=0.396479
Epoch[11] Batch [1149]	Speed: 4025.286391 samples/sec	accuracy=0.384513	lr=0.396439
Epoch[11] Batch [1199]	Speed: 4110.055331 samples/sec	accuracy=0.384484	lr=0.396397
Epoch[11] Batch [1249]	Speed: 3400.980092 samples/sec	accuracy=0.384458	lr=0.396356
[Epoch 11] training: accuracy=0.384471
[Epoch 11] speed: 5231 samples/sec	time cost: 255.184449
[Epoch 11] validation: err-top1=0.592265 err-top5=0.327311
Epoch[12] Batch [49]	Speed: 5016.818555 samples/sec	accuracy=0.393613	lr=0.396314
Epoch[12] Batch [99]	Speed: 4634.099402 samples/sec	accuracy=0.390293	lr=0.396272
Epoch[12] Batch [149]	Speed: 4260.288716 samples/sec	accuracy=0.390911	lr=0.396230
Epoch[12] Batch [199]	Speed: 4880.275528 samples/sec	accuracy=0.390493	lr=0.396187
Epoch[12] Batch [249]	Speed: 3826.860854 samples/sec	accuracy=0.390590	lr=0.396145
Epoch[12] Batch [299]	Speed: 4006.182959 samples/sec	accuracy=0.390479	lr=0.396102
Epoch[12] Batch [349]	Speed: 4483.791223 samples/sec	accuracy=0.390544	lr=0.396059
Epoch[12] Batch [399]	Speed: 3692.593614 samples/sec	accuracy=0.389995	lr=0.396016
Epoch[12] Batch [449]	Speed: 3319.325156 samples/sec	accuracy=0.389826	lr=0.395972
Epoch[12] Batch [499]	Speed: 3762.284145 samples/sec	accuracy=0.389652	lr=0.395928
Epoch[12] Batch [549]	Speed: 5487.240091 samples/sec	accuracy=0.389688	lr=0.395885
Epoch[12] Batch [599]	Speed: 5616.370415 samples/sec	accuracy=0.389867	lr=0.395840
Epoch[12] Batch [649]	Speed: 5353.226785 samples/sec	accuracy=0.389856	lr=0.395796
Epoch[12] Batch [699]	Speed: 4555.159366 samples/sec	accuracy=0.389937	lr=0.395751
Epoch[12] Batch [749]	Speed: 3780.409619 samples/sec	accuracy=0.389840	lr=0.395706
Epoch[12] Batch [799]	Speed: 5099.252628 samples/sec	accuracy=0.389694	lr=0.395661
Epoch[12] Batch [849]	Speed: 4446.512996 samples/sec	accuracy=0.389474	lr=0.395616
Epoch[12] Batch [899]	Speed: 3144.555778 samples/sec	accuracy=0.389183	lr=0.395570
Epoch[12] Batch [949]	Speed: 5580.240409 samples/sec	accuracy=0.389067	lr=0.395524
Epoch[12] Batch [999]	Speed: 3919.052558 samples/sec	accuracy=0.389112	lr=0.395478
Epoch[12] Batch [1049]	Speed: 3981.026640 samples/sec	accuracy=0.389207	lr=0.395432
Epoch[12] Batch [1099]	Speed: 5573.126958 samples/sec	accuracy=0.389019	lr=0.395386
Epoch[12] Batch [1149]	Speed: 3833.291974 samples/sec	accuracy=0.388875	lr=0.395339
Epoch[12] Batch [1199]	Speed: 4065.408554 samples/sec	accuracy=0.388849	lr=0.395292
Epoch[12] Batch [1249]	Speed: 4775.805733 samples/sec	accuracy=0.388989	lr=0.395245
[Epoch 12] training: accuracy=0.388971
[Epoch 12] speed: 4323 samples/sec	time cost: 306.617165
[Epoch 12] validation: err-top1=0.590262 err-top5=0.329739
Epoch[13] Batch [49]	Speed: 3963.297113 samples/sec	accuracy=0.394004	lr=0.395196
Epoch[13] Batch [99]	Speed: 4273.925564 samples/sec	accuracy=0.395244	lr=0.395149
Epoch[13] Batch [149]	Speed: 5009.216674 samples/sec	accuracy=0.396042	lr=0.395101
Epoch[13] Batch [199]	Speed: 5461.088067 samples/sec	accuracy=0.395591	lr=0.395052
Epoch[13] Batch [249]	Speed: 4695.594786 samples/sec	accuracy=0.395797	lr=0.395004
Epoch[13] Batch [299]	Speed: 5017.715411 samples/sec	accuracy=0.395371	lr=0.394955
Epoch[13] Batch [349]	Speed: 5141.434575 samples/sec	accuracy=0.395533	lr=0.394907
Epoch[13] Batch [399]	Speed: 4707.368454 samples/sec	accuracy=0.395227	lr=0.394858
Epoch[13] Batch [449]	Speed: 3826.894884 samples/sec	accuracy=0.395482	lr=0.394808
Epoch[13] Batch [499]	Speed: 3835.932939 samples/sec	accuracy=0.395887	lr=0.394759
Epoch[13] Batch [549]	Speed: 3489.749991 samples/sec	accuracy=0.395648	lr=0.394709
Epoch[13] Batch [599]	Speed: 3972.381404 samples/sec	accuracy=0.395464	lr=0.394659
Epoch[13] Batch [649]	Speed: 5453.144104 samples/sec	accuracy=0.395000	lr=0.394609
Epoch[13] Batch [699]	Speed: 4612.487920 samples/sec	accuracy=0.395043	lr=0.394558
Epoch[13] Batch [749]	Speed: 3968.685251 samples/sec	accuracy=0.394956	lr=0.394507
Epoch[13] Batch [799]	Speed: 5772.952623 samples/sec	accuracy=0.394567	lr=0.394457
Epoch[13] Batch [849]	Speed: 4664.018891 samples/sec	accuracy=0.394421	lr=0.394405
Epoch[13] Batch [899]	Speed: 4989.703933 samples/sec	accuracy=0.394248	lr=0.394354
Epoch[13] Batch [949]	Speed: 4390.129230 samples/sec	accuracy=0.394113	lr=0.394302
Epoch[13] Batch [999]	Speed: 5119.126248 samples/sec	accuracy=0.394192	lr=0.394250
Epoch[13] Batch [1049]	Speed: 5476.304819 samples/sec	accuracy=0.394055	lr=0.394198
Epoch[13] Batch [1099]	Speed: 5187.627963 samples/sec	accuracy=0.394152	lr=0.394146
Epoch[13] Batch [1149]	Speed: 5802.864343 samples/sec	accuracy=0.394152	lr=0.394093
Epoch[13] Batch [1199]	Speed: 4965.066125 samples/sec	accuracy=0.394141	lr=0.394041
Epoch[13] Batch [1249]	Speed: 5144.152338 samples/sec	accuracy=0.394027	lr=0.393988
[Epoch 13] training: accuracy=0.394020
[Epoch 13] speed: 4664 samples/sec	time cost: 284.676643
[Epoch 13] validation: err-top1=0.750060 err-top5=0.523338
Epoch[14] Batch [49]	Speed: 4794.697094 samples/sec	accuracy=0.399551	lr=0.393933
Epoch[14] Batch [99]	Speed: 4903.176646 samples/sec	accuracy=0.401475	lr=0.393880
Epoch[14] Batch [149]	Speed: 4707.564622 samples/sec	accuracy=0.400553	lr=0.393826
Epoch[14] Batch [199]	Speed: 5840.718936 samples/sec	accuracy=0.400313	lr=0.393772
Epoch[14] Batch [249]	Speed: 5736.282042 samples/sec	accuracy=0.400074	lr=0.393718
Epoch[14] Batch [299]	Speed: 5535.287794 samples/sec	accuracy=0.399365	lr=0.393664
Epoch[14] Batch [349]	Speed: 5523.925346 samples/sec	accuracy=0.398485	lr=0.393609
Epoch[14] Batch [399]	Speed: 5499.403827 samples/sec	accuracy=0.398567	lr=0.393554
Epoch[14] Batch [449]	Speed: 3913.502121 samples/sec	accuracy=0.398405	lr=0.393499
Epoch[14] Batch [499]	Speed: 4530.195237 samples/sec	accuracy=0.398131	lr=0.393444
Epoch[14] Batch [549]	Speed: 4461.592147 samples/sec	accuracy=0.398168	lr=0.393388
Epoch[14] Batch [599]	Speed: 4945.537297 samples/sec	accuracy=0.397983	lr=0.393332
Epoch[14] Batch [649]	Speed: 5261.263337 samples/sec	accuracy=0.397946	lr=0.393276
Epoch[14] Batch [699]	Speed: 4160.176576 samples/sec	accuracy=0.397913	lr=0.393220
Epoch[14] Batch [749]	Speed: 5175.352516 samples/sec	accuracy=0.397699	lr=0.393163
Epoch[14] Batch [799]	Speed: 4777.679792 samples/sec	accuracy=0.397727	lr=0.393107
Epoch[14] Batch [849]	Speed: 5002.379026 samples/sec	accuracy=0.397624	lr=0.393050
Epoch[14] Batch [899]	Speed: 5022.818030 samples/sec	accuracy=0.397635	lr=0.392993
Epoch[14] Batch [949]	Speed: 5326.492142 samples/sec	accuracy=0.397729	lr=0.392935
Epoch[14] Batch [999]	Speed: 5518.713833 samples/sec	accuracy=0.397792	lr=0.392877
Epoch[14] Batch [1049]	Speed: 5061.901452 samples/sec	accuracy=0.397887	lr=0.392820
Epoch[14] Batch [1099]	Speed: 5519.456375 samples/sec	accuracy=0.397830	lr=0.392762
Epoch[14] Batch [1149]	Speed: 5628.620228 samples/sec	accuracy=0.397744	lr=0.392703
Epoch[14] Batch [1199]	Speed: 5307.322374 samples/sec	accuracy=0.397738	lr=0.392645
Epoch[14] Batch [1249]	Speed: 5483.151820 samples/sec	accuracy=0.397684	lr=0.392586
[Epoch 14] training: accuracy=0.397662
[Epoch 14] speed: 5050 samples/sec	time cost: 263.656569
[Epoch 14] validation: err-top1=0.670978 err-top5=0.412528
Epoch[15] Batch [49]	Speed: 5367.150365 samples/sec	accuracy=0.407754	lr=0.392526
Epoch[15] Batch [99]	Speed: 5702.256180 samples/sec	accuracy=0.403965	lr=0.392466
Epoch[15] Batch [149]	Speed: 5473.386403 samples/sec	accuracy=0.403724	lr=0.392407
Epoch[15] Batch [199]	Speed: 5190.779327 samples/sec	accuracy=0.403516	lr=0.392347
Epoch[15] Batch [249]	Speed: 5906.106975 samples/sec	accuracy=0.402762	lr=0.392287
Epoch[15] Batch [299]	Speed: 5103.781761 samples/sec	accuracy=0.402064	lr=0.392227
Epoch[15] Batch [349]	Speed: 5256.651374 samples/sec	accuracy=0.401473	lr=0.392167
Epoch[15] Batch [399]	Speed: 5151.520130 samples/sec	accuracy=0.401550	lr=0.392106
Epoch[15] Batch [449]	Speed: 5318.494159 samples/sec	accuracy=0.401576	lr=0.392045
Epoch[15] Batch [499]	Speed: 5320.474892 samples/sec	accuracy=0.401121	lr=0.391984
Epoch[15] Batch [549]	Speed: 5382.417599 samples/sec	accuracy=0.400936	lr=0.391923
Epoch[15] Batch [599]	Speed: 5743.482557 samples/sec	accuracy=0.400711	lr=0.391861
Epoch[15] Batch [649]	Speed: 5741.097209 samples/sec	accuracy=0.400849	lr=0.391799
Epoch[15] Batch [699]	Speed: 5051.072549 samples/sec	accuracy=0.400735	lr=0.391737
Epoch[15] Batch [749]	Speed: 5757.069824 samples/sec	accuracy=0.400637	lr=0.391675
Epoch[15] Batch [799]	Speed: 5414.270477 samples/sec	accuracy=0.400731	lr=0.391613
Epoch[15] Batch [849]	Speed: 5109.689749 samples/sec	accuracy=0.400692	lr=0.391550
Epoch[15] Batch [899]	Speed: 5643.843529 samples/sec	accuracy=0.400611	lr=0.391487
Epoch[15] Batch [949]	Speed: 5119.138939 samples/sec	accuracy=0.400652	lr=0.391424
Epoch[15] Batch [999]	Speed: 5749.914662 samples/sec	accuracy=0.400549	lr=0.391361
Epoch[15] Batch [1049]	Speed: 5663.582853 samples/sec	accuracy=0.400439	lr=0.391297
Epoch[15] Batch [1099]	Speed: 5725.836408 samples/sec	accuracy=0.400216	lr=0.391233
Epoch[15] Batch [1149]	Speed: 5421.661813 samples/sec	accuracy=0.400298	lr=0.391169
Epoch[15] Batch [1199]	Speed: 5373.250995 samples/sec	accuracy=0.400250	lr=0.391105
Epoch[15] Batch [1249]	Speed: 5336.899597 samples/sec	accuracy=0.400208	lr=0.391040
[Epoch 15] training: accuracy=0.400200
[Epoch 15] speed: 5421 samples/sec	time cost: 246.241286
[Epoch 15] validation: err-top1=0.698561 err-top5=0.450654
Epoch[16] Batch [49]	Speed: 5026.050112 samples/sec	accuracy=0.408516	lr=0.390974
Epoch[16] Batch [99]	Speed: 5319.738138 samples/sec	accuracy=0.409473	lr=0.390909
Epoch[16] Batch [149]	Speed: 5159.995680 samples/sec	accuracy=0.408366	lr=0.390844
Epoch[16] Batch [199]	Speed: 5415.364243 samples/sec	accuracy=0.408169	lr=0.390779
Epoch[16] Batch [249]	Speed: 5395.741786 samples/sec	accuracy=0.406941	lr=0.390713
Epoch[16] Batch [299]	Speed: 5501.428891 samples/sec	accuracy=0.405928	lr=0.390647
Epoch[16] Batch [349]	Speed: 5469.482419 samples/sec	accuracy=0.405594	lr=0.390581
Epoch[16] Batch [399]	Speed: 5081.783974 samples/sec	accuracy=0.405122	lr=0.390515
Epoch[16] Batch [449]	Speed: 5814.846218 samples/sec	accuracy=0.404946	lr=0.390448
Epoch[16] Batch [499]	Speed: 5334.286834 samples/sec	accuracy=0.404629	lr=0.390381
Epoch[16] Batch [549]	Speed: 5164.756128 samples/sec	accuracy=0.404561	lr=0.390314
Epoch[16] Batch [599]	Speed: 5139.371846 samples/sec	accuracy=0.404414	lr=0.390247
Epoch[16] Batch [649]	Speed: 4506.587260 samples/sec	accuracy=0.404414	lr=0.390180
Epoch[16] Batch [699]	Speed: 5826.824080 samples/sec	accuracy=0.404273	lr=0.390112
Epoch[16] Batch [749]	Speed: 5424.730310 samples/sec	accuracy=0.404048	lr=0.390044
Epoch[16] Batch [799]	Speed: 3184.465184 samples/sec	accuracy=0.404127	lr=0.389976
Epoch[16] Batch [849]	Speed: 4178.146731 samples/sec	accuracy=0.404027	lr=0.389907
Epoch[16] Batch [899]	Speed: 4240.899977 samples/sec	accuracy=0.403882	lr=0.389839
Epoch[16] Batch [949]	Speed: 3928.144728 samples/sec	accuracy=0.403486	lr=0.389770
Epoch[16] Batch [999]	Speed: 5697.209357 samples/sec	accuracy=0.403195	lr=0.389701
Epoch[16] Batch [1049]	Speed: 4388.154227 samples/sec	accuracy=0.402906	lr=0.389632
Epoch[16] Batch [1099]	Speed: 4615.205515 samples/sec	accuracy=0.402879	lr=0.389562
Epoch[16] Batch [1149]	Speed: 2564.562469 samples/sec	accuracy=0.402732	lr=0.389492
Epoch[16] Batch [1199]	Speed: 3467.533440 samples/sec	accuracy=0.402681	lr=0.389422
Epoch[16] Batch [1249]	Speed: 3496.073978 samples/sec	accuracy=0.402649	lr=0.389352
[Epoch 16] training: accuracy=0.402644
[Epoch 16] speed: 4568 samples/sec	time cost: 290.283763
[Epoch 16] validation: err-top1=0.584722 err-top5=0.323441
Epoch[17] Batch [49]	Speed: 3012.532762 samples/sec	accuracy=0.414180	lr=0.389280
Epoch[17] Batch [99]	Speed: 4485.264604 samples/sec	accuracy=0.409795	lr=0.389210
Epoch[17] Batch [149]	Speed: 1162.256074 samples/sec	accuracy=0.408008	lr=0.389139
Epoch[17] Batch [199]	Speed: 4032.860936 samples/sec	accuracy=0.408008	lr=0.389068
Epoch[17] Batch [249]	Speed: 2299.883324 samples/sec	accuracy=0.407418	lr=0.388996
Epoch[17] Batch [299]	Speed: 2933.473249 samples/sec	accuracy=0.406901	lr=0.388925
Epoch[17] Batch [349]	Speed: 3720.118126 samples/sec	accuracy=0.406844	lr=0.388853
Epoch[17] Batch [399]	Speed: 2876.338737 samples/sec	accuracy=0.406721	lr=0.388781
Epoch[17] Batch [449]	Speed: 4920.754282 samples/sec	accuracy=0.406582	lr=0.388709
Epoch[17] Batch [499]	Speed: 5764.259216 samples/sec	accuracy=0.406361	lr=0.388636
Epoch[17] Batch [549]	Speed: 6061.133461 samples/sec	accuracy=0.406261	lr=0.388564
Epoch[17] Batch [599]	Speed: 2345.040148 samples/sec	accuracy=0.406260	lr=0.388491
Epoch[17] Batch [649]	Speed: 2078.272860 samples/sec	accuracy=0.406258	lr=0.388418
Epoch[17] Batch [699]	Speed: 5558.062730 samples/sec	accuracy=0.406073	lr=0.388344
Epoch[17] Batch [749]	Speed: 5420.151781 samples/sec	accuracy=0.405858	lr=0.388271
Epoch[17] Batch [799]	Speed: 5503.815687 samples/sec	accuracy=0.405580	lr=0.388197
Epoch[17] Batch [849]	Speed: 5850.404197 samples/sec	accuracy=0.405387	lr=0.388123
Epoch[17] Batch [899]	Speed: 5561.055030 samples/sec	accuracy=0.405179	lr=0.388049
Epoch[17] Batch [949]	Speed: 6034.441194 samples/sec	accuracy=0.405206	lr=0.387974
Epoch[17] Batch [999]	Speed: 5931.894941 samples/sec	accuracy=0.405211	lr=0.387900
Epoch[17] Batch [1049]	Speed: 5730.382498 samples/sec	accuracy=0.405189	lr=0.387825
Epoch[17] Batch [1099]	Speed: 5672.386125 samples/sec	accuracy=0.405076	lr=0.387750
Epoch[17] Batch [1149]	Speed: 5654.504605 samples/sec	accuracy=0.404977	lr=0.387674
Epoch[17] Batch [1199]	Speed: 5890.547737 samples/sec	accuracy=0.405184	lr=0.387599
Epoch[17] Batch [1249]	Speed: 5977.954273 samples/sec	accuracy=0.404992	lr=0.387523
[Epoch 17] training: accuracy=0.404997
[Epoch 17] speed: 3783 samples/sec	time cost: 347.963562
[Epoch 17] validation: err-top1=0.655986 err-top5=0.405599
Epoch[18] Batch [49]	Speed: 5360.475831 samples/sec	accuracy=0.403770	lr=0.387445
Epoch[18] Batch [99]	Speed: 5507.938373 samples/sec	accuracy=0.405693	lr=0.387369
Epoch[18] Batch [149]	Speed: 5695.371570 samples/sec	accuracy=0.406921	lr=0.387293
Epoch[18] Batch [199]	Speed: 5797.930494 samples/sec	accuracy=0.407993	lr=0.387216
Epoch[18] Batch [249]	Speed: 5720.377629 samples/sec	accuracy=0.408102	lr=0.387139
Epoch[18] Batch [299]	Speed: 5695.505100 samples/sec	accuracy=0.408405	lr=0.387062
Epoch[18] Batch [349]	Speed: 5586.511709 samples/sec	accuracy=0.408326	lr=0.386984
Epoch[18] Batch [399]	Speed: 5700.775593 samples/sec	accuracy=0.408582	lr=0.386907
Epoch[18] Batch [449]	Speed: 5701.509508 samples/sec	accuracy=0.408709	lr=0.386829
Epoch[18] Batch [499]	Speed: 5845.578240 samples/sec	accuracy=0.408586	lr=0.386751
Epoch[18] Batch [549]	Speed: 4397.912488 samples/sec	accuracy=0.408240	lr=0.386673
Epoch[18] Batch [599]	Speed: 4736.275862 samples/sec	accuracy=0.408278	lr=0.386594
Epoch[18] Batch [649]	Speed: 3586.233282 samples/sec	accuracy=0.408251	lr=0.386515
Epoch[18] Batch [699]	Speed: 5594.774808 samples/sec	accuracy=0.408318	lr=0.386436
Epoch[18] Batch [749]	Speed: 5376.397637 samples/sec	accuracy=0.408277	lr=0.386357
Epoch[18] Batch [799]	Speed: 4854.769835 samples/sec	accuracy=0.408416	lr=0.386278
Epoch[18] Batch [849]	Speed: 3044.016398 samples/sec	accuracy=0.408381	lr=0.386198
Epoch[18] Batch [899]	Speed: 4094.033601 samples/sec	accuracy=0.408218	lr=0.386119
Epoch[18] Batch [949]	Speed: 3994.413708 samples/sec	accuracy=0.408302	lr=0.386038
Epoch[18] Batch [999]	Speed: 4023.603487 samples/sec	accuracy=0.408259	lr=0.385958
Epoch[18] Batch [1049]	Speed: 3823.031559 samples/sec	accuracy=0.408265	lr=0.385878
Epoch[18] Batch [1099]	Speed: 5485.609236 samples/sec	accuracy=0.408269	lr=0.385797
Epoch[18] Batch [1149]	Speed: 5948.524767 samples/sec	accuracy=0.408364	lr=0.385716
Epoch[18] Batch [1199]	Speed: 3993.480299 samples/sec	accuracy=0.408317	lr=0.385635
Epoch[18] Batch [1249]	Speed: 3264.259577 samples/sec	accuracy=0.408488	lr=0.385554
[Epoch 18] training: accuracy=0.408488
[Epoch 18] speed: 4715 samples/sec	time cost: 287.747651
[Epoch 18] validation: err-top1=0.627252 err-top5=0.363700
Epoch[19] Batch [49]	Speed: 4623.488634 samples/sec	accuracy=0.416465	lr=0.385470
Epoch[19] Batch [99]	Speed: 3728.428227 samples/sec	accuracy=0.415654	lr=0.385388
Epoch[19] Batch [149]	Speed: 4188.312423 samples/sec	accuracy=0.414238	lr=0.385306
Epoch[19] Batch [199]	Speed: 5124.081865 samples/sec	accuracy=0.413828	lr=0.385224
Epoch[19] Batch [249]	Speed: 3959.619590 samples/sec	accuracy=0.413871	lr=0.385142
Epoch[19] Batch [299]	Speed: 5106.597310 samples/sec	accuracy=0.413698	lr=0.385059
Epoch[19] Batch [349]	Speed: 4679.834345 samples/sec	accuracy=0.413170	lr=0.384976
Epoch[19] Batch [399]	Speed: 4023.794453 samples/sec	accuracy=0.412754	lr=0.384893
Epoch[19] Batch [449]	Speed: 4663.009699 samples/sec	accuracy=0.412389	lr=0.384810
Epoch[19] Batch [499]	Speed: 4191.082596 samples/sec	accuracy=0.412145	lr=0.384726
Epoch[19] Batch [549]	Speed: 3917.896194 samples/sec	accuracy=0.412159	lr=0.384642
Epoch[19] Batch [599]	Speed: 4556.214919 samples/sec	accuracy=0.412262	lr=0.384558
Epoch[19] Batch [649]	Speed: 3454.001403 samples/sec	accuracy=0.411686	lr=0.384474
Epoch[19] Batch [699]	Speed: 3997.065002 samples/sec	accuracy=0.411292	lr=0.384389
Epoch[19] Batch [749]	Speed: 4190.900776 samples/sec	accuracy=0.411048	lr=0.384305
Epoch[19] Batch [799]	Speed: 5509.410088 samples/sec	accuracy=0.410692	lr=0.384220
Epoch[19] Batch [849]	Speed: 4297.638622 samples/sec	accuracy=0.410692	lr=0.384135
Epoch[19] Batch [899]	Speed: 4010.654561 samples/sec	accuracy=0.410180	lr=0.384049
Epoch[19] Batch [949]	Speed: 3480.604825 samples/sec	accuracy=0.410050	lr=0.383964
Epoch[19] Batch [999]	Speed: 4665.504764 samples/sec	accuracy=0.409841	lr=0.383878
Epoch[19] Batch [1049]	Speed: 4638.679032 samples/sec	accuracy=0.409650	lr=0.383792
Epoch[19] Batch [1099]	Speed: 6049.608258 samples/sec	accuracy=0.409565	lr=0.383706
Epoch[19] Batch [1149]	Speed: 5709.361454 samples/sec	accuracy=0.409454	lr=0.383619
Epoch[19] Batch [1199]	Speed: 4702.260553 samples/sec	accuracy=0.409179	lr=0.383533
Epoch[19] Batch [1249]	Speed: 3411.850796 samples/sec	accuracy=0.409273	lr=0.383446
[Epoch 19] training: accuracy=0.409294
[Epoch 19] speed: 4333 samples/sec	time cost: 308.000921
[Epoch 19] validation: err-top1=0.571169 err-top5=0.310547
Epoch[20] Batch [49]	Speed: 2440.351861 samples/sec	accuracy=0.410684	lr=0.383357
Epoch[20] Batch [99]	Speed: 4716.543977 samples/sec	accuracy=0.410840	lr=0.383270
Epoch[20] Batch [149]	Speed: 4817.328211 samples/sec	accuracy=0.412292	lr=0.383182
Epoch[20] Batch [199]	Speed: 4793.519603 samples/sec	accuracy=0.412632	lr=0.383094
Epoch[20] Batch [249]	Speed: 4792.335098 samples/sec	accuracy=0.413426	lr=0.383006
Epoch[20] Batch [299]	Speed: 3302.017892 samples/sec	accuracy=0.412523	lr=0.382918
Epoch[20] Batch [349]	Speed: 2955.115016 samples/sec	accuracy=0.412313	lr=0.382830
Epoch[20] Batch [399]	Speed: 4002.404868 samples/sec	accuracy=0.411868	lr=0.382741
Epoch[20] Batch [449]	Speed: 2550.387457 samples/sec	accuracy=0.411875	lr=0.382652
Epoch[20] Batch [499]	Speed: 4334.572108 samples/sec	accuracy=0.411824	lr=0.382563
Epoch[20] Batch [549]	Speed: 4600.298787 samples/sec	accuracy=0.411942	lr=0.382474
Epoch[20] Batch [599]	Speed: 4816.270061 samples/sec	accuracy=0.411619	lr=0.382384
Epoch[20] Batch [649]	Speed: 3422.592377 samples/sec	accuracy=0.411532	lr=0.382295
Epoch[20] Batch [699]	Speed: 4852.484913 samples/sec	accuracy=0.411313	lr=0.382205
Epoch[20] Batch [749]	Speed: 4278.666518 samples/sec	accuracy=0.411354	lr=0.382115
Epoch[20] Batch [799]	Speed: 4706.370741 samples/sec	accuracy=0.411279	lr=0.382024
Epoch[20] Batch [849]	Speed: 5161.938759 samples/sec	accuracy=0.411225	lr=0.381934
Epoch[20] Batch [899]	Speed: 5496.593131 samples/sec	accuracy=0.411221	lr=0.381843
Epoch[20] Batch [949]	Speed: 3256.180077 samples/sec	accuracy=0.411150	lr=0.381752
Epoch[20] Batch [999]	Speed: 3518.857052 samples/sec	accuracy=0.411144	lr=0.381661
Epoch[20] Batch [1049]	Speed: 5690.491174 samples/sec	accuracy=0.411213	lr=0.381569
Epoch[20] Batch [1099]	Speed: 5311.841475 samples/sec	accuracy=0.411202	lr=0.381477
Epoch[20] Batch [1149]	Speed: 4994.219099 samples/sec	accuracy=0.411221	lr=0.381385
Epoch[20] Batch [1199]	Speed: 4797.338222 samples/sec	accuracy=0.410977	lr=0.381293
Epoch[20] Batch [1249]	Speed: 3971.038121 samples/sec	accuracy=0.410927	lr=0.381201
[Epoch 20] training: accuracy=0.410915
[Epoch 20] speed: 4081 samples/sec	time cost: 324.579560
[Epoch 20] validation: err-top1=0.580895 err-top5=0.314632
Epoch[21] Batch [49]	Speed: 3960.347624 samples/sec	accuracy=0.416367	lr=0.381107
Epoch[21] Batch [99]	Speed: 4253.762123 samples/sec	accuracy=0.415732	lr=0.381014
Epoch[21] Batch [149]	Speed: 4857.817329 samples/sec	accuracy=0.415475	lr=0.380921
Epoch[21] Batch [199]	Speed: 4552.614019 samples/sec	accuracy=0.414668	lr=0.380828
Epoch[21] Batch [249]	Speed: 4397.933294 samples/sec	accuracy=0.414508	lr=0.380734
Epoch[21] Batch [299]	Speed: 3049.032274 samples/sec	accuracy=0.414059	lr=0.380641
Epoch[21] Batch [349]	Speed: 3238.782214 samples/sec	accuracy=0.413446	lr=0.380547
Epoch[21] Batch [399]	Speed: 4369.207560 samples/sec	accuracy=0.413320	lr=0.380453
Epoch[21] Batch [449]	Speed: 3999.629383 samples/sec	accuracy=0.413785	lr=0.380359
Epoch[21] Batch [499]	Speed: 4751.865324 samples/sec	accuracy=0.413742	lr=0.380264
Epoch[21] Batch [549]	Speed: 4191.671352 samples/sec	accuracy=0.413427	lr=0.380169
Epoch[21] Batch [599]	Speed: 4627.416940 samples/sec	accuracy=0.413662	lr=0.380074
Epoch[21] Batch [649]	Speed: 5294.673437 samples/sec	accuracy=0.413633	lr=0.379979
Epoch[21] Batch [699]	Speed: 3030.541583 samples/sec	accuracy=0.413824	lr=0.379884
Epoch[21] Batch [749]	Speed: 5097.497529 samples/sec	accuracy=0.413729	lr=0.379788
Epoch[21] Batch [799]	Speed: 4956.588718 samples/sec	accuracy=0.413641	lr=0.379693
Epoch[21] Batch [849]	Speed: 4586.505392 samples/sec	accuracy=0.413448	lr=0.379597
Epoch[21] Batch [899]	Speed: 5592.294361 samples/sec	accuracy=0.413447	lr=0.379501
Epoch[21] Batch [949]	Speed: 4484.297100 samples/sec	accuracy=0.413305	lr=0.379404
Epoch[21] Batch [999]	Speed: 5811.354784 samples/sec	accuracy=0.413340	lr=0.379307
Epoch[21] Batch [1049]	Speed: 4304.063676 samples/sec	accuracy=0.413150	lr=0.379211
Epoch[21] Batch [1099]	Speed: 5032.050240 samples/sec	accuracy=0.413146	lr=0.379114
Epoch[21] Batch [1149]	Speed: 4727.374992 samples/sec	accuracy=0.413174	lr=0.379016
Epoch[21] Batch [1199]	Speed: 4738.235166 samples/sec	accuracy=0.413062	lr=0.378919
Epoch[21] Batch [1249]	Speed: 4466.327509 samples/sec	accuracy=0.413212	lr=0.378821
[Epoch 21] training: accuracy=0.413201
[Epoch 21] speed: 4378 samples/sec	time cost: 302.746084
[Epoch 21] validation: err-top1=0.618383 err-top5=0.358199
Epoch[22] Batch [49]	Speed: 4787.309741 samples/sec	accuracy=0.420371	lr=0.378721
Epoch[22] Batch [99]	Speed: 5289.019724 samples/sec	accuracy=0.419590	lr=0.378623
Epoch[22] Batch [149]	Speed: 5763.924878 samples/sec	accuracy=0.418457	lr=0.378525
Epoch[22] Batch [199]	Speed: 4395.319922 samples/sec	accuracy=0.418296	lr=0.378426
Epoch[22] Batch [249]	Speed: 4828.960944 samples/sec	accuracy=0.418570	lr=0.378328
Epoch[22] Batch [299]	Speed: 5098.467162 samples/sec	accuracy=0.417796	lr=0.378229
Epoch[22] Batch [349]	Speed: 5847.277341 samples/sec	accuracy=0.417720	lr=0.378129
Epoch[22] Batch [399]	Speed: 5021.578567 samples/sec	accuracy=0.417415	lr=0.378030
Epoch[22] Batch [449]	Speed: 5324.795927 samples/sec	accuracy=0.417051	lr=0.377930
Epoch[22] Batch [499]	Speed: 4824.047504 samples/sec	accuracy=0.416957	lr=0.377831
Epoch[22] Batch [549]	Speed: 4575.942814 samples/sec	accuracy=0.416907	lr=0.377731
Epoch[22] Batch [599]	Speed: 6075.331195 samples/sec	accuracy=0.417005	lr=0.377630
Epoch[22] Batch [649]	Speed: 5340.053459 samples/sec	accuracy=0.416623	lr=0.377530
Epoch[22] Batch [699]	Speed: 5507.861382 samples/sec	accuracy=0.416112	lr=0.377429
Epoch[22] Batch [749]	Speed: 5082.250004 samples/sec	accuracy=0.415962	lr=0.377328
Epoch[22] Batch [799]	Speed: 5394.879410 samples/sec	accuracy=0.415769	lr=0.377227
Epoch[22] Batch [849]	Speed: 4763.438518 samples/sec	accuracy=0.415693	lr=0.377126
Epoch[22] Batch [899]	Speed: 5024.776015 samples/sec	accuracy=0.415414	lr=0.377024
Epoch[22] Batch [949]	Speed: 3291.670126 samples/sec	accuracy=0.415334	lr=0.376923
Epoch[22] Batch [999]	Speed: 5487.439476 samples/sec	accuracy=0.415405	lr=0.376821
Epoch[22] Batch [1049]	Speed: 5319.831044 samples/sec	accuracy=0.415253	lr=0.376719
Epoch[22] Batch [1099]	Speed: 4985.902148 samples/sec	accuracy=0.415164	lr=0.376616
Epoch[22] Batch [1149]	Speed: 3439.200027 samples/sec	accuracy=0.415150	lr=0.376514
Epoch[22] Batch [1199]	Speed: 3352.899345 samples/sec	accuracy=0.415050	lr=0.376411
Epoch[22] Batch [1249]	Speed: 5551.656895 samples/sec	accuracy=0.415070	lr=0.376308
[Epoch 22] training: accuracy=0.415049
[Epoch 22] speed: 4844 samples/sec	time cost: 274.362628
[Epoch 22] validation: err-top1=0.594507 err-top5=0.327288
Epoch[23] Batch [49]	Speed: 3469.683281 samples/sec	accuracy=0.421914	lr=0.376203
Epoch[23] Batch [99]	Speed: 3399.725636 samples/sec	accuracy=0.423438	lr=0.376099
Epoch[23] Batch [149]	Speed: 2649.339302 samples/sec	accuracy=0.422240	lr=0.375996
Epoch[23] Batch [199]	Speed: 3926.129854 samples/sec	accuracy=0.421318	lr=0.375892
Epoch[23] Batch [249]	Speed: 3540.672427 samples/sec	accuracy=0.420664	lr=0.375788
Epoch[23] Batch [299]	Speed: 5009.790565 samples/sec	accuracy=0.419460	lr=0.375683
Epoch[23] Batch [349]	Speed: 2773.419638 samples/sec	accuracy=0.419417	lr=0.375579
Epoch[23] Batch [399]	Speed: 2772.367812 samples/sec	accuracy=0.419329	lr=0.375474
Epoch[23] Batch [449]	Speed: 3507.926689 samples/sec	accuracy=0.419449	lr=0.375369
Epoch[23] Batch [499]	Speed: 4118.223599 samples/sec	accuracy=0.419172	lr=0.375264
Epoch[23] Batch [549]	Speed: 3167.642113 samples/sec	accuracy=0.418667	lr=0.375159
Epoch[23] Batch [599]	Speed: 3839.945641 samples/sec	accuracy=0.418618	lr=0.375053
Epoch[23] Batch [649]	Speed: 4233.163615 samples/sec	accuracy=0.418646	lr=0.374948
Epoch[23] Batch [699]	Speed: 4990.661983 samples/sec	accuracy=0.418331	lr=0.374842
Epoch[23] Batch [749]	Speed: 2856.175395 samples/sec	accuracy=0.418358	lr=0.374736
Epoch[23] Batch [799]	Speed: 2636.951921 samples/sec	accuracy=0.418351	lr=0.374629
Epoch[23] Batch [849]	Speed: 3975.364231 samples/sec	accuracy=0.418397	lr=0.374523
Epoch[23] Batch [899]	Speed: 3625.596294 samples/sec	accuracy=0.418402	lr=0.374416
Epoch[23] Batch [949]	Speed: 2163.639980 samples/sec	accuracy=0.418007	lr=0.374309
Epoch[23] Batch [999]	Speed: 3524.690523 samples/sec	accuracy=0.417857	lr=0.374202
Epoch[23] Batch [1049]	Speed: 3963.148049 samples/sec	accuracy=0.417735	lr=0.374095
Epoch[23] Batch [1099]	Speed: 3669.306066 samples/sec	accuracy=0.417654	lr=0.373987
Epoch[23] Batch [1149]	Speed: 2027.663018 samples/sec	accuracy=0.417515	lr=0.373879
Epoch[23] Batch [1199]	Speed: 2687.825584 samples/sec	accuracy=0.417469	lr=0.373771
Epoch[23] Batch [1249]	Speed: 3048.838907 samples/sec	accuracy=0.417394	lr=0.373663
[Epoch 23] training: accuracy=0.417409
[Epoch 23] speed: 3251 samples/sec	time cost: 407.505877
[Epoch 23] validation: err-top1=0.566549 err-top5=0.302816
Epoch[24] Batch [49]	Speed: 2541.645005 samples/sec	accuracy=0.423730	lr=0.373552
Epoch[24] Batch [99]	Speed: 2541.485132 samples/sec	accuracy=0.423174	lr=0.373444
Epoch[24] Batch [149]	Speed: 5366.088859 samples/sec	accuracy=0.422025	lr=0.373335
Epoch[24] Batch [199]	Speed: 3075.474227 samples/sec	accuracy=0.421406	lr=0.373226
Epoch[24] Batch [249]	Speed: 2995.804291 samples/sec	accuracy=0.420871	lr=0.373117
Epoch[24] Batch [299]	Speed: 5119.960572 samples/sec	accuracy=0.420703	lr=0.373007
Epoch[24] Batch [349]	Speed: 2275.426223 samples/sec	accuracy=0.420312	lr=0.372897
Epoch[24] Batch [399]	Speed: 3480.259160 samples/sec	accuracy=0.420217	lr=0.372788
Epoch[24] Batch [449]	Speed: 2826.481237 samples/sec	accuracy=0.420182	lr=0.372678
Epoch[24] Batch [499]	Speed: 3703.645034 samples/sec	accuracy=0.419943	lr=0.372567
Epoch[24] Batch [549]	Speed: 4013.647305 samples/sec	accuracy=0.419982	lr=0.372457
Epoch[24] Batch [599]	Speed: 4026.176756 samples/sec	accuracy=0.419852	lr=0.372346
Epoch[24] Batch [649]	Speed: 3196.896038 samples/sec	accuracy=0.419764	lr=0.372235
Epoch[24] Batch [699]	Speed: 3233.925713 samples/sec	accuracy=0.419632	lr=0.372124
Epoch[24] Batch [749]	Speed: 3919.321996 samples/sec	accuracy=0.419729	lr=0.372013
Epoch[24] Batch [799]	Speed: 3754.360945 samples/sec	accuracy=0.419543	lr=0.371901
Epoch[24] Batch [849]	Speed: 4556.155276 samples/sec	accuracy=0.419312	lr=0.371790
Epoch[24] Batch [899]	Speed: 5780.539463 samples/sec	accuracy=0.419191	lr=0.371678
Epoch[24] Batch [949]	Speed: 5667.881889 samples/sec	accuracy=0.419126	lr=0.371565
Epoch[24] Batch [999]	Speed: 4048.272226 samples/sec	accuracy=0.419103	lr=0.371453
Epoch[24] Batch [1049]	Speed: 3820.789717 samples/sec	accuracy=0.419066	lr=0.371341
Epoch[24] Batch [1099]	Speed: 4063.450190 samples/sec	accuracy=0.419190	lr=0.371228
Epoch[24] Batch [1149]	Speed: 3394.381154 samples/sec	accuracy=0.419131	lr=0.371115
Epoch[24] Batch [1199]	Speed: 2457.699013 samples/sec	accuracy=0.419177	lr=0.371002
Epoch[24] Batch [1249]	Speed: 2831.920149 samples/sec	accuracy=0.419141	lr=0.370888
[Epoch 24] training: accuracy=0.419143
[Epoch 24] speed: 3467 samples/sec	time cost: 395.912897
[Epoch 24] validation: err-top1=0.556023 err-top5=0.292909
Epoch[25] Batch [49]	Speed: 3978.401585 samples/sec	accuracy=0.424023	lr=0.370773
Epoch[25] Batch [99]	Speed: 2996.113210 samples/sec	accuracy=0.424980	lr=0.370659
Epoch[25] Batch [149]	Speed: 2547.331860 samples/sec	accuracy=0.424082	lr=0.370545
Epoch[25] Batch [199]	Speed: 4004.643759 samples/sec	accuracy=0.423740	lr=0.370431
Epoch[25] Batch [249]	Speed: 4414.043445 samples/sec	accuracy=0.424047	lr=0.370316
Epoch[25] Batch [299]	Speed: 3368.010933 samples/sec	accuracy=0.424150	lr=0.370202
Epoch[25] Batch [349]	Speed: 3582.140471 samples/sec	accuracy=0.423823	lr=0.370087
Epoch[25] Batch [399]	Speed: 3635.395892 samples/sec	accuracy=0.423582	lr=0.369972
Epoch[25] Batch [449]	Speed: 2981.793989 samples/sec	accuracy=0.423496	lr=0.369857
Epoch[25] Batch [499]	Speed: 3122.868859 samples/sec	accuracy=0.423467	lr=0.369741
Epoch[25] Batch [549]	Speed: 2636.272379 samples/sec	accuracy=0.423061	lr=0.369626
Epoch[25] Batch [599]	Speed: 801.528603 samples/sec	accuracy=0.422873	lr=0.369510
Epoch[25] Batch [649]	Speed: 3148.550288 samples/sec	accuracy=0.422775	lr=0.369394
Epoch[25] Batch [699]	Speed: 2985.510298 samples/sec	accuracy=0.422782	lr=0.369278
Epoch[25] Batch [749]	Speed: 4635.308920 samples/sec	accuracy=0.422671	lr=0.369161
Epoch[25] Batch [799]	Speed: 3932.610436 samples/sec	accuracy=0.422493	lr=0.369045
Epoch[25] Batch [849]	Speed: 3733.942833 samples/sec	accuracy=0.422448	lr=0.368928
Epoch[25] Batch [899]	Speed: 3791.321545 samples/sec	accuracy=0.422600	lr=0.368811
Epoch[25] Batch [949]	Speed: 2743.173060 samples/sec	accuracy=0.422390	lr=0.368694
Epoch[25] Batch [999]	Speed: 3339.929696 samples/sec	accuracy=0.422358	lr=0.368576
Epoch[25] Batch [1049]	Speed: 4146.848663 samples/sec	accuracy=0.422148	lr=0.368459
Epoch[25] Batch [1099]	Speed: 3834.484511 samples/sec	accuracy=0.422016	lr=0.368341
Epoch[25] Batch [1149]	Speed: 3270.321733 samples/sec	accuracy=0.421818	lr=0.368223
Epoch[25] Batch [1199]	Speed: 3830.005816 samples/sec	accuracy=0.421604	lr=0.368105
Epoch[25] Batch [1249]	Speed: 2730.267662 samples/sec	accuracy=0.421384	lr=0.367986
[Epoch 25] training: accuracy=0.421405
[Epoch 25] speed: 2993 samples/sec	time cost: 441.804043
[Epoch 25] validation: err-top1=0.696847 err-top5=0.458626
Epoch[26] Batch [49]	Speed: 3250.580637 samples/sec	accuracy=0.427656	lr=0.367865
Epoch[26] Batch [99]	Speed: 3640.293492 samples/sec	accuracy=0.426348	lr=0.367747
Epoch[26] Batch [149]	Speed: 3518.443564 samples/sec	accuracy=0.426641	lr=0.367628
Epoch[26] Batch [199]	Speed: 2873.535147 samples/sec	accuracy=0.426226	lr=0.367508
Epoch[26] Batch [249]	Speed: 3820.184253 samples/sec	accuracy=0.425680	lr=0.367389
Epoch[26] Batch [299]	Speed: 3579.827576 samples/sec	accuracy=0.425179	lr=0.367269
Epoch[26] Batch [349]	Speed: 4119.106572 samples/sec	accuracy=0.424570	lr=0.367150
Epoch[26] Batch [399]	Speed: 3752.479120 samples/sec	accuracy=0.424236	lr=0.367030
Epoch[26] Batch [449]	Speed: 3278.964028 samples/sec	accuracy=0.424227	lr=0.366909
Epoch[26] Batch [499]	Speed: 3237.743386 samples/sec	accuracy=0.424027	lr=0.366789
Epoch[26] Batch [549]	Speed: 3440.888692 samples/sec	accuracy=0.423835	lr=0.366668
Epoch[26] Batch [599]	Speed: 4239.594043 samples/sec	accuracy=0.423857	lr=0.366548
Epoch[26] Batch [649]	Speed: 5500.878592 samples/sec	accuracy=0.423576	lr=0.366426
Epoch[26] Batch [699]	Speed: 5573.887109 samples/sec	accuracy=0.423785	lr=0.366305
Epoch[26] Batch [749]	Speed: 3631.091865 samples/sec	accuracy=0.423426	lr=0.366184
Epoch[26] Batch [799]	Speed: 5289.599848 samples/sec	accuracy=0.423269	lr=0.366062
Epoch[26] Batch [849]	Speed: 2471.990947 samples/sec	accuracy=0.423054	lr=0.365940
Epoch[26] Batch [899]	Speed: 4873.046136 samples/sec	accuracy=0.422987	lr=0.365819
Epoch[26] Batch [949]	Speed: 3749.384399 samples/sec	accuracy=0.422870	lr=0.365696
Epoch[26] Batch [999]	Speed: 3943.528043 samples/sec	accuracy=0.422635	lr=0.365574
Epoch[26] Batch [1049]	Speed: 2734.715508 samples/sec	accuracy=0.422613	lr=0.365451
Epoch[26] Batch [1099]	Speed: 2427.908322 samples/sec	accuracy=0.422483	lr=0.365329
Epoch[26] Batch [1149]	Speed: 2585.293731 samples/sec	accuracy=0.422291	lr=0.365206
Epoch[26] Batch [1199]	Speed: 4486.231026 samples/sec	accuracy=0.422238	lr=0.365082
Epoch[26] Batch [1249]	Speed: 4715.907191 samples/sec	accuracy=0.422064	lr=0.364959
[Epoch 26] training: accuracy=0.422065
[Epoch 26] speed: 3588 samples/sec	time cost: 369.549965
[Epoch 26] validation: err-top1=0.587372 err-top5=0.324936
Epoch[27] Batch [49]	Speed: 3403.607868 samples/sec	accuracy=0.429824	lr=0.364833
Epoch[27] Batch [99]	Speed: 4678.343410 samples/sec	accuracy=0.430107	lr=0.364709
Epoch[27] Batch [149]	Speed: 3615.429833 samples/sec	accuracy=0.428262	lr=0.364585
Epoch[27] Batch [199]	Speed: 3707.662954 samples/sec	accuracy=0.426909	lr=0.364461
Epoch[27] Batch [249]	Speed: 4425.991334 samples/sec	accuracy=0.426508	lr=0.364337
Epoch[27] Batch [299]	Speed: 4444.907080 samples/sec	accuracy=0.426318	lr=0.364212
Epoch[27] Batch [349]	Speed: 4588.319287 samples/sec	accuracy=0.426230	lr=0.364087
Epoch[27] Batch [399]	Speed: 3647.063664 samples/sec	accuracy=0.426189	lr=0.363962
Epoch[27] Batch [449]	Speed: 5035.010053 samples/sec	accuracy=0.425974	lr=0.363837
Epoch[27] Batch [499]	Speed: 5110.468461 samples/sec	accuracy=0.425625	lr=0.363712
Epoch[27] Batch [549]	Speed: 5834.205341 samples/sec	accuracy=0.425634	lr=0.363586
Epoch[27] Batch [599]	Speed: 5694.422091 samples/sec	accuracy=0.425563	lr=0.363461
Epoch[27] Batch [649]	Speed: 3964.227513 samples/sec	accuracy=0.425356	lr=0.363335
Epoch[27] Batch [699]	Speed: 5567.503658 samples/sec	accuracy=0.425384	lr=0.363209
Epoch[27] Batch [749]	Speed: 4029.043420 samples/sec	accuracy=0.425164	lr=0.363082
Epoch[27] Batch [799]	Speed: 4485.691357 samples/sec	accuracy=0.424935	lr=0.362956
Epoch[27] Batch [849]	Speed: 3542.983094 samples/sec	accuracy=0.424730	lr=0.362829
Epoch[27] Batch [899]	Speed: 4517.251009 samples/sec	accuracy=0.424428	lr=0.362702
Epoch[27] Batch [949]	Speed: 4194.435731 samples/sec	accuracy=0.424244	lr=0.362575
Epoch[27] Batch [999]	Speed: 3920.199730 samples/sec	accuracy=0.424293	lr=0.362448
Epoch[27] Batch [1049]	Speed: 4915.913316 samples/sec	accuracy=0.424340	lr=0.362320
Epoch[27] Batch [1099]	Speed: 2691.680945 samples/sec	accuracy=0.424339	lr=0.362193
Epoch[27] Batch [1149]	Speed: 5248.087471 samples/sec	accuracy=0.424251	lr=0.362065
Epoch[27] Batch [1199]	Speed: 4438.666459 samples/sec	accuracy=0.424036	lr=0.361937
Epoch[27] Batch [1249]	Speed: 3426.262220 samples/sec	accuracy=0.424109	lr=0.361809
[Epoch 27] training: accuracy=0.424104
[Epoch 27] speed: 4217 samples/sec	time cost: 314.831343
[Epoch 27] validation: err-top1=0.570771 err-top5=0.304648
Epoch[28] Batch [49]	Speed: 3510.152686 samples/sec	accuracy=0.428906	lr=0.361678
Epoch[28] Batch [99]	Speed: 5661.076551 samples/sec	accuracy=0.428887	lr=0.361549
Epoch[28] Batch [149]	Speed: 3728.272423 samples/sec	accuracy=0.429056	lr=0.361420
Epoch[28] Batch [199]	Speed: 5172.296588 samples/sec	accuracy=0.428535	lr=0.361291
Epoch[28] Batch [249]	Speed: 3191.525503 samples/sec	accuracy=0.427895	lr=0.361162
Epoch[28] Batch [299]	Speed: 3233.556218 samples/sec	accuracy=0.428057	lr=0.361032
Epoch[28] Batch [349]	Speed: 2952.962690 samples/sec	accuracy=0.428030	lr=0.360903
Epoch[28] Batch [399]	Speed: 5235.445182 samples/sec	accuracy=0.427952	lr=0.360773
Epoch[28] Batch [449]	Speed: 5516.406774 samples/sec	accuracy=0.427190	lr=0.360643
Epoch[28] Batch [499]	Speed: 4879.566603 samples/sec	accuracy=0.427115	lr=0.360513
Epoch[28] Batch [549]	Speed: 5310.090631 samples/sec	accuracy=0.426895	lr=0.360382
Epoch[28] Batch [599]	Speed: 4951.167167 samples/sec	accuracy=0.426927	lr=0.360252
Epoch[28] Batch [649]	Speed: 4925.948152 samples/sec	accuracy=0.426779	lr=0.360121
Epoch[28] Batch [699]	Speed: 4951.765284 samples/sec	accuracy=0.426532	lr=0.359990
Epoch[28] Batch [749]	Speed: 4666.485224 samples/sec	accuracy=0.426193	lr=0.359859
Epoch[28] Batch [799]	Speed: 5758.119978 samples/sec	accuracy=0.425919	lr=0.359728
Epoch[28] Batch [849]	Speed: 5234.462050 samples/sec	accuracy=0.425939	lr=0.359596
Epoch[28] Batch [899]	Speed: 5574.367898 samples/sec	accuracy=0.425715	lr=0.359465
Epoch[28] Batch [949]	Speed: 4065.841206 samples/sec	accuracy=0.425783	lr=0.359333
Epoch[28] Batch [999]	Speed: 5035.112051 samples/sec	accuracy=0.425802	lr=0.359201
Epoch[28] Batch [1049]	Speed: 5509.176454 samples/sec	accuracy=0.425896	lr=0.359068
Epoch[28] Batch [1099]	Speed: 5377.693767 samples/sec	accuracy=0.425975	lr=0.358936
Epoch[28] Batch [1149]	Speed: 4859.471605 samples/sec	accuracy=0.425852	lr=0.358803
Epoch[28] Batch [1199]	Speed: 5954.872531 samples/sec	accuracy=0.425663	lr=0.358670
Epoch[28] Batch [1249]	Speed: 5337.560584 samples/sec	accuracy=0.425725	lr=0.358537
[Epoch 28] training: accuracy=0.425709
[Epoch 28] speed: 4636 samples/sec	time cost: 286.626077
[Epoch 28] validation: err-top1=0.615772 err-top5=0.355768
Epoch[29] Batch [49]	Speed: 4888.241092 samples/sec	accuracy=0.431797	lr=0.358401
Epoch[29] Batch [99]	Speed: 5568.495605 samples/sec	accuracy=0.431113	lr=0.358268
Epoch[29] Batch [149]	Speed: 4185.256081 samples/sec	accuracy=0.430195	lr=0.358134
Epoch[29] Batch [199]	Speed: 5368.163849 samples/sec	accuracy=0.429438	lr=0.358001
Epoch[29] Batch [249]	Speed: 5524.630064 samples/sec	accuracy=0.428539	lr=0.357867
Epoch[29] Batch [299]	Speed: 5680.067712 samples/sec	accuracy=0.428490	lr=0.357732
Epoch[29] Batch [349]	Speed: 5437.509107 samples/sec	accuracy=0.428770	lr=0.357598
Epoch[29] Batch [399]	Speed: 5038.190356 samples/sec	accuracy=0.429421	lr=0.357464
Epoch[29] Batch [449]	Speed: 5298.350470 samples/sec	accuracy=0.428780	lr=0.357329
Epoch[29] Batch [499]	Speed: 5689.603316 samples/sec	accuracy=0.428301	lr=0.357194
Epoch[29] Batch [549]	Speed: 5551.171119 samples/sec	accuracy=0.428251	lr=0.357059
Epoch[29] Batch [599]	Speed: 4397.122026 samples/sec	accuracy=0.428511	lr=0.356924
Epoch[29] Batch [649]	Speed: 4838.770594 samples/sec	accuracy=0.428389	lr=0.356788
Epoch[29] Batch [699]	Speed: 5384.876674 samples/sec	accuracy=0.428198	lr=0.356652
Epoch[29] Batch [749]	Speed: 4592.806872 samples/sec	accuracy=0.428152	lr=0.356517
Epoch[29] Batch [799]	Speed: 5014.547080 samples/sec	accuracy=0.428029	lr=0.356380
Epoch[29] Batch [849]	Speed: 5157.876904 samples/sec	accuracy=0.427973	lr=0.356244
Epoch[29] Batch [899]	Speed: 5239.141739 samples/sec	accuracy=0.427711	lr=0.356108
Epoch[29] Batch [949]	Speed: 3114.304094 samples/sec	accuracy=0.427441	lr=0.355971
Epoch[29] Batch [999]	Speed: 5103.716504 samples/sec	accuracy=0.427452	lr=0.355834
Epoch[29] Batch [1049]	Speed: 3460.303257 samples/sec	accuracy=0.427141	lr=0.355698
Epoch[29] Batch [1099]	Speed: 3002.828466 samples/sec	accuracy=0.427023	lr=0.355560
Epoch[29] Batch [1149]	Speed: 5865.957382 samples/sec	accuracy=0.426879	lr=0.355423
Epoch[29] Batch [1199]	Speed: 2588.053977 samples/sec	accuracy=0.426838	lr=0.355285
Epoch[29] Batch [1249]	Speed: 4779.744253 samples/sec	accuracy=0.426684	lr=0.355148
[Epoch 29] training: accuracy=0.426687
[Epoch 29] speed: 4608 samples/sec	time cost: 290.016370
[Epoch 29] validation: err-top1=0.616943 err-top5=0.358948
Epoch[30] Batch [49]	Speed: 3876.602914 samples/sec	accuracy=0.431719	lr=0.355007
Epoch[30] Batch [99]	Speed: 2334.767438 samples/sec	accuracy=0.431699	lr=0.354869
Epoch[30] Batch [149]	Speed: 2994.312127 samples/sec	accuracy=0.431387	lr=0.354731
Epoch[30] Batch [199]	Speed: 3017.579955 samples/sec	accuracy=0.431216	lr=0.354592
Epoch[30] Batch [249]	Speed: 4328.236197 samples/sec	accuracy=0.430660	lr=0.354454
Epoch[30] Batch [299]	Speed: 3632.863408 samples/sec	accuracy=0.431009	lr=0.354315
Epoch[30] Batch [349]	Speed: 5256.826891 samples/sec	accuracy=0.430884	lr=0.354176
Epoch[30] Batch [399]	Speed: 2430.258571 samples/sec	accuracy=0.430339	lr=0.354037
Epoch[30] Batch [449]	Speed: 5882.171196 samples/sec	accuracy=0.430464	lr=0.353897
Epoch[30] Batch [499]	Speed: 5708.290164 samples/sec	accuracy=0.430357	lr=0.353758
Epoch[30] Batch [549]	Speed: 5898.127556 samples/sec	accuracy=0.430520	lr=0.353618
Epoch[30] Batch [599]	Speed: 3646.609901 samples/sec	accuracy=0.430345	lr=0.353478
Epoch[30] Batch [649]	Speed: 3705.388547 samples/sec	accuracy=0.430152	lr=0.353338
Epoch[30] Batch [699]	Speed: 5817.443279 samples/sec	accuracy=0.429920	lr=0.353198
Epoch[30] Batch [749]	Speed: 4983.672108 samples/sec	accuracy=0.429411	lr=0.353057
Epoch[30] Batch [799]	Speed: 2209.911072 samples/sec	accuracy=0.429299	lr=0.352917
Epoch[30] Batch [849]	Speed: 5289.731055 samples/sec	accuracy=0.429214	lr=0.352776
Epoch[30] Batch [899]	Speed: 3435.497100 samples/sec	accuracy=0.429256	lr=0.352635
Epoch[30] Batch [949]	Speed: 2975.764439 samples/sec	accuracy=0.429191	lr=0.352493
Epoch[30] Batch [999]	Speed: 3068.337297 samples/sec	accuracy=0.429097	lr=0.352352
Epoch[30] Batch [1049]	Speed: 2426.573637 samples/sec	accuracy=0.429095	lr=0.352211
Epoch[30] Batch [1099]	Speed: 4102.212769 samples/sec	accuracy=0.429095	lr=0.352069
Epoch[30] Batch [1149]	Speed: 2325.230348 samples/sec	accuracy=0.428925	lr=0.351927
Epoch[30] Batch [1199]	Speed: 2805.109181 samples/sec	accuracy=0.428946	lr=0.351785
Epoch[30] Batch [1249]	Speed: 3339.747171 samples/sec	accuracy=0.428848	lr=0.351642
[Epoch 30] training: accuracy=0.428852
[Epoch 30] speed: 3455 samples/sec	time cost: 382.938928
[Epoch 30] validation: err-top1=0.692044 err-top5=0.450733
Epoch[31] Batch [49]	Speed: 5304.236785 samples/sec	accuracy=0.431914	lr=0.351497
Epoch[31] Batch [99]	Speed: 2495.764023 samples/sec	accuracy=0.431758	lr=0.351354
Epoch[31] Batch [149]	Speed: 3007.014340 samples/sec	accuracy=0.431413	lr=0.351212
Epoch[31] Batch [199]	Speed: 3931.262533 samples/sec	accuracy=0.431030	lr=0.351069
Epoch[31] Batch [249]	Speed: 2134.416982 samples/sec	accuracy=0.431152	lr=0.350925
Epoch[31] Batch [299]	Speed: 3133.938713 samples/sec	accuracy=0.431058	lr=0.350782
Epoch[31] Batch [349]	Speed: 5976.785316 samples/sec	accuracy=0.431381	lr=0.350639
Epoch[31] Batch [399]	Speed: 6024.607371 samples/sec	accuracy=0.431582	lr=0.350495
Epoch[31] Batch [449]	Speed: 3193.931333 samples/sec	accuracy=0.431599	lr=0.350351
Epoch[31] Batch [499]	Speed: 3216.901410 samples/sec	accuracy=0.431092	lr=0.350207
Epoch[31] Batch [549]	Speed: 2554.029308 samples/sec	accuracy=0.430815	lr=0.350062
Epoch[31] Batch [599]	Speed: 3342.117083 samples/sec	accuracy=0.430672	lr=0.349918
Epoch[31] Batch [649]	Speed: 3850.791718 samples/sec	accuracy=0.430640	lr=0.349773
Epoch[31] Batch [699]	Speed: 2150.708377 samples/sec	accuracy=0.430437	lr=0.349629
Epoch[31] Batch [749]	Speed: 5053.922523 samples/sec	accuracy=0.430370	lr=0.349484
Epoch[31] Batch [799]	Speed: 5953.337422 samples/sec	accuracy=0.430217	lr=0.349338
Epoch[31] Batch [849]	Speed: 5620.172204 samples/sec	accuracy=0.430296	lr=0.349193
Epoch[31] Batch [899]	Speed: 3499.766574 samples/sec	accuracy=0.430217	lr=0.349048
Epoch[31] Batch [949]	Speed: 1891.985799 samples/sec	accuracy=0.430209	lr=0.348902
Epoch[31] Batch [999]	Speed: 2415.356842 samples/sec	accuracy=0.430220	lr=0.348756
Epoch[31] Batch [1049]	Speed: 2931.912658 samples/sec	accuracy=0.430116	lr=0.348610
Epoch[31] Batch [1099]	Speed: 3445.830732 samples/sec	accuracy=0.430153	lr=0.348464
Epoch[31] Batch [1149]	Speed: 5085.656122 samples/sec	accuracy=0.430198	lr=0.348317
Epoch[31] Batch [1199]	Speed: 1879.956067 samples/sec	accuracy=0.430146	lr=0.348171
Epoch[31] Batch [1249]	Speed: 3299.070690 samples/sec	accuracy=0.429958	lr=0.348024
[Epoch 31] training: accuracy=0.429963
[Epoch 31] speed: 3214 samples/sec	time cost: 412.499612
[Epoch 31] validation: err-top1=0.596839 err-top5=0.334343
Epoch[32] Batch [49]	Speed: 3198.978321 samples/sec	accuracy=0.437129	lr=0.347874
Epoch[32] Batch [99]	Speed: 2642.859063 samples/sec	accuracy=0.436367	lr=0.347727
Epoch[32] Batch [149]	Speed: 3616.059501 samples/sec	accuracy=0.435352	lr=0.347580
Epoch[32] Batch [199]	Speed: 2272.599163 samples/sec	accuracy=0.434976	lr=0.347432
Epoch[32] Batch [249]	Speed: 3642.401307 samples/sec	accuracy=0.434934	lr=0.347285
Epoch[32] Batch [299]	Speed: 5097.402788 samples/sec	accuracy=0.434824	lr=0.347137
Epoch[32] Batch [349]	Speed: 5768.575372 samples/sec	accuracy=0.434520	lr=0.346989
Epoch[32] Batch [399]	Speed: 5274.325721 samples/sec	accuracy=0.434163	lr=0.346841
Epoch[32] Batch [449]	Speed: 3931.048154 samples/sec	accuracy=0.433859	lr=0.346692
Epoch[32] Batch [499]	Speed: 4424.940908 samples/sec	accuracy=0.433777	lr=0.346544
Epoch[32] Batch [549]	Speed: 1714.087069 samples/sec	accuracy=0.433887	lr=0.346395
Epoch[32] Batch [599]	Speed: 4226.671054 samples/sec	accuracy=0.433870	lr=0.346246
Epoch[32] Batch [649]	Speed: 3401.755766 samples/sec	accuracy=0.433552	lr=0.346097
Epoch[32] Batch [699]	Speed: 4665.639070 samples/sec	accuracy=0.433284	lr=0.345948
Epoch[32] Batch [749]	Speed: 3197.124207 samples/sec	accuracy=0.433108	lr=0.345799
Epoch[32] Batch [799]	Speed: 3450.759596 samples/sec	accuracy=0.432727	lr=0.345649
Epoch[32] Batch [849]	Speed: 4288.355681 samples/sec	accuracy=0.432492	lr=0.345499
Epoch[32] Batch [899]	Speed: 2934.855240 samples/sec	accuracy=0.432326	lr=0.345349
Epoch[32] Batch [949]	Speed: 3954.450657 samples/sec	accuracy=0.432163	lr=0.345199
Epoch[32] Batch [999]	Speed: 2399.473875 samples/sec	accuracy=0.431969	lr=0.345049
Epoch[32] Batch [1049]	Speed: 3317.478751 samples/sec	accuracy=0.431850	lr=0.344899
Epoch[32] Batch [1099]	Speed: 2561.478209 samples/sec	accuracy=0.431657	lr=0.344748
Epoch[32] Batch [1149]	Speed: 4134.934259 samples/sec	accuracy=0.431630	lr=0.344597
Epoch[32] Batch [1199]	Speed: 3175.486025 samples/sec	accuracy=0.431509	lr=0.344446
Epoch[32] Batch [1249]	Speed: 2917.977505 samples/sec	accuracy=0.431283	lr=0.344295
[Epoch 32] training: accuracy=0.431291
[Epoch 32] speed: 3336 samples/sec	time cost: 399.960256
[Epoch 32] validation: err-top1=0.582231 err-top5=0.321648
Epoch[33] Batch [49]	Speed: 3090.671372 samples/sec	accuracy=0.438848	lr=0.344141
Epoch[33] Batch [99]	Speed: 3423.465644 samples/sec	accuracy=0.437197	lr=0.343989
Epoch[33] Batch [149]	Speed: 5181.641874 samples/sec	accuracy=0.435840	lr=0.343838
Epoch[33] Batch [199]	Speed: 3118.616940 samples/sec	accuracy=0.434736	lr=0.343686
Epoch[33] Batch [249]	Speed: 3063.477999 samples/sec	accuracy=0.434648	lr=0.343534
Epoch[33] Batch [299]	Speed: 3065.471618 samples/sec	accuracy=0.434958	lr=0.343382
Epoch[33] Batch [349]	Speed: 3628.509028 samples/sec	accuracy=0.434741	lr=0.343229
Epoch[33] Batch [399]	Speed: 3277.818266 samples/sec	accuracy=0.434277	lr=0.343077
Epoch[33] Batch [449]	Speed: 4186.817774 samples/sec	accuracy=0.433895	lr=0.342924
Epoch[33] Batch [499]	Speed: 3505.709008 samples/sec	accuracy=0.433840	lr=0.342771
Epoch[33] Batch [549]	Speed: 3307.596287 samples/sec	accuracy=0.433775	lr=0.342618
Epoch[33] Batch [599]	Speed: 4367.995107 samples/sec	accuracy=0.433794	lr=0.342465
Epoch[33] Batch [649]	Speed: 2944.946629 samples/sec	accuracy=0.433612	lr=0.342312
Epoch[33] Batch [699]	Speed: 3323.083018 samples/sec	accuracy=0.433675	lr=0.342158
Epoch[33] Batch [749]	Speed: 3088.844456 samples/sec	accuracy=0.433464	lr=0.342005
Epoch[33] Batch [799]	Speed: 4270.559256 samples/sec	accuracy=0.433485	lr=0.341851
Epoch[33] Batch [849]	Speed: 2829.193233 samples/sec	accuracy=0.433339	lr=0.341697
Epoch[33] Batch [899]	Speed: 3032.574601 samples/sec	accuracy=0.433241	lr=0.341543
Epoch[33] Batch [949]	Speed: 4162.171548 samples/sec	accuracy=0.432954	lr=0.341388
Epoch[33] Batch [999]	Speed: 4241.059863 samples/sec	accuracy=0.432937	lr=0.341234
Epoch[33] Batch [1049]	Speed: 3293.563977 samples/sec	accuracy=0.432887	lr=0.341079
Epoch[33] Batch [1099]	Speed: 3907.952448 samples/sec	accuracy=0.432771	lr=0.340924
Epoch[33] Batch [1149]	Speed: 2724.822327 samples/sec	accuracy=0.432764	lr=0.340769
Epoch[33] Batch [1199]	Speed: 4339.187961 samples/sec	accuracy=0.432847	lr=0.340614
Epoch[33] Batch [1249]	Speed: 5001.534238 samples/sec	accuracy=0.432865	lr=0.340458
[Epoch 33] training: accuracy=0.432872
[Epoch 33] speed: 3501 samples/sec	time cost: 380.879008
[Epoch 33] validation: err-top1=0.636998 err-top5=0.383470
Epoch[34] Batch [49]	Speed: 3297.609214 samples/sec	accuracy=0.443320	lr=0.340300
Epoch[34] Batch [99]	Speed: 4308.319900 samples/sec	accuracy=0.439707	lr=0.340144
Epoch[34] Batch [149]	Speed: 3279.960696 samples/sec	accuracy=0.437839	lr=0.339988
Epoch[34] Batch [199]	Speed: 2561.236559 samples/sec	accuracy=0.437515	lr=0.339832
Epoch[34] Batch [249]	Speed: 3657.398351 samples/sec	accuracy=0.437234	lr=0.339676
Epoch[34] Batch [299]	Speed: 3567.511644 samples/sec	accuracy=0.437298	lr=0.339520
Epoch[34] Batch [349]	Speed: 3482.069134 samples/sec	accuracy=0.437360	lr=0.339363
Epoch[34] Batch [399]	Speed: 2542.881085 samples/sec	accuracy=0.437188	lr=0.339206
Epoch[34] Batch [449]	Speed: 3823.363921 samples/sec	accuracy=0.436758	lr=0.339049
Epoch[34] Batch [499]	Speed: 4060.406466 samples/sec	accuracy=0.436398	lr=0.338892
Epoch[34] Batch [549]	Speed: 3477.524468 samples/sec	accuracy=0.436238	lr=0.338735
Epoch[34] Batch [599]	Speed: 3365.532224 samples/sec	accuracy=0.435811	lr=0.338578
Epoch[34] Batch [649]	Speed: 4968.921383 samples/sec	accuracy=0.435648	lr=0.338420
Epoch[34] Batch [699]	Speed: 3113.216158 samples/sec	accuracy=0.435268	lr=0.338263
Epoch[34] Batch [749]	Speed: 4230.190141 samples/sec	accuracy=0.435023	lr=0.338105
Epoch[34] Batch [799]	Speed: 3697.479971 samples/sec	accuracy=0.435125	lr=0.337947
Epoch[34] Batch [849]	Speed: 3748.909139 samples/sec	accuracy=0.435026	lr=0.337788
Epoch[34] Batch [899]	Speed: 4210.934642 samples/sec	accuracy=0.434855	lr=0.337630
Epoch[34] Batch [949]	Speed: 2825.295076 samples/sec	accuracy=0.434727	lr=0.337472
Epoch[34] Batch [999]	Speed: 1579.016896 samples/sec	accuracy=0.434823	lr=0.337313
Epoch[34] Batch [1049]	Speed: 3889.309020 samples/sec	accuracy=0.434813	lr=0.337154
Epoch[34] Batch [1099]	Speed: 4173.465852 samples/sec	accuracy=0.434664	lr=0.336995
Epoch[34] Batch [1149]	Speed: 2896.272111 samples/sec	accuracy=0.434688	lr=0.336836
Epoch[34] Batch [1199]	Speed: 4082.180611 samples/sec	accuracy=0.434551	lr=0.336677
Epoch[34] Batch [1249]	Speed: 3514.397131 samples/sec	accuracy=0.434285	lr=0.336517
[Epoch 34] training: accuracy=0.434283
[Epoch 34] speed: 3351 samples/sec	time cost: 393.136187
[Epoch 34] validation: err-top1=0.621033 err-top5=0.369425
Epoch[35] Batch [49]	Speed: 4212.066664 samples/sec	accuracy=0.438867	lr=0.336354
Epoch[35] Batch [99]	Speed: 4251.117192 samples/sec	accuracy=0.440039	lr=0.336194
Epoch[35] Batch [149]	Speed: 3774.885338 samples/sec	accuracy=0.438249	lr=0.336034
Epoch[35] Batch [199]	Speed: 5281.700758 samples/sec	accuracy=0.437651	lr=0.335874
Epoch[35] Batch [249]	Speed: 3501.283245 samples/sec	accuracy=0.437664	lr=0.335714
Epoch[35] Batch [299]	Speed: 4503.049333 samples/sec	accuracy=0.437367	lr=0.335553
Epoch[35] Batch [349]	Speed: 5911.058652 samples/sec	accuracy=0.437606	lr=0.335393
Epoch[35] Batch [399]	Speed: 5296.380033 samples/sec	accuracy=0.437114	lr=0.335232
Epoch[35] Batch [449]	Speed: 5232.180599 samples/sec	accuracy=0.436977	lr=0.335071
Epoch[35] Batch [499]	Speed: 3651.024495 samples/sec	accuracy=0.436943	lr=0.334910
Epoch[35] Batch [549]	Speed: 3955.655226 samples/sec	accuracy=0.436765	lr=0.334749
Epoch[35] Batch [599]	Speed: 3909.921866 samples/sec	accuracy=0.436499	lr=0.334587
Epoch[35] Batch [649]	Speed: 1634.348543 samples/sec	accuracy=0.436393	lr=0.334425
Epoch[35] Batch [699]	Speed: 4334.514802 samples/sec	accuracy=0.436027	lr=0.334264
Epoch[35] Batch [749]	Speed: 3944.420349 samples/sec	accuracy=0.435840	lr=0.334102
Epoch[35] Batch [799]	Speed: 4143.015251 samples/sec	accuracy=0.435630	lr=0.333940
Epoch[35] Batch [849]	Speed: 4104.699375 samples/sec	accuracy=0.435770	lr=0.333777
Epoch[35] Batch [899]	Speed: 4510.316981 samples/sec	accuracy=0.435663	lr=0.333615
Epoch[35] Batch [949]	Speed: 4188.888144 samples/sec	accuracy=0.435752	lr=0.333452
Epoch[35] Batch [999]	Speed: 3560.577234 samples/sec	accuracy=0.435603	lr=0.333290
Epoch[35] Batch [1049]	Speed: 3598.043413 samples/sec	accuracy=0.435769	lr=0.333127
Epoch[35] Batch [1099]	Speed: 4549.926524 samples/sec	accuracy=0.435510	lr=0.332964
Epoch[35] Batch [1149]	Speed: 3937.326684 samples/sec	accuracy=0.435357	lr=0.332801
Epoch[35] Batch [1199]	Speed: 4170.150683 samples/sec	accuracy=0.435326	lr=0.332637
Epoch[35] Batch [1249]	Speed: 2999.071249 samples/sec	accuracy=0.435283	lr=0.332474
[Epoch 35] training: accuracy=0.435293
[Epoch 35] speed: 3894 samples/sec	time cost: 344.632430
[Epoch 35] validation: err-top1=0.602639 err-top5=0.340641
Epoch[36] Batch [49]	Speed: 4155.244539 samples/sec	accuracy=0.439785	lr=0.332307
Epoch[36] Batch [99]	Speed: 4877.051598 samples/sec	accuracy=0.440850	lr=0.332143
Epoch[36] Batch [149]	Speed: 3883.118751 samples/sec	accuracy=0.439948	lr=0.331979
Epoch[36] Batch [199]	Speed: 3752.787784 samples/sec	accuracy=0.439888	lr=0.331815
Epoch[36] Batch [249]	Speed: 3463.558839 samples/sec	accuracy=0.440230	lr=0.331650
Epoch[36] Batch [299]	Speed: 4123.631639 samples/sec	accuracy=0.439411	lr=0.331486
Epoch[36] Batch [349]	Speed: 3512.202799 samples/sec	accuracy=0.439408	lr=0.331321
Epoch[36] Batch [399]	Speed: 5045.781619 samples/sec	accuracy=0.439543	lr=0.331157
Epoch[36] Batch [449]	Speed: 3898.500975 samples/sec	accuracy=0.439086	lr=0.330992
Epoch[36] Batch [499]	Speed: 4219.073754 samples/sec	accuracy=0.439090	lr=0.330827
Epoch[36] Batch [549]	Speed: 4614.430206 samples/sec	accuracy=0.438640	lr=0.330661
Epoch[36] Batch [599]	Speed: 3018.363028 samples/sec	accuracy=0.438175	lr=0.330496
Epoch[36] Batch [649]	Speed: 5079.830704 samples/sec	accuracy=0.437979	lr=0.330330
Epoch[36] Batch [699]	Speed: 4653.944260 samples/sec	accuracy=0.438005	lr=0.330165
Epoch[36] Batch [749]	Speed: 4823.151483 samples/sec	accuracy=0.437823	lr=0.329999
Epoch[36] Batch [799]	Speed: 3820.511158 samples/sec	accuracy=0.437560	lr=0.329833
Epoch[36] Batch [849]	Speed: 2011.279668 samples/sec	accuracy=0.437422	lr=0.329667
Epoch[36] Batch [899]	Speed: 2738.200450 samples/sec	accuracy=0.437250	lr=0.329500
Epoch[36] Batch [949]	Speed: 4514.668302 samples/sec	accuracy=0.437299	lr=0.329334
Epoch[36] Batch [999]	Speed: 2914.520036 samples/sec	accuracy=0.437263	lr=0.329167
Epoch[36] Batch [1049]	Speed: 3160.307542 samples/sec	accuracy=0.437197	lr=0.329000
Epoch[36] Batch [1099]	Speed: 4207.993870 samples/sec	accuracy=0.437039	lr=0.328833
Epoch[36] Batch [1149]	Speed: 3833.268641 samples/sec	accuracy=0.436915	lr=0.328666
Epoch[36] Batch [1199]	Speed: 4449.148166 samples/sec	accuracy=0.436708	lr=0.328499
Epoch[36] Batch [1249]	Speed: 5283.435137 samples/sec	accuracy=0.436705	lr=0.328332
[Epoch 36] training: accuracy=0.436712
[Epoch 36] speed: 3806 samples/sec	time cost: 348.823039
[Epoch 36] validation: err-top1=0.621991 err-top5=0.362584
Epoch[37] Batch [49]	Speed: 4460.224038 samples/sec	accuracy=0.443555	lr=0.328161
Epoch[37] Batch [99]	Speed: 5093.587280 samples/sec	accuracy=0.443623	lr=0.327993
Epoch[37] Batch [149]	Speed: 5157.973535 samples/sec	accuracy=0.442533	lr=0.327825
Epoch[37] Batch [199]	Speed: 4282.023424 samples/sec	accuracy=0.442559	lr=0.327657
Epoch[37] Batch [249]	Speed: 5064.378464 samples/sec	accuracy=0.441629	lr=0.327489
Epoch[37] Batch [299]	Speed: 3732.135574 samples/sec	accuracy=0.441781	lr=0.327321
Epoch[37] Batch [349]	Speed: 4270.740580 samples/sec	accuracy=0.441588	lr=0.327152
Epoch[37] Batch [399]	Speed: 2790.058181 samples/sec	accuracy=0.441523	lr=0.326983
Epoch[37] Batch [449]	Speed: 4685.517554 samples/sec	accuracy=0.441254	lr=0.326815
Epoch[37] Batch [499]	Speed: 5411.877093 samples/sec	accuracy=0.440713	lr=0.326646
Epoch[37] Batch [549]	Speed: 3742.321013 samples/sec	accuracy=0.440716	lr=0.326477
Epoch[37] Batch [599]	Speed: 3500.127877 samples/sec	accuracy=0.440251	lr=0.326307
Epoch[37] Batch [649]	Speed: 4241.862403 samples/sec	accuracy=0.440044	lr=0.326138
Epoch[37] Batch [699]	Speed: 5263.988517 samples/sec	accuracy=0.439901	lr=0.325968
Epoch[37] Batch [749]	Speed: 4166.623299 samples/sec	accuracy=0.439879	lr=0.325799
Epoch[37] Batch [799]	Speed: 5660.925231 samples/sec	accuracy=0.439596	lr=0.325629
Epoch[37] Batch [849]	Speed: 5199.250092 samples/sec	accuracy=0.439264	lr=0.325459
Epoch[37] Batch [899]	Speed: 4773.526397 samples/sec	accuracy=0.439240	lr=0.325289
Epoch[37] Batch [949]	Speed: 4957.442423 samples/sec	accuracy=0.439037	lr=0.325118
Epoch[37] Batch [999]	Speed: 4794.555683 samples/sec	accuracy=0.438914	lr=0.324948
Epoch[37] Batch [1049]	Speed: 5579.396331 samples/sec	accuracy=0.438737	lr=0.324777
Epoch[37] Batch [1099]	Speed: 5285.457613 samples/sec	accuracy=0.438697	lr=0.324607
Epoch[37] Batch [1149]	Speed: 5929.175393 samples/sec	accuracy=0.438651	lr=0.324436
Epoch[37] Batch [1199]	Speed: 3593.873215 samples/sec	accuracy=0.438593	lr=0.324265
Epoch[37] Batch [1249]	Speed: 4940.749092 samples/sec	accuracy=0.438762	lr=0.324094
[Epoch 37] training: accuracy=0.438765
[Epoch 37] speed: 4516 samples/sec	time cost: 297.062345
[Epoch 37] validation: err-top1=0.611029 err-top5=0.348633
Epoch[38] Batch [49]	Speed: 5333.319479 samples/sec	accuracy=0.446836	lr=0.323919
Epoch[38] Batch [99]	Speed: 5659.922758 samples/sec	accuracy=0.443252	lr=0.323747
Epoch[38] Batch [149]	Speed: 4136.653353 samples/sec	accuracy=0.443548	lr=0.323576
Epoch[38] Batch [199]	Speed: 5579.670462 samples/sec	accuracy=0.443237	lr=0.323404
Epoch[38] Batch [249]	Speed: 5591.436442 samples/sec	accuracy=0.442945	lr=0.323232
Epoch[38] Batch [299]	Speed: 4009.586051 samples/sec	accuracy=0.442337	lr=0.323060
Epoch[38] Batch [349]	Speed: 5177.689776 samples/sec	accuracy=0.441635	lr=0.322888
Epoch[38] Batch [399]	Speed: 3527.715493 samples/sec	accuracy=0.441431	lr=0.322715
Epoch[38] Batch [449]	Speed: 5527.446893 samples/sec	accuracy=0.441482	lr=0.322543
Epoch[38] Batch [499]	Speed: 5946.379689 samples/sec	accuracy=0.441846	lr=0.322370
Epoch[38] Batch [549]	Speed: 5273.907987 samples/sec	accuracy=0.441728	lr=0.322197
Epoch[38] Batch [599]	Speed: 5502.443534 samples/sec	accuracy=0.441484	lr=0.322025
Epoch[38] Batch [649]	Speed: 5363.204193 samples/sec	accuracy=0.441253	lr=0.321851
Epoch[38] Batch [699]	Speed: 4901.374685 samples/sec	accuracy=0.441582	lr=0.321678
Epoch[38] Batch [749]	Speed: 5331.805958 samples/sec	accuracy=0.441512	lr=0.321505
Epoch[38] Batch [799]	Speed: 5184.657411 samples/sec	accuracy=0.441440	lr=0.321331
Epoch[38] Batch [849]	Speed: 5187.592874 samples/sec	accuracy=0.441321	lr=0.321158
Epoch[38] Batch [899]	Speed: 5682.963354 samples/sec	accuracy=0.441107	lr=0.320984
Epoch[38] Batch [949]	Speed: 5417.184525 samples/sec	accuracy=0.440847	lr=0.320810
Epoch[38] Batch [999]	Speed: 5406.300343 samples/sec	accuracy=0.440676	lr=0.320636
Epoch[38] Batch [1049]	Speed: 5861.594960 samples/sec	accuracy=0.440569	lr=0.320462
Epoch[38] Batch [1099]	Speed: 4912.932222 samples/sec	accuracy=0.440544	lr=0.320287
Epoch[38] Batch [1149]	Speed: 5694.087048 samples/sec	accuracy=0.440510	lr=0.320113
Epoch[38] Batch [1199]	Speed: 5781.642560 samples/sec	accuracy=0.440409	lr=0.319938
Epoch[38] Batch [1249]	Speed: 5798.853112 samples/sec	accuracy=0.440262	lr=0.319763
[Epoch 38] training: accuracy=0.440270
[Epoch 38] speed: 5186 samples/sec	time cost: 257.162690
[Epoch 38] validation: err-top1=0.596460 err-top5=0.335658
Epoch[39] Batch [49]	Speed: 5510.134576 samples/sec	accuracy=0.445684	lr=0.319585
Epoch[39] Batch [99]	Speed: 4997.359810 samples/sec	accuracy=0.444639	lr=0.319410
Epoch[39] Batch [149]	Speed: 5819.001022 samples/sec	accuracy=0.445228	lr=0.319234
Epoch[39] Batch [199]	Speed: 5108.032798 samples/sec	accuracy=0.444053	lr=0.319059
Epoch[39] Batch [249]	Speed: 5136.676210 samples/sec	accuracy=0.444801	lr=0.318883
Epoch[39] Batch [299]	Speed: 5346.495101 samples/sec	accuracy=0.444069	lr=0.318708
Epoch[39] Batch [349]	Speed: 5605.818129 samples/sec	accuracy=0.443750	lr=0.318532
Epoch[39] Batch [399]	Speed: 4873.655168 samples/sec	accuracy=0.443711	lr=0.318356
Epoch[39] Batch [449]	Speed: 5470.157986 samples/sec	accuracy=0.443262	lr=0.318180
Epoch[39] Batch [499]	Speed: 5311.782876 samples/sec	accuracy=0.443195	lr=0.318004
Epoch[39] Batch [549]	Speed: 5268.993420 samples/sec	accuracy=0.443043	lr=0.317827
Epoch[39] Batch [599]	Speed: 3541.583517 samples/sec	accuracy=0.442788	lr=0.317651
Epoch[39] Batch [649]	Speed: 1479.408467 samples/sec	accuracy=0.442556	lr=0.317474
Epoch[39] Batch [699]	Speed: 4096.862994 samples/sec	accuracy=0.442259	lr=0.317297
Epoch[39] Batch [749]	Speed: 4099.218699 samples/sec	accuracy=0.442223	lr=0.317120
Epoch[39] Batch [799]	Speed: 2828.690993 samples/sec	accuracy=0.442045	lr=0.316943
Epoch[39] Batch [849]	Speed: 3516.835910 samples/sec	accuracy=0.441865	lr=0.316766
Epoch[39] Batch [899]	Speed: 2624.274935 samples/sec	accuracy=0.441871	lr=0.316589
Epoch[39] Batch [949]	Speed: 2708.885209 samples/sec	accuracy=0.441612	lr=0.316411
Epoch[39] Batch [999]	Speed: 3609.520689 samples/sec	accuracy=0.441420	lr=0.316233
Epoch[39] Batch [1049]	Speed: 3455.490780 samples/sec	accuracy=0.441278	lr=0.316056
Epoch[39] Batch [1099]	Speed: 4021.512640 samples/sec	accuracy=0.441191	lr=0.315878
Epoch[39] Batch [1149]	Speed: 5218.049296 samples/sec	accuracy=0.441146	lr=0.315700
Epoch[39] Batch [1199]	Speed: 4497.147348 samples/sec	accuracy=0.441093	lr=0.315522
Epoch[39] Batch [1249]	Speed: 5575.356792 samples/sec	accuracy=0.441104	lr=0.315343
[Epoch 39] training: accuracy=0.441092
[Epoch 39] speed: 3952 samples/sec	time cost: 338.973559
[Epoch 39] validation: err-top1=0.557577 err-top5=0.298131
Epoch[40] Batch [49]	Speed: 5247.953449 samples/sec	accuracy=0.446309	lr=0.315161
Epoch[40] Batch [99]	Speed: 5988.238336 samples/sec	accuracy=0.447510	lr=0.314983
Epoch[40] Batch [149]	Speed: 5616.035240 samples/sec	accuracy=0.447493	lr=0.314804
Epoch[40] Batch [199]	Speed: 5461.384028 samples/sec	accuracy=0.446968	lr=0.314625
Epoch[40] Batch [249]	Speed: 5068.026407 samples/sec	accuracy=0.446695	lr=0.314446
Epoch[40] Batch [299]	Speed: 3749.140895 samples/sec	accuracy=0.446273	lr=0.314267
Epoch[40] Batch [349]	Speed: 2250.822723 samples/sec	accuracy=0.445756	lr=0.314087
Epoch[40] Batch [399]	Speed: 4556.870996 samples/sec	accuracy=0.445195	lr=0.313908
Epoch[40] Batch [449]	Speed: 4836.562465 samples/sec	accuracy=0.444959	lr=0.313728
Epoch[40] Batch [499]	Speed: 4032.740521 samples/sec	accuracy=0.445268	lr=0.313549
Epoch[40] Batch [549]	Speed: 2399.268901 samples/sec	accuracy=0.444915	lr=0.313369
Epoch[40] Batch [599]	Speed: 2771.464674 samples/sec	accuracy=0.444331	lr=0.313189
Epoch[40] Batch [649]	Speed: 1180.029085 samples/sec	accuracy=0.443998	lr=0.313009
Epoch[40] Batch [699]	Speed: 1633.463022 samples/sec	accuracy=0.443862	lr=0.312829
Epoch[40] Batch [749]	Speed: 4694.623916 samples/sec	accuracy=0.443794	lr=0.312648
Epoch[40] Batch [799]	Speed: 4413.682376 samples/sec	accuracy=0.443884	lr=0.312468
Epoch[40] Batch [849]	Speed: 2942.034651 samples/sec	accuracy=0.443857	lr=0.312287
Epoch[40] Batch [899]	Speed: 1859.228636 samples/sec	accuracy=0.443641	lr=0.312106
Epoch[40] Batch [949]	Speed: 3955.885340 samples/sec	accuracy=0.443570	lr=0.311925
Epoch[40] Batch [999]	Speed: 3913.105203 samples/sec	accuracy=0.443644	lr=0.311744
Epoch[40] Batch [1049]	Speed: 2144.199528 samples/sec	accuracy=0.443543	lr=0.311563
Epoch[40] Batch [1099]	Speed: 4364.444096 samples/sec	accuracy=0.443564	lr=0.311382
Epoch[40] Batch [1149]	Speed: 4244.472672 samples/sec	accuracy=0.443468	lr=0.311201
Epoch[40] Batch [1199]	Speed: 3477.235886 samples/sec	accuracy=0.443262	lr=0.311019
Epoch[40] Batch [1249]	Speed: 2897.326140 samples/sec	accuracy=0.443043	lr=0.310837
[Epoch 40] training: accuracy=0.443042
[Epoch 40] speed: 3147 samples/sec	time cost: 425.978539
[Epoch 40] validation: err-top1=0.552327 err-top5=0.291565
Epoch[41] Batch [49]	Speed: 3564.760062 samples/sec	accuracy=0.450156	lr=0.310652
Epoch[41] Batch [99]	Speed: 5096.646680 samples/sec	accuracy=0.449121	lr=0.310470
Epoch[41] Batch [149]	Speed: 2229.753433 samples/sec	accuracy=0.448874	lr=0.310288
Epoch[41] Batch [199]	Speed: 4967.610121 samples/sec	accuracy=0.448169	lr=0.310105
Epoch[41] Batch [249]	Speed: 2959.370446 samples/sec	accuracy=0.447656	lr=0.309923
Epoch[41] Batch [299]	Speed: 3957.965165 samples/sec	accuracy=0.447461	lr=0.309741
Epoch[41] Batch [349]	Speed: 2924.364383 samples/sec	accuracy=0.447137	lr=0.309558
Epoch[41] Batch [399]	Speed: 2435.813443 samples/sec	accuracy=0.446558	lr=0.309375
Epoch[41] Batch [449]	Speed: 3246.624563 samples/sec	accuracy=0.446426	lr=0.309192
Epoch[41] Batch [499]	Speed: 2315.112194 samples/sec	accuracy=0.446082	lr=0.309009
Epoch[41] Batch [549]	Speed: 3533.329444 samples/sec	accuracy=0.445996	lr=0.308826
Epoch[41] Batch [599]	Speed: 5819.938402 samples/sec	accuracy=0.445804	lr=0.308643
Epoch[41] Batch [649]	Speed: 5173.826469 samples/sec	accuracy=0.445616	lr=0.308459
Epoch[41] Batch [699]	Speed: 5386.268490 samples/sec	accuracy=0.445120	lr=0.308276
Epoch[41] Batch [749]	Speed: 2076.908310 samples/sec	accuracy=0.445020	lr=0.308092
Epoch[41] Batch [799]	Speed: 1985.119898 samples/sec	accuracy=0.444739	lr=0.307908
Epoch[41] Batch [849]	Speed: 5768.134092 samples/sec	accuracy=0.444866	lr=0.307725
Epoch[41] Batch [899]	Speed: 5622.722055 samples/sec	accuracy=0.444632	lr=0.307540
Epoch[41] Batch [949]	Speed: 2090.485352 samples/sec	accuracy=0.444635	lr=0.307356
Epoch[41] Batch [999]	Speed: 2392.183548 samples/sec	accuracy=0.444628	lr=0.307172
Epoch[41] Batch [1049]	Speed: 4182.010604 samples/sec	accuracy=0.444726	lr=0.306988
Epoch[41] Batch [1099]	Speed: 5435.070516 samples/sec	accuracy=0.444632	lr=0.306803
Epoch[41] Batch [1149]	Speed: 1600.093907 samples/sec	accuracy=0.444530	lr=0.306618
Epoch[41] Batch [1199]	Speed: 5394.470684 samples/sec	accuracy=0.444307	lr=0.306433
Epoch[41] Batch [1249]	Speed: 5116.688544 samples/sec	accuracy=0.444366	lr=0.306249
[Epoch 41] training: accuracy=0.444380
[Epoch 41] speed: 3242 samples/sec	time cost: 413.937875
[Epoch 41] validation: err-top1=0.601742 err-top5=0.340422
Epoch[42] Batch [49]	Speed: 4289.901094 samples/sec	accuracy=0.451504	lr=0.306060
Epoch[42] Batch [99]	Speed: 5070.906088 samples/sec	accuracy=0.447754	lr=0.305875
Epoch[42] Batch [149]	Speed: 5480.690171 samples/sec	accuracy=0.447988	lr=0.305689
Epoch[42] Batch [199]	Speed: 2896.846156 samples/sec	accuracy=0.447852	lr=0.305504
Epoch[42] Batch [249]	Speed: 4336.880820 samples/sec	accuracy=0.448047	lr=0.305318
Epoch[42] Batch [299]	Speed: 1340.966086 samples/sec	accuracy=0.448389	lr=0.305133
Epoch[42] Batch [349]	Speed: 4770.578094 samples/sec	accuracy=0.448075	lr=0.304947
Epoch[42] Batch [399]	Speed: 5492.838481 samples/sec	accuracy=0.447869	lr=0.304761
Epoch[42] Batch [449]	Speed: 4693.236879 samples/sec	accuracy=0.447426	lr=0.304575
Epoch[42] Batch [499]	Speed: 3482.969065 samples/sec	accuracy=0.447193	lr=0.304388
Epoch[42] Batch [549]	Speed: 1711.318949 samples/sec	accuracy=0.447047	lr=0.304202
Epoch[42] Batch [599]	Speed: 4203.309509 samples/sec	accuracy=0.446624	lr=0.304016
Epoch[42] Batch [649]	Speed: 4291.901248 samples/sec	accuracy=0.446531	lr=0.303829
Epoch[42] Batch [699]	Speed: 3715.578811 samples/sec	accuracy=0.446737	lr=0.303642
Epoch[42] Batch [749]	Speed: 3164.620164 samples/sec	accuracy=0.446728	lr=0.303456
Epoch[42] Batch [799]	Speed: 3925.727428 samples/sec	accuracy=0.446630	lr=0.303269
Epoch[42] Batch [849]	Speed: 2450.331122 samples/sec	accuracy=0.446573	lr=0.303082
Epoch[42] Batch [899]	Speed: 3521.396550 samples/sec	accuracy=0.446620	lr=0.302894
Epoch[42] Batch [949]	Speed: 3280.697981 samples/sec	accuracy=0.446468	lr=0.302707
Epoch[42] Batch [999]	Speed: 3239.987902 samples/sec	accuracy=0.446187	lr=0.302520
Epoch[42] Batch [1049]	Speed: 4459.000827 samples/sec	accuracy=0.446067	lr=0.302332
Epoch[42] Batch [1099]	Speed: 3655.785267 samples/sec	accuracy=0.446032	lr=0.302144
Epoch[42] Batch [1149]	Speed: 3367.784710 samples/sec	accuracy=0.445987	lr=0.301957
Epoch[42] Batch [1199]	Speed: 4032.912285 samples/sec	accuracy=0.445811	lr=0.301769
Epoch[42] Batch [1249]	Speed: 2241.304357 samples/sec	accuracy=0.445688	lr=0.301581
[Epoch 42] training: accuracy=0.445663
[Epoch 42] speed: 3322 samples/sec	time cost: 397.037406
[Epoch 42] validation: err-top1=0.595085 err-top5=0.339146
Epoch[43] Batch [49]	Speed: 2077.708403 samples/sec	accuracy=0.453184	lr=0.301389
Epoch[43] Batch [99]	Speed: 4260.900206 samples/sec	accuracy=0.451709	lr=0.301200
Epoch[43] Batch [149]	Speed: 1739.034804 samples/sec	accuracy=0.451087	lr=0.301012
Epoch[43] Batch [199]	Speed: 3435.299803 samples/sec	accuracy=0.450835	lr=0.300823
Epoch[43] Batch [249]	Speed: 3387.034120 samples/sec	accuracy=0.449848	lr=0.300635
Epoch[43] Batch [299]	Speed: 1892.217441 samples/sec	accuracy=0.449284	lr=0.300446
Epoch[43] Batch [349]	Speed: 5588.608718 samples/sec	accuracy=0.449146	lr=0.300257
Epoch[43] Batch [399]	Speed: 4227.887381 samples/sec	accuracy=0.449009	lr=0.300068
Epoch[43] Batch [449]	Speed: 3114.119204 samples/sec	accuracy=0.448750	lr=0.299879
Epoch[43] Batch [499]	Speed: 5711.560686 samples/sec	accuracy=0.448516	lr=0.299690
Epoch[43] Batch [549]	Speed: 4896.952061 samples/sec	accuracy=0.448173	lr=0.299500
Epoch[43] Batch [599]	Speed: 2171.067301 samples/sec	accuracy=0.448172	lr=0.299311
Epoch[43] Batch [649]	Speed: 2880.301623 samples/sec	accuracy=0.448006	lr=0.299121
Epoch[43] Batch [699]	Speed: 4667.347104 samples/sec	accuracy=0.448030	lr=0.298932
Epoch[43] Batch [749]	Speed: 3399.815628 samples/sec	accuracy=0.447982	lr=0.298742
Epoch[43] Batch [799]	Speed: 1856.192637 samples/sec	accuracy=0.447866	lr=0.298552
Epoch[43] Batch [849]	Speed: 3188.730633 samples/sec	accuracy=0.447800	lr=0.298362
Epoch[43] Batch [899]	Speed: 2334.388164 samples/sec	accuracy=0.447742	lr=0.298171
Epoch[43] Batch [949]	Speed: 4313.306007 samples/sec	accuracy=0.447461	lr=0.297981
Epoch[43] Batch [999]	Speed: 3128.797189 samples/sec	accuracy=0.447342	lr=0.297791
Epoch[43] Batch [1049]	Speed: 4303.733312 samples/sec	accuracy=0.447114	lr=0.297600
Epoch[43] Batch [1099]	Speed: 2405.694999 samples/sec	accuracy=0.447044	lr=0.297409
Epoch[43] Batch [1149]	Speed: 3941.214956 samples/sec	accuracy=0.446984	lr=0.297219
Epoch[43] Batch [1199]	Speed: 3394.800018 samples/sec	accuracy=0.446892	lr=0.297028
Epoch[43] Batch [1249]	Speed: 3767.688486 samples/sec	accuracy=0.446850	lr=0.296837
[Epoch 43] training: accuracy=0.446856
[Epoch 43] speed: 3071 samples/sec	time cost: 447.226879
[Epoch 43] validation: err-top1=0.783681 err-top5=0.577706
Epoch[44] Batch [49]	Speed: 2515.245959 samples/sec	accuracy=0.452461	lr=0.296642
Epoch[44] Batch [99]	Speed: 2969.031457 samples/sec	accuracy=0.451504	lr=0.296451
Epoch[44] Batch [149]	Speed: 4920.847870 samples/sec	accuracy=0.451576	lr=0.296259
Epoch[44] Batch [199]	Speed: 4897.168145 samples/sec	accuracy=0.451543	lr=0.296068
Epoch[44] Batch [249]	Speed: 2992.359091 samples/sec	accuracy=0.450645	lr=0.295876
Epoch[44] Batch [299]	Speed: 3573.878902 samples/sec	accuracy=0.451276	lr=0.295684
Epoch[44] Batch [349]	Speed: 5691.184284 samples/sec	accuracy=0.450547	lr=0.295493
Epoch[44] Batch [399]	Speed: 3539.124654 samples/sec	accuracy=0.449817	lr=0.295301
Epoch[44] Batch [449]	Speed: 2225.535907 samples/sec	accuracy=0.449944	lr=0.295109
Epoch[44] Batch [499]	Speed: 2441.876332 samples/sec	accuracy=0.449484	lr=0.294917
Epoch[44] Batch [549]	Speed: 3704.153610 samples/sec	accuracy=0.449339	lr=0.294724
Epoch[44] Batch [599]	Speed: 2155.381613 samples/sec	accuracy=0.449409	lr=0.294532
Epoch[44] Batch [649]	Speed: 2475.123064 samples/sec	accuracy=0.449246	lr=0.294339
Epoch[44] Batch [699]	Speed: 2982.204882 samples/sec	accuracy=0.449054	lr=0.294147
Epoch[44] Batch [749]	Speed: 2060.372045 samples/sec	accuracy=0.448952	lr=0.293954
Epoch[44] Batch [799]	Speed: 1999.734727 samples/sec	accuracy=0.448793	lr=0.293761
Epoch[44] Batch [849]	Speed: 3994.500118 samples/sec	accuracy=0.448657	lr=0.293568
Epoch[44] Batch [899]	Speed: 4445.874688 samples/sec	accuracy=0.448456	lr=0.293375
Epoch[44] Batch [949]	Speed: 3893.561295 samples/sec	accuracy=0.448521	lr=0.293182
Epoch[44] Batch [999]	Speed: 2724.226613 samples/sec	accuracy=0.448155	lr=0.292989
Epoch[44] Batch [1049]	Speed: 3625.244059 samples/sec	accuracy=0.448038	lr=0.292795
Epoch[44] Batch [1099]	Speed: 4148.166023 samples/sec	accuracy=0.447964	lr=0.292602
Epoch[44] Batch [1149]	Speed: 3391.489862 samples/sec	accuracy=0.448088	lr=0.292408
Epoch[44] Batch [1199]	Speed: 2707.610805 samples/sec	accuracy=0.448010	lr=0.292215
Epoch[44] Batch [1249]	Speed: 3194.739849 samples/sec	accuracy=0.447808	lr=0.292021
[Epoch 44] training: accuracy=0.447811
[Epoch 44] speed: 3074 samples/sec	time cost: 432.258089
[Epoch 44] validation: err-top1=0.553711 err-top5=0.296177
Epoch[45] Batch [49]	Speed: 5253.545750 samples/sec	accuracy=0.456094	lr=0.291823
Epoch[45] Batch [99]	Speed: 4143.641588 samples/sec	accuracy=0.455869	lr=0.291629
Epoch[45] Batch [149]	Speed: 3618.249007 samples/sec	accuracy=0.455911	lr=0.291435
Epoch[45] Batch [199]	Speed: 2611.204029 samples/sec	accuracy=0.454663	lr=0.291240
Epoch[45] Batch [249]	Speed: 2174.410482 samples/sec	accuracy=0.454074	lr=0.291046
Epoch[45] Batch [299]	Speed: 3471.844826 samples/sec	accuracy=0.454121	lr=0.290852
Epoch[45] Batch [349]	Speed: 2224.008673 samples/sec	accuracy=0.453083	lr=0.290657
Epoch[45] Batch [399]	Speed: 3577.066384 samples/sec	accuracy=0.452329	lr=0.290462
Epoch[45] Batch [449]	Speed: 3721.407069 samples/sec	accuracy=0.452040	lr=0.290267
Epoch[45] Batch [499]	Speed: 3879.879477 samples/sec	accuracy=0.452166	lr=0.290073
Epoch[45] Batch [549]	Speed: 3205.201867 samples/sec	accuracy=0.451763	lr=0.289878
Epoch[45] Batch [599]	Speed: 3742.181261 samples/sec	accuracy=0.451774	lr=0.289682
Epoch[45] Batch [649]	Speed: 3922.829990 samples/sec	accuracy=0.451378	lr=0.289487
Epoch[45] Batch [699]	Speed: 4243.763740 samples/sec	accuracy=0.451387	lr=0.289292
Epoch[45] Batch [749]	Speed: 3667.522488 samples/sec	accuracy=0.451150	lr=0.289096
Epoch[45] Batch [799]	Speed: 3435.329754 samples/sec	accuracy=0.451090	lr=0.288901
Epoch[45] Batch [849]	Speed: 2962.194153 samples/sec	accuracy=0.450872	lr=0.288705
Epoch[45] Batch [899]	Speed: 3013.511493 samples/sec	accuracy=0.450961	lr=0.288509
Epoch[45] Batch [949]	Speed: 3304.654939 samples/sec	accuracy=0.450989	lr=0.288314
Epoch[45] Batch [999]	Speed: 3270.575347 samples/sec	accuracy=0.450909	lr=0.288118
Epoch[45] Batch [1049]	Speed: 3266.085350 samples/sec	accuracy=0.450695	lr=0.287921
Epoch[45] Batch [1099]	Speed: 4346.170047 samples/sec	accuracy=0.450612	lr=0.287725
Epoch[45] Batch [1149]	Speed: 2380.114623 samples/sec	accuracy=0.450476	lr=0.287529
Epoch[45] Batch [1199]	Speed: 3754.296557 samples/sec	accuracy=0.450476	lr=0.287333
Epoch[45] Batch [1249]	Speed: 2763.885468 samples/sec	accuracy=0.450395	lr=0.287136
[Epoch 45] training: accuracy=0.450401
[Epoch 45] speed: 3289 samples/sec	time cost: 406.837795
[Epoch 45] validation: err-top1=0.602918 err-top5=0.345085
Epoch[46] Batch [49]	Speed: 4566.522400 samples/sec	accuracy=0.456445	lr=0.286936
Epoch[46] Batch [99]	Speed: 2746.760477 samples/sec	accuracy=0.456299	lr=0.286739
Epoch[46] Batch [149]	Speed: 3445.659392 samples/sec	accuracy=0.455788	lr=0.286542
Epoch[46] Batch [199]	Speed: 4154.455148 samples/sec	accuracy=0.454575	lr=0.286345
Epoch[46] Batch [249]	Speed: 2869.456376 samples/sec	accuracy=0.454695	lr=0.286148
Epoch[46] Batch [299]	Speed: 4170.687644 samples/sec	accuracy=0.454150	lr=0.285951
Epoch[46] Batch [349]	Speed: 4220.111797 samples/sec	accuracy=0.453186	lr=0.285754
Epoch[46] Batch [399]	Speed: 3001.516477 samples/sec	accuracy=0.453079	lr=0.285556
Epoch[46] Batch [449]	Speed: 2815.942866 samples/sec	accuracy=0.453114	lr=0.285359
Epoch[46] Batch [499]	Speed: 4197.297936 samples/sec	accuracy=0.452955	lr=0.285161
Epoch[46] Batch [549]	Speed: 3781.177095 samples/sec	accuracy=0.452805	lr=0.284964
Epoch[46] Batch [599]	Speed: 3782.258349 samples/sec	accuracy=0.452708	lr=0.284766
Epoch[46] Batch [649]	Speed: 2843.060374 samples/sec	accuracy=0.452728	lr=0.284568
Epoch[46] Batch [699]	Speed: 5240.472146 samples/sec	accuracy=0.452508	lr=0.284370
Epoch[46] Batch [749]	Speed: 3912.743369 samples/sec	accuracy=0.452437	lr=0.284172
Epoch[46] Batch [799]	Speed: 3267.132606 samples/sec	accuracy=0.452307	lr=0.283974
Epoch[46] Batch [849]	Speed: 3837.629882 samples/sec	accuracy=0.452426	lr=0.283776
Epoch[46] Batch [899]	Speed: 3743.961581 samples/sec	accuracy=0.452523	lr=0.283577
Epoch[46] Batch [949]	Speed: 4528.737081 samples/sec	accuracy=0.452492	lr=0.283379
Epoch[46] Batch [999]	Speed: 3563.773964 samples/sec	accuracy=0.452277	lr=0.283181
Epoch[46] Batch [1049]	Speed: 4292.911077 samples/sec	accuracy=0.452217	lr=0.282982
Epoch[46] Batch [1099]	Speed: 3528.113948 samples/sec	accuracy=0.451958	lr=0.282783
Epoch[46] Batch [1149]	Speed: 5107.137133 samples/sec	accuracy=0.451842	lr=0.282584
Epoch[46] Batch [1199]	Speed: 3647.553969 samples/sec	accuracy=0.451701	lr=0.282385
Epoch[46] Batch [1249]	Speed: 3544.067148 samples/sec	accuracy=0.451649	lr=0.282186
[Epoch 46] training: accuracy=0.451661
[Epoch 46] speed: 3671 samples/sec	time cost: 363.882371
[Epoch 46] validation: err-top1=0.575033 err-top5=0.317118
Epoch[47] Batch [49]	Speed: 3409.323509 samples/sec	accuracy=0.459590	lr=0.281983
Epoch[47] Batch [99]	Speed: 2400.643375 samples/sec	accuracy=0.458564	lr=0.281784
Epoch[47] Batch [149]	Speed: 2888.314398 samples/sec	accuracy=0.457865	lr=0.281585
Epoch[47] Batch [199]	Speed: 3814.149574 samples/sec	accuracy=0.456475	lr=0.281385
Epoch[47] Batch [249]	Speed: 3085.412435 samples/sec	accuracy=0.456824	lr=0.281186
Epoch[47] Batch [299]	Speed: 3973.849867 samples/sec	accuracy=0.456673	lr=0.280986
Epoch[47] Batch [349]	Speed: 2560.574470 samples/sec	accuracy=0.456842	lr=0.280786
Epoch[47] Batch [399]	Speed: 2427.354297 samples/sec	accuracy=0.456499	lr=0.280587
Epoch[47] Batch [449]	Speed: 3048.950067 samples/sec	accuracy=0.456283	lr=0.280387
Epoch[47] Batch [499]	Speed: 3346.815424 samples/sec	accuracy=0.455965	lr=0.280187
Epoch[47] Batch [549]	Speed: 3432.457847 samples/sec	accuracy=0.455650	lr=0.279987
Epoch[47] Batch [599]	Speed: 4490.379637 samples/sec	accuracy=0.455288	lr=0.279786
Epoch[47] Batch [649]	Speed: 3411.169938 samples/sec	accuracy=0.455098	lr=0.279586
Epoch[47] Batch [699]	Speed: 3598.645029 samples/sec	accuracy=0.454881	lr=0.279386
Epoch[47] Batch [749]	Speed: 4061.540568 samples/sec	accuracy=0.454828	lr=0.279185
Epoch[47] Batch [799]	Speed: 4076.780375 samples/sec	accuracy=0.454752	lr=0.278985
Epoch[47] Batch [849]	Speed: 2519.198696 samples/sec	accuracy=0.454545	lr=0.278784
Epoch[47] Batch [899]	Speed: 4381.954080 samples/sec	accuracy=0.454425	lr=0.278583
Epoch[47] Batch [949]	Speed: 3571.740493 samples/sec	accuracy=0.454389	lr=0.278382
Epoch[47] Batch [999]	Speed: 3801.262245 samples/sec	accuracy=0.454116	lr=0.278181
Epoch[47] Batch [1049]	Speed: 3588.486404 samples/sec	accuracy=0.454098	lr=0.277980
Epoch[47] Batch [1099]	Speed: 5134.715145 samples/sec	accuracy=0.453895	lr=0.277779
Epoch[47] Batch [1149]	Speed: 2979.884121 samples/sec	accuracy=0.453845	lr=0.277578
Epoch[47] Batch [1199]	Speed: 3805.671974 samples/sec	accuracy=0.453626	lr=0.277377
Epoch[47] Batch [1249]	Speed: 2670.146909 samples/sec	accuracy=0.453550	lr=0.277175
[Epoch 47] training: accuracy=0.453545
[Epoch 47] speed: 3321 samples/sec	time cost: 396.785834
[Epoch 47] validation: err-top1=0.572724 err-top5=0.313915
Epoch[48] Batch [49]	Speed: 3434.264452 samples/sec	accuracy=0.463125	lr=0.276970
Epoch[48] Batch [99]	Speed: 2976.049112 samples/sec	accuracy=0.460615	lr=0.276768
Epoch[48] Batch [149]	Speed: 4679.170219 samples/sec	accuracy=0.459850	lr=0.276566
Epoch[48] Batch [199]	Speed: 5803.270648 samples/sec	accuracy=0.458696	lr=0.276365
Epoch[48] Batch [249]	Speed: 3136.424474 samples/sec	accuracy=0.457699	lr=0.276163
Epoch[48] Batch [299]	Speed: 3805.816239 samples/sec	accuracy=0.457454	lr=0.275961
Epoch[48] Batch [349]	Speed: 5643.510703 samples/sec	accuracy=0.456911	lr=0.275759
Epoch[48] Batch [399]	Speed: 3887.965503 samples/sec	accuracy=0.456279	lr=0.275557
Epoch[48] Batch [449]	Speed: 3461.557907 samples/sec	accuracy=0.456261	lr=0.275354
Epoch[48] Batch [499]	Speed: 5385.177937 samples/sec	accuracy=0.456363	lr=0.275152
Epoch[48] Batch [549]	Speed: 3378.179084 samples/sec	accuracy=0.455820	lr=0.274950
Epoch[48] Batch [599]	Speed: 4303.512178 samples/sec	accuracy=0.455575	lr=0.274747
Epoch[48] Batch [649]	Speed: 4250.860537 samples/sec	accuracy=0.455353	lr=0.274545
Epoch[48] Batch [699]	Speed: 3691.784432 samples/sec	accuracy=0.455172	lr=0.274342
Epoch[48] Batch [749]	Speed: 4889.734560 samples/sec	accuracy=0.455081	lr=0.274139
Epoch[48] Batch [799]	Speed: 3848.444379 samples/sec	accuracy=0.455104	lr=0.273936
Epoch[48] Batch [849]	Speed: 4033.152234 samples/sec	accuracy=0.455017	lr=0.273733
Epoch[48] Batch [899]	Speed: 2327.473202 samples/sec	accuracy=0.455107	lr=0.273530
Epoch[48] Batch [949]	Speed: 4118.484234 samples/sec	accuracy=0.455032	lr=0.273327
Epoch[48] Batch [999]	Speed: 4283.211252 samples/sec	accuracy=0.455103	lr=0.273124
Epoch[48] Batch [1049]	Speed: 3931.439724 samples/sec	accuracy=0.454925	lr=0.272921
Epoch[48] Batch [1099]	Speed: 4041.011162 samples/sec	accuracy=0.454849	lr=0.272717
Epoch[48] Batch [1149]	Speed: 3098.496907 samples/sec	accuracy=0.454716	lr=0.272514
Epoch[48] Batch [1199]	Speed: 3051.258388 samples/sec	accuracy=0.454559	lr=0.272310
Epoch[48] Batch [1249]	Speed: 3839.145820 samples/sec	accuracy=0.454503	lr=0.272107
[Epoch 48] training: accuracy=0.454505
[Epoch 48] speed: 3799 samples/sec	time cost: 350.899937
[Epoch 48] validation: err-top1=0.576730 err-top5=0.322086
Epoch[49] Batch [49]	Speed: 3254.375715 samples/sec	accuracy=0.459395	lr=0.271899
Epoch[49] Batch [99]	Speed: 2848.953073 samples/sec	accuracy=0.460596	lr=0.271695
Epoch[49] Batch [149]	Speed: 4006.953639 samples/sec	accuracy=0.459720	lr=0.271491
Epoch[49] Batch [199]	Speed: 5120.398835 samples/sec	accuracy=0.459854	lr=0.271287
Epoch[49] Batch [249]	Speed: 5566.066821 samples/sec	accuracy=0.459961	lr=0.271083
Epoch[49] Batch [299]	Speed: 4931.228638 samples/sec	accuracy=0.459821	lr=0.270879
Epoch[49] Batch [349]	Speed: 3527.430168 samples/sec	accuracy=0.459269	lr=0.270675
Epoch[49] Batch [399]	Speed: 5299.346896 samples/sec	accuracy=0.458972	lr=0.270470
Epoch[49] Batch [449]	Speed: 2682.679193 samples/sec	accuracy=0.458639	lr=0.270266
Epoch[49] Batch [499]	Speed: 3830.227555 samples/sec	accuracy=0.458160	lr=0.270061
Epoch[49] Batch [549]	Speed: 4691.054616 samples/sec	accuracy=0.457974	lr=0.269857
Epoch[49] Batch [599]	Speed: 2611.043159 samples/sec	accuracy=0.458055	lr=0.269652
Epoch[49] Batch [649]	Speed: 3558.978866 samples/sec	accuracy=0.457819	lr=0.269447
Epoch[49] Batch [699]	Speed: 5419.147839 samples/sec	accuracy=0.457753	lr=0.269243
Epoch[49] Batch [749]	Speed: 3890.955954 samples/sec	accuracy=0.457579	lr=0.269038
Epoch[49] Batch [799]	Speed: 3875.380542 samples/sec	accuracy=0.457708	lr=0.268833
Epoch[49] Batch [849]	Speed: 3915.222291 samples/sec	accuracy=0.457248	lr=0.268628
Epoch[49] Batch [899]	Speed: 3982.524472 samples/sec	accuracy=0.456924	lr=0.268422
Epoch[49] Batch [949]	Speed: 4070.708852 samples/sec	accuracy=0.456758	lr=0.268217
Epoch[49] Batch [999]	Speed: 5743.710063 samples/sec	accuracy=0.456608	lr=0.268012
Epoch[49] Batch [1049]	Speed: 5094.491855 samples/sec	accuracy=0.456455	lr=0.267807
Epoch[49] Batch [1099]	Speed: 4397.309215 samples/sec	accuracy=0.456398	lr=0.267601
Epoch[49] Batch [1149]	Speed: 2530.270332 samples/sec	accuracy=0.456449	lr=0.267396
Epoch[49] Batch [1199]	Speed: 3615.060948 samples/sec	accuracy=0.456376	lr=0.267190
Epoch[49] Batch [1249]	Speed: 5284.956694 samples/sec	accuracy=0.456146	lr=0.266984
[Epoch 49] training: accuracy=0.456140
[Epoch 49] speed: 3915 samples/sec	time cost: 338.215348
[Epoch 49] validation: err-top1=0.579978 err-top5=0.314573
Epoch[50] Batch [49]	Speed: 3958.763597 samples/sec	accuracy=0.463281	lr=0.266774
Epoch[50] Batch [99]	Speed: 4682.727373 samples/sec	accuracy=0.464600	lr=0.266568
Epoch[50] Batch [149]	Speed: 4177.385913 samples/sec	accuracy=0.462689	lr=0.266362
Epoch[50] Batch [199]	Speed: 3686.470004 samples/sec	accuracy=0.461968	lr=0.266156
Epoch[50] Batch [249]	Speed: 2529.987142 samples/sec	accuracy=0.461922	lr=0.265950
Epoch[50] Batch [299]	Speed: 2952.089279 samples/sec	accuracy=0.461055	lr=0.265744
Epoch[50] Batch [349]	Speed: 4284.823234 samples/sec	accuracy=0.460759	lr=0.265538
Epoch[50] Batch [399]	Speed: 3213.900944 samples/sec	accuracy=0.460417	lr=0.265331
Epoch[50] Batch [449]	Speed: 4234.806450 samples/sec	accuracy=0.460577	lr=0.265125
Epoch[50] Batch [499]	Speed: 2096.188020 samples/sec	accuracy=0.460174	lr=0.264918
Epoch[50] Batch [549]	Speed: 3073.801965 samples/sec	accuracy=0.459560	lr=0.264712
Epoch[50] Batch [599]	Speed: 5377.398324 samples/sec	accuracy=0.459432	lr=0.264505
Epoch[50] Batch [649]	Speed: 4943.574675 samples/sec	accuracy=0.458920	lr=0.264298
Epoch[50] Batch [699]	Speed: 5308.500637 samples/sec	accuracy=0.458700	lr=0.264092
Epoch[50] Batch [749]	Speed: 2964.567561 samples/sec	accuracy=0.458628	lr=0.263885
Epoch[50] Batch [799]	Speed: 3677.752461 samples/sec	accuracy=0.458263	lr=0.263678
Epoch[50] Batch [849]	Speed: 3666.441229 samples/sec	accuracy=0.458189	lr=0.263471
Epoch[50] Batch [899]	Speed: 5810.059545 samples/sec	accuracy=0.457995	lr=0.263264
Epoch[50] Batch [949]	Speed: 4089.002599 samples/sec	accuracy=0.457867	lr=0.263056
Epoch[50] Batch [999]	Speed: 3759.586205 samples/sec	accuracy=0.457764	lr=0.262849
Epoch[50] Batch [1049]	Speed: 3935.934933 samples/sec	accuracy=0.457514	lr=0.262642
Epoch[50] Batch [1099]	Speed: 4109.257459 samples/sec	accuracy=0.457463	lr=0.262434
Epoch[50] Batch [1149]	Speed: 4322.410975 samples/sec	accuracy=0.457255	lr=0.262227
Epoch[50] Batch [1199]	Speed: 4137.084246 samples/sec	accuracy=0.457223	lr=0.262019
Epoch[50] Batch [1249]	Speed: 3988.184054 samples/sec	accuracy=0.456966	lr=0.261812
[Epoch 50] training: accuracy=0.456981
[Epoch 50] speed: 3747 samples/sec	time cost: 353.943023
[Epoch 50] validation: err-top1=0.534518 err-top5=0.276846
Epoch[51] Batch [49]	Speed: 4405.019845 samples/sec	accuracy=0.462734	lr=0.261600
Epoch[51] Batch [99]	Speed: 4155.010745 samples/sec	accuracy=0.461143	lr=0.261392
Epoch[51] Batch [149]	Speed: 4217.236790 samples/sec	accuracy=0.460840	lr=0.261184
Epoch[51] Batch [199]	Speed: 5578.400497 samples/sec	accuracy=0.461792	lr=0.260976
Epoch[51] Batch [249]	Speed: 3836.947428 samples/sec	accuracy=0.461441	lr=0.260768
Epoch[51] Batch [299]	Speed: 4442.106108 samples/sec	accuracy=0.461377	lr=0.260560
Epoch[51] Batch [349]	Speed: 4542.522700 samples/sec	accuracy=0.460536	lr=0.260352
Epoch[51] Batch [399]	Speed: 4312.724075 samples/sec	accuracy=0.460840	lr=0.260144
Epoch[51] Batch [449]	Speed: 3541.133315 samples/sec	accuracy=0.460306	lr=0.259935
Epoch[51] Batch [499]	Speed: 3978.103993 samples/sec	accuracy=0.459867	lr=0.259727
Epoch[51] Batch [549]	Speed: 4301.327066 samples/sec	accuracy=0.459542	lr=0.259519
Epoch[51] Batch [599]	Speed: 2009.797224 samples/sec	accuracy=0.459536	lr=0.259310
Epoch[51] Batch [649]	Speed: 5368.518403 samples/sec	accuracy=0.459587	lr=0.259102
Epoch[51] Batch [699]	Speed: 4860.417695 samples/sec	accuracy=0.459466	lr=0.258893
Epoch[51] Batch [749]	Speed: 5964.918866 samples/sec	accuracy=0.459418	lr=0.258684
Epoch[51] Batch [799]	Speed: 5585.868703 samples/sec	accuracy=0.459232	lr=0.258475
Epoch[51] Batch [849]	Speed: 4423.275732 samples/sec	accuracy=0.459351	lr=0.258267
Epoch[51] Batch [899]	Speed: 3876.260393 samples/sec	accuracy=0.459206	lr=0.258058
Epoch[51] Batch [949]	Speed: 2873.939397 samples/sec	accuracy=0.459175	lr=0.257849
Epoch[51] Batch [999]	Speed: 5229.161876 samples/sec	accuracy=0.459177	lr=0.257640
Epoch[51] Batch [1049]	Speed: 4753.957205 samples/sec	accuracy=0.459026	lr=0.257430
Epoch[51] Batch [1099]	Speed: 3651.119531 samples/sec	accuracy=0.458766	lr=0.257221
Epoch[51] Batch [1149]	Speed: 4268.350545 samples/sec	accuracy=0.458460	lr=0.257012
Epoch[51] Batch [1199]	Speed: 3515.722460 samples/sec	accuracy=0.458419	lr=0.256803
Epoch[51] Batch [1249]	Speed: 5328.702421 samples/sec	accuracy=0.458392	lr=0.256593
[Epoch 51] training: accuracy=0.458404
[Epoch 51] speed: 4136 samples/sec	time cost: 321.194111
[Epoch 51] validation: err-top1=0.547772 err-top5=0.291793
Epoch[52] Batch [49]	Speed: 5814.248594 samples/sec	accuracy=0.467344	lr=0.256379
Epoch[52] Batch [99]	Speed: 3225.859080 samples/sec	accuracy=0.467080	lr=0.256170
Epoch[52] Batch [149]	Speed: 5630.663492 samples/sec	accuracy=0.465397	lr=0.255960
Epoch[52] Batch [199]	Speed: 4562.587315 samples/sec	accuracy=0.464312	lr=0.255751
Epoch[52] Batch [249]	Speed: 3495.187571 samples/sec	accuracy=0.463602	lr=0.255541
Epoch[52] Batch [299]	Speed: 5873.959670 samples/sec	accuracy=0.463258	lr=0.255331
Epoch[52] Batch [349]	Speed: 5137.295779 samples/sec	accuracy=0.462813	lr=0.255121
Epoch[52] Batch [399]	Speed: 5766.009364 samples/sec	accuracy=0.462676	lr=0.254911
Epoch[52] Batch [449]	Speed: 5748.691137 samples/sec	accuracy=0.462741	lr=0.254701
Epoch[52] Batch [499]	Speed: 4645.836276 samples/sec	accuracy=0.462096	lr=0.254491
Epoch[52] Batch [549]	Speed: 4897.500739 samples/sec	accuracy=0.461944	lr=0.254281
Epoch[52] Batch [599]	Speed: 4638.116890 samples/sec	accuracy=0.462082	lr=0.254071
Epoch[52] Batch [649]	Speed: 4890.265140 samples/sec	accuracy=0.462078	lr=0.253861
Epoch[52] Batch [699]	Speed: 3454.100681 samples/sec	accuracy=0.461943	lr=0.253650
Epoch[52] Batch [749]	Speed: 3466.698772 samples/sec	accuracy=0.461652	lr=0.253440
Epoch[52] Batch [799]	Speed: 5107.368520 samples/sec	accuracy=0.461573	lr=0.253229
Epoch[52] Batch [849]	Speed: 4967.699754 samples/sec	accuracy=0.461186	lr=0.253019
Epoch[52] Batch [899]	Speed: 4219.292513 samples/sec	accuracy=0.461061	lr=0.252808
Epoch[52] Batch [949]	Speed: 4420.153923 samples/sec	accuracy=0.460895	lr=0.252598
Epoch[52] Batch [999]	Speed: 4330.874752 samples/sec	accuracy=0.460739	lr=0.252387
Epoch[52] Batch [1049]	Speed: 5403.206976 samples/sec	accuracy=0.460609	lr=0.252176
Epoch[52] Batch [1099]	Speed: 3561.233236 samples/sec	accuracy=0.460461	lr=0.251965
Epoch[52] Batch [1149]	Speed: 4903.188288 samples/sec	accuracy=0.460284	lr=0.251754
Epoch[52] Batch [1199]	Speed: 5562.360619 samples/sec	accuracy=0.460129	lr=0.251543
Epoch[52] Batch [1249]	Speed: 5043.214508 samples/sec	accuracy=0.459937	lr=0.251332
[Epoch 52] training: accuracy=0.459925
[Epoch 52] speed: 4594 samples/sec	time cost: 289.773259
[Epoch 52] validation: err-top1=0.541361 err-top5=0.279582
Epoch[53] Batch [49]	Speed: 4798.142020 samples/sec	accuracy=0.470898	lr=0.251117
Epoch[53] Batch [99]	Speed: 5112.365649 samples/sec	accuracy=0.468906	lr=0.250906
Epoch[53] Batch [149]	Speed: 4469.315936 samples/sec	accuracy=0.468665	lr=0.250695
Epoch[53] Batch [199]	Speed: 4812.149395 samples/sec	accuracy=0.467637	lr=0.250483
Epoch[53] Batch [249]	Speed: 5941.734775 samples/sec	accuracy=0.466305	lr=0.250272
Epoch[53] Batch [299]	Speed: 4716.720501 samples/sec	accuracy=0.466035	lr=0.250061
Epoch[53] Batch [349]	Speed: 4514.706172 samples/sec	accuracy=0.465882	lr=0.249849
Epoch[53] Batch [399]	Speed: 4174.292020 samples/sec	accuracy=0.465400	lr=0.249638
Epoch[53] Batch [449]	Speed: 5327.586281 samples/sec	accuracy=0.465213	lr=0.249426
Epoch[53] Batch [499]	Speed: 4260.622926 samples/sec	accuracy=0.464738	lr=0.249215
Epoch[53] Batch [549]	Speed: 4901.008792 samples/sec	accuracy=0.464551	lr=0.249003
Epoch[53] Batch [599]	Speed: 4471.894213 samples/sec	accuracy=0.464069	lr=0.248791
Epoch[53] Batch [649]	Speed: 3916.205097 samples/sec	accuracy=0.463767	lr=0.248579
Epoch[53] Batch [699]	Speed: 4960.981206 samples/sec	accuracy=0.463640	lr=0.248367
Epoch[53] Batch [749]	Speed: 4798.614735 samples/sec	accuracy=0.463298	lr=0.248156
Epoch[53] Batch [799]	Speed: 5361.543417 samples/sec	accuracy=0.463232	lr=0.247944
Epoch[53] Batch [849]	Speed: 5223.919316 samples/sec	accuracy=0.463104	lr=0.247732
Epoch[53] Batch [899]	Speed: 3832.369213 samples/sec	accuracy=0.462907	lr=0.247519
Epoch[53] Batch [949]	Speed: 4638.333576 samples/sec	accuracy=0.462698	lr=0.247307
Epoch[53] Batch [999]	Speed: 5593.241843 samples/sec	accuracy=0.462517	lr=0.247095
Epoch[53] Batch [1049]	Speed: 4816.335412 samples/sec	accuracy=0.462491	lr=0.246883
Epoch[53] Batch [1099]	Speed: 5355.004325 samples/sec	accuracy=0.462543	lr=0.246670
Epoch[53] Batch [1149]	Speed: 5840.889552 samples/sec	accuracy=0.462593	lr=0.246458
Epoch[53] Batch [1199]	Speed: 5617.252019 samples/sec	accuracy=0.462546	lr=0.246246
Epoch[53] Batch [1249]	Speed: 4504.480315 samples/sec	accuracy=0.462539	lr=0.246033
[Epoch 53] training: accuracy=0.462535
[Epoch 53] speed: 4811 samples/sec	time cost: 278.177128
[Epoch 53] validation: err-top1=0.565131 err-top5=0.307657
Epoch[54] Batch [49]	Speed: 5043.104246 samples/sec	accuracy=0.468184	lr=0.245816
Epoch[54] Batch [99]	Speed: 4597.161525 samples/sec	accuracy=0.466670	lr=0.245604
Epoch[54] Batch [149]	Speed: 3854.344561 samples/sec	accuracy=0.468092	lr=0.245391
Epoch[54] Batch [199]	Speed: 3847.972634 samples/sec	accuracy=0.467119	lr=0.245178
Epoch[54] Batch [249]	Speed: 4689.021708 samples/sec	accuracy=0.466957	lr=0.244966
Epoch[54] Batch [299]	Speed: 5521.446123 samples/sec	accuracy=0.467083	lr=0.244753
Epoch[54] Batch [349]	Speed: 4913.982339 samples/sec	accuracy=0.466978	lr=0.244540
Epoch[54] Batch [399]	Speed: 4489.647949 samples/sec	accuracy=0.466465	lr=0.244327
Epoch[54] Batch [449]	Speed: 4865.460547 samples/sec	accuracy=0.466222	lr=0.244114
Epoch[54] Batch [499]	Speed: 5394.304691 samples/sec	accuracy=0.465889	lr=0.243901
Epoch[54] Batch [549]	Speed: 5542.891578 samples/sec	accuracy=0.465598	lr=0.243688
Epoch[54] Batch [599]	Speed: 4851.871951 samples/sec	accuracy=0.465381	lr=0.243475
Epoch[54] Batch [649]	Speed: 4909.352428 samples/sec	accuracy=0.465018	lr=0.243262
Epoch[54] Batch [699]	Speed: 4827.217668 samples/sec	accuracy=0.465000	lr=0.243049
Epoch[54] Batch [749]	Speed: 4855.720682 samples/sec	accuracy=0.464948	lr=0.242835
Epoch[54] Batch [799]	Speed: 4319.063437 samples/sec	accuracy=0.464523	lr=0.242622
Epoch[54] Batch [849]	Speed: 5305.181036 samples/sec	accuracy=0.464337	lr=0.242409
Epoch[54] Batch [899]	Speed: 5162.941754 samples/sec	accuracy=0.464258	lr=0.242195
Epoch[54] Batch [949]	Speed: 5843.852933 samples/sec	accuracy=0.464096	lr=0.241982
Epoch[54] Batch [999]	Speed: 5414.670740 samples/sec	accuracy=0.464272	lr=0.241768
Epoch[54] Batch [1049]	Speed: 4849.848448 samples/sec	accuracy=0.464096	lr=0.241555
Epoch[54] Batch [1099]	Speed: 5539.240739 samples/sec	accuracy=0.464107	lr=0.241341
Epoch[54] Batch [1149]	Speed: 5258.394060 samples/sec	accuracy=0.463798	lr=0.241127
Epoch[54] Batch [1199]	Speed: 5004.495220 samples/sec	accuracy=0.463582	lr=0.240914
Epoch[54] Batch [1249]	Speed: 5468.337443 samples/sec	accuracy=0.463402	lr=0.240700
[Epoch 54] training: accuracy=0.463400
[Epoch 54] speed: 4915 samples/sec	time cost: 270.254606
[Epoch 54] validation: err-top1=0.548430 err-top5=0.292909
Epoch[55] Batch [49]	Speed: 4982.691536 samples/sec	accuracy=0.472500	lr=0.240482
Epoch[55] Batch [99]	Speed: 5654.102339 samples/sec	accuracy=0.470957	lr=0.240268
Epoch[55] Batch [149]	Speed: 6069.313495 samples/sec	accuracy=0.470150	lr=0.240054
Epoch[55] Batch [199]	Speed: 3063.392782 samples/sec	accuracy=0.469800	lr=0.239840
Epoch[55] Batch [249]	Speed: 5080.946297 samples/sec	accuracy=0.469047	lr=0.239626
Epoch[55] Batch [299]	Speed: 5293.347076 samples/sec	accuracy=0.468018	lr=0.239412
Epoch[55] Batch [349]	Speed: 5042.254527 samples/sec	accuracy=0.467503	lr=0.239198
Epoch[55] Batch [399]	Speed: 5411.335427 samples/sec	accuracy=0.467380	lr=0.238984
Epoch[55] Batch [449]	Speed: 5384.765143 samples/sec	accuracy=0.466929	lr=0.238769
Epoch[55] Batch [499]	Speed: 5350.500697 samples/sec	accuracy=0.467127	lr=0.238555
Epoch[55] Batch [549]	Speed: 5367.266801 samples/sec	accuracy=0.467044	lr=0.238341
Epoch[55] Batch [599]	Speed: 5322.285221 samples/sec	accuracy=0.466854	lr=0.238126
Epoch[55] Batch [649]	Speed: 5435.210139 samples/sec	accuracy=0.466668	lr=0.237912
Epoch[55] Batch [699]	Speed: 5208.018066 samples/sec	accuracy=0.466383	lr=0.237698
Epoch[55] Batch [749]	Speed: 5451.915851 samples/sec	accuracy=0.465785	lr=0.237483
Epoch[55] Batch [799]	Speed: 4930.545812 samples/sec	accuracy=0.465825	lr=0.237269
Epoch[55] Batch [849]	Speed: 5303.051247 samples/sec	accuracy=0.465728	lr=0.237054
Epoch[55] Batch [899]	Speed: 4876.702284 samples/sec	accuracy=0.465598	lr=0.236839
Epoch[55] Batch [949]	Speed: 5080.033907 samples/sec	accuracy=0.465455	lr=0.236625
Epoch[55] Batch [999]	Speed: 5474.966727 samples/sec	accuracy=0.465511	lr=0.236410
Epoch[55] Batch [1049]	Speed: 6042.351730 samples/sec	accuracy=0.465375	lr=0.236195
Epoch[55] Batch [1099]	Speed: 5189.244295 samples/sec	accuracy=0.465361	lr=0.235981
Epoch[55] Batch [1149]	Speed: 4872.618124 samples/sec	accuracy=0.465491	lr=0.235766
Epoch[55] Batch [1199]	Speed: 5054.849472 samples/sec	accuracy=0.465478	lr=0.235551
Epoch[55] Batch [1249]	Speed: 5011.546593 samples/sec	accuracy=0.465192	lr=0.235336
[Epoch 55] training: accuracy=0.465190
[Epoch 55] speed: 5117 samples/sec	time cost: 260.456078
[Epoch 55] validation: err-top1=0.586555 err-top5=0.331872
Epoch[56] Batch [49]	Speed: 5090.039027 samples/sec	accuracy=0.474922	lr=0.235117
Epoch[56] Batch [99]	Speed: 5458.680595 samples/sec	accuracy=0.471934	lr=0.234902
Epoch[56] Batch [149]	Speed: 5449.842561 samples/sec	accuracy=0.470840	lr=0.234687
Epoch[56] Batch [199]	Speed: 5000.438570 samples/sec	accuracy=0.470786	lr=0.234472
Epoch[56] Batch [249]	Speed: 4444.403152 samples/sec	accuracy=0.470262	lr=0.234256
Epoch[56] Batch [299]	Speed: 5217.433547 samples/sec	accuracy=0.470684	lr=0.234041
Epoch[56] Batch [349]	Speed: 4958.344619 samples/sec	accuracy=0.470248	lr=0.233826
Epoch[56] Batch [399]	Speed: 5089.795575 samples/sec	accuracy=0.469729	lr=0.233611
Epoch[56] Batch [449]	Speed: 5182.265960 samples/sec	accuracy=0.469264	lr=0.233396
Epoch[56] Batch [499]	Speed: 2084.343445 samples/sec	accuracy=0.468773	lr=0.233180
Epoch[56] Batch [549]	Speed: 1250.691539 samples/sec	accuracy=0.468681	lr=0.232965
Epoch[56] Batch [599]	Speed: 1091.607276 samples/sec	accuracy=0.468307	lr=0.232749
Epoch[56] Batch [649]	Speed: 3164.828451 samples/sec	accuracy=0.468328	lr=0.232534
Epoch[56] Batch [699]	Speed: 1695.026750 samples/sec	accuracy=0.467937	lr=0.232319
Epoch[56] Batch [749]	Speed: 5458.082353 samples/sec	accuracy=0.467900	lr=0.232103
Epoch[56] Batch [799]	Speed: 5789.191095 samples/sec	accuracy=0.467924	lr=0.231887
Epoch[56] Batch [849]	Speed: 5844.153826 samples/sec	accuracy=0.467514	lr=0.231672
Epoch[56] Batch [899]	Speed: 1051.462346 samples/sec	accuracy=0.467373	lr=0.231456
Epoch[56] Batch [949]	Speed: 1693.059048 samples/sec	accuracy=0.467234	lr=0.231241
Epoch[56] Batch [999]	Speed: 5273.742855 samples/sec	accuracy=0.467063	lr=0.231025
Epoch[56] Batch [1049]	Speed: 4859.982659 samples/sec	accuracy=0.466976	lr=0.230809
Epoch[56] Batch [1099]	Speed: 1456.255523 samples/sec	accuracy=0.466936	lr=0.230593
Epoch[56] Batch [1149]	Speed: 4529.292413 samples/sec	accuracy=0.466890	lr=0.230377
Epoch[56] Batch [1199]	Speed: 5421.579140 samples/sec	accuracy=0.466797	lr=0.230162
Epoch[56] Batch [1249]	Speed: 4001.272010 samples/sec	accuracy=0.466684	lr=0.229946
[Epoch 56] training: accuracy=0.466683
[Epoch 56] speed: 2877 samples/sec	time cost: 461.213981
[Epoch 56] validation: err-top1=0.527025 err-top5=0.270986
Epoch[57] Batch [49]	Speed: 1787.098105 samples/sec	accuracy=0.473535	lr=0.229725
Epoch[57] Batch [99]	Speed: 2417.568910 samples/sec	accuracy=0.471562	lr=0.229510
Epoch[57] Batch [149]	Speed: 3183.976607 samples/sec	accuracy=0.471354	lr=0.229294
Epoch[57] Batch [199]	Speed: 2446.052818 samples/sec	accuracy=0.471338	lr=0.229077
Epoch[57] Batch [249]	Speed: 5570.038438 samples/sec	accuracy=0.471309	lr=0.228861
Epoch[57] Batch [299]	Speed: 2588.474519 samples/sec	accuracy=0.470700	lr=0.228645
Epoch[57] Batch [349]	Speed: 4345.236204 samples/sec	accuracy=0.470979	lr=0.228429
Epoch[57] Batch [399]	Speed: 5603.491052 samples/sec	accuracy=0.470818	lr=0.228213
Epoch[57] Batch [449]	Speed: 2617.583825 samples/sec	accuracy=0.470697	lr=0.227997
Epoch[57] Batch [499]	Speed: 2279.830184 samples/sec	accuracy=0.469980	lr=0.227781
Epoch[57] Batch [549]	Speed: 1854.967001 samples/sec	accuracy=0.470004	lr=0.227564
Epoch[57] Batch [599]	Speed: 4663.073792 samples/sec	accuracy=0.469787	lr=0.227348
Epoch[57] Batch [649]	Speed: 5418.129368 samples/sec	accuracy=0.469657	lr=0.227132
Epoch[57] Batch [699]	Speed: 3640.778584 samples/sec	accuracy=0.469757	lr=0.226915
Epoch[57] Batch [749]	Speed: 1632.779423 samples/sec	accuracy=0.469579	lr=0.226699
Epoch[57] Batch [799]	Speed: 5303.818099 samples/sec	accuracy=0.469282	lr=0.226482
Epoch[57] Batch [849]	Speed: 3819.051802 samples/sec	accuracy=0.469011	lr=0.226266
Epoch[57] Batch [899]	Speed: 5285.175989 samples/sec	accuracy=0.468912	lr=0.226049
Epoch[57] Batch [949]	Speed: 5382.412068 samples/sec	accuracy=0.468959	lr=0.225833
Epoch[57] Batch [999]	Speed: 3435.702938 samples/sec	accuracy=0.468732	lr=0.225616
Epoch[57] Batch [1049]	Speed: 3047.540688 samples/sec	accuracy=0.468693	lr=0.225400
Epoch[57] Batch [1099]	Speed: 2949.020486 samples/sec	accuracy=0.468691	lr=0.225183
Epoch[57] Batch [1149]	Speed: 1236.211801 samples/sec	accuracy=0.468538	lr=0.224967
Epoch[57] Batch [1199]	Speed: 5706.659341 samples/sec	accuracy=0.468598	lr=0.224750
Epoch[57] Batch [1249]	Speed: 5420.702055 samples/sec	accuracy=0.468359	lr=0.224533
[Epoch 57] training: accuracy=0.468386
[Epoch 57] speed: 3022 samples/sec	time cost: 433.695713
[Epoch 57] validation: err-top1=0.597557 err-top5=0.334064
Epoch[58] Batch [49]	Speed: 5392.443014 samples/sec	accuracy=0.474590	lr=0.224312
Epoch[58] Batch [99]	Speed: 818.540625 samples/sec	accuracy=0.473086	lr=0.224095
Epoch[58] Batch [149]	Speed: 2456.279435 samples/sec	accuracy=0.473053	lr=0.223879
Epoch[58] Batch [199]	Speed: 1802.045327 samples/sec	accuracy=0.472075	lr=0.223662
Epoch[58] Batch [249]	Speed: 1793.054164 samples/sec	accuracy=0.471555	lr=0.223445
Epoch[58] Batch [299]	Speed: 5694.563278 samples/sec	accuracy=0.471647	lr=0.223228
Epoch[58] Batch [349]	Speed: 5831.548005 samples/sec	accuracy=0.471883	lr=0.223011
Epoch[58] Batch [399]	Speed: 4906.292746 samples/sec	accuracy=0.472112	lr=0.222794
Epoch[58] Batch [449]	Speed: 4348.411007 samples/sec	accuracy=0.472094	lr=0.222577
Epoch[58] Batch [499]	Speed: 5008.876562 samples/sec	accuracy=0.472082	lr=0.222360
Epoch[58] Batch [549]	Speed: 4391.494183 samples/sec	accuracy=0.471618	lr=0.222143
Epoch[58] Batch [599]	Speed: 1617.413483 samples/sec	accuracy=0.471439	lr=0.221926
Epoch[58] Batch [649]	Speed: 3756.417200 samples/sec	accuracy=0.471188	lr=0.221709
Epoch[58] Batch [699]	Speed: 4725.761370 samples/sec	accuracy=0.470964	lr=0.221492
Epoch[58] Batch [749]	Speed: 5584.667224 samples/sec	accuracy=0.470926	lr=0.221275
Epoch[58] Batch [799]	Speed: 6277.079954 samples/sec	accuracy=0.470795	lr=0.221058
Epoch[58] Batch [849]	Speed: 3306.361300 samples/sec	accuracy=0.470727	lr=0.220841
Epoch[58] Batch [899]	Speed: 5428.127027 samples/sec	accuracy=0.470750	lr=0.220623
Epoch[58] Batch [949]	Speed: 1701.263283 samples/sec	accuracy=0.470610	lr=0.220406
Epoch[58] Batch [999]	Speed: 5763.467449 samples/sec	accuracy=0.470637	lr=0.220189
Epoch[58] Batch [1049]	Speed: 5185.933863 samples/sec	accuracy=0.470373	lr=0.219972
Epoch[58] Batch [1099]	Speed: 5013.717608 samples/sec	accuracy=0.470320	lr=0.219754
Epoch[58] Batch [1149]	Speed: 1902.529258 samples/sec	accuracy=0.470145	lr=0.219537
Epoch[58] Batch [1199]	Speed: 5011.590217 samples/sec	accuracy=0.469972	lr=0.219320
Epoch[58] Batch [1249]	Speed: 2485.440040 samples/sec	accuracy=0.469827	lr=0.219102
[Epoch 58] training: accuracy=0.469830
[Epoch 58] speed: 2996 samples/sec	time cost: 441.906562
[Epoch 58] validation: err-top1=0.534770 err-top5=0.279358
Epoch[59] Batch [49]	Speed: 5071.587623 samples/sec	accuracy=0.482187	lr=0.218881
Epoch[59] Batch [99]	Speed: 1659.005324 samples/sec	accuracy=0.479209	lr=0.218663
Epoch[59] Batch [149]	Speed: 2497.520775 samples/sec	accuracy=0.479499	lr=0.218446
Epoch[59] Batch [199]	Speed: 3483.257638 samples/sec	accuracy=0.478184	lr=0.218228
Epoch[59] Batch [249]	Speed: 711.591317 samples/sec	accuracy=0.477367	lr=0.218011
Epoch[59] Batch [299]	Speed: 1044.221901 samples/sec	accuracy=0.477087	lr=0.217793
Epoch[59] Batch [349]	Speed: 1656.319800 samples/sec	accuracy=0.476236	lr=0.217576
Epoch[59] Batch [399]	Speed: 2512.328312 samples/sec	accuracy=0.475210	lr=0.217358
Epoch[59] Batch [449]	Speed: 5525.971073 samples/sec	accuracy=0.474889	lr=0.217141
Epoch[59] Batch [499]	Speed: 4347.579794 samples/sec	accuracy=0.474676	lr=0.216923
Epoch[59] Batch [549]	Speed: 1766.697511 samples/sec	accuracy=0.474125	lr=0.216706
Epoch[59] Batch [599]	Speed: 961.781219 samples/sec	accuracy=0.473722	lr=0.216488
Epoch[59] Batch [649]	Speed: 5540.432899 samples/sec	accuracy=0.473625	lr=0.216270
Epoch[59] Batch [699]	Speed: 4720.512532 samples/sec	accuracy=0.473273	lr=0.216053
Epoch[59] Batch [749]	Speed: 2467.581164 samples/sec	accuracy=0.473424	lr=0.215835
Epoch[59] Batch [799]	Speed: 4254.229728 samples/sec	accuracy=0.473186	lr=0.215617
Epoch[59] Batch [849]	Speed: 3664.200406 samples/sec	accuracy=0.473055	lr=0.215400
Epoch[59] Batch [899]	Speed: 3896.511651 samples/sec	accuracy=0.472988	lr=0.215182
Epoch[59] Batch [949]	Speed: 3626.239490 samples/sec	accuracy=0.472860	lr=0.214964
Epoch[59] Batch [999]	Speed: 2306.510984 samples/sec	accuracy=0.472600	lr=0.214746
Epoch[59] Batch [1049]	Speed: 4991.789804 samples/sec	accuracy=0.472490	lr=0.214529
Epoch[59] Batch [1099]	Speed: 1609.108714 samples/sec	accuracy=0.472553	lr=0.214311
Epoch[59] Batch [1149]	Speed: 4990.096293 samples/sec	accuracy=0.472621	lr=0.214093
Epoch[59] Batch [1199]	Speed: 1097.533211 samples/sec	accuracy=0.472563	lr=0.213875
Epoch[59] Batch [1249]	Speed: 1426.450656 samples/sec	accuracy=0.472371	lr=0.213657
[Epoch 59] training: accuracy=0.472352
[Epoch 59] speed: 2122 samples/sec	time cost: 613.620860
[Epoch 59] validation: err-top1=0.534718 err-top5=0.275670
Epoch[60] Batch [49]	Speed: 2742.847252 samples/sec	accuracy=0.479434	lr=0.213435
Epoch[60] Batch [99]	Speed: 5794.622102 samples/sec	accuracy=0.480029	lr=0.213217
Epoch[60] Batch [149]	Speed: 5964.593647 samples/sec	accuracy=0.478145	lr=0.212999
Epoch[60] Batch [199]	Speed: 1048.097437 samples/sec	accuracy=0.477476	lr=0.212781
Epoch[60] Batch [249]	Speed: 5838.836626 samples/sec	accuracy=0.476883	lr=0.212563
Epoch[60] Batch [299]	Speed: 1385.948746 samples/sec	accuracy=0.476380	lr=0.212345
Epoch[60] Batch [349]	Speed: 1070.074587 samples/sec	accuracy=0.476105	lr=0.212127
Epoch[60] Batch [399]	Speed: 2163.972927 samples/sec	accuracy=0.476270	lr=0.211909
Epoch[60] Batch [449]	Speed: 1064.390288 samples/sec	accuracy=0.475658	lr=0.211691
Epoch[60] Batch [499]	Speed: 1866.919544 samples/sec	accuracy=0.475041	lr=0.211473
Epoch[60] Batch [549]	Speed: 5475.211427 samples/sec	accuracy=0.475138	lr=0.211255
Epoch[60] Batch [599]	Speed: 2405.488476 samples/sec	accuracy=0.474868	lr=0.211037
Epoch[60] Batch [649]	Speed: 5495.605396 samples/sec	accuracy=0.474581	lr=0.210819
Epoch[60] Batch [699]	Speed: 5629.642486 samples/sec	accuracy=0.474213	lr=0.210601
Epoch[60] Batch [749]	Speed: 5562.452973 samples/sec	accuracy=0.473961	lr=0.210383
Epoch[60] Batch [799]	Speed: 5548.222984 samples/sec	accuracy=0.473771	lr=0.210165
Epoch[60] Batch [849]	Speed: 2725.795268 samples/sec	accuracy=0.473749	lr=0.209947
Epoch[60] Batch [899]	Speed: 1101.772396 samples/sec	accuracy=0.473604	lr=0.209729
Epoch[60] Batch [949]	Speed: 5466.173870 samples/sec	accuracy=0.473598	lr=0.209511
Epoch[60] Batch [999]	Speed: 5766.874926 samples/sec	accuracy=0.473461	lr=0.209293
Epoch[60] Batch [1049]	Speed: 1262.734850 samples/sec	accuracy=0.473506	lr=0.209075
Epoch[60] Batch [1099]	Speed: 1146.093192 samples/sec	accuracy=0.473414	lr=0.208856
Epoch[60] Batch [1149]	Speed: 748.530209 samples/sec	accuracy=0.473161	lr=0.208638
Epoch[60] Batch [1199]	Speed: 2783.480927 samples/sec	accuracy=0.473211	lr=0.208420
Epoch[60] Batch [1249]	Speed: 5346.317672 samples/sec	accuracy=0.473030	lr=0.208202
[Epoch 60] training: accuracy=0.473033
[Epoch 60] speed: 2096 samples/sec	time cost: 637.999379
[Epoch 60] validation: err-top1=0.544603 err-top5=0.285415
Epoch[61] Batch [49]	Speed: 5437.256889 samples/sec	accuracy=0.480469	lr=0.207979
Epoch[61] Batch [99]	Speed: 5874.682930 samples/sec	accuracy=0.478633	lr=0.207761
Epoch[61] Batch [149]	Speed: 3455.886378 samples/sec	accuracy=0.479707	lr=0.207543
Epoch[61] Batch [199]	Speed: 5227.357963 samples/sec	accuracy=0.479033	lr=0.207325
Epoch[61] Batch [249]	Speed: 5951.078872 samples/sec	accuracy=0.478070	lr=0.207106
Epoch[61] Batch [299]	Speed: 1966.271227 samples/sec	accuracy=0.477728	lr=0.206888
Epoch[61] Batch [349]	Speed: 5564.962821 samples/sec	accuracy=0.477377	lr=0.206670
Epoch[61] Batch [399]	Speed: 5705.857392 samples/sec	accuracy=0.477485	lr=0.206452
Epoch[61] Batch [449]	Speed: 1185.161696 samples/sec	accuracy=0.477125	lr=0.206233
Epoch[61] Batch [499]	Speed: 1944.252438 samples/sec	accuracy=0.477164	lr=0.206015
Epoch[61] Batch [549]	Speed: 1991.185188 samples/sec	accuracy=0.476752	lr=0.205797
Epoch[61] Batch [599]	Speed: 5454.029226 samples/sec	accuracy=0.476564	lr=0.205579
Epoch[61] Batch [649]	Speed: 1204.393902 samples/sec	accuracy=0.476406	lr=0.205360
Epoch[61] Batch [699]	Speed: 1988.232231 samples/sec	accuracy=0.475811	lr=0.205142
Epoch[61] Batch [749]	Speed: 1459.356403 samples/sec	accuracy=0.475747	lr=0.204924
Epoch[61] Batch [799]	Speed: 1940.714966 samples/sec	accuracy=0.475667	lr=0.204705
Epoch[61] Batch [849]	Speed: 1199.681985 samples/sec	accuracy=0.475460	lr=0.204487
Epoch[61] Batch [899]	Speed: 5617.076440 samples/sec	accuracy=0.475438	lr=0.204269
Epoch[61] Batch [949]	Speed: 5717.732369 samples/sec	accuracy=0.475283	lr=0.204051
Epoch[61] Batch [999]	Speed: 810.809086 samples/sec	accuracy=0.475143	lr=0.203832
Epoch[61] Batch [1049]	Speed: 4691.539980 samples/sec	accuracy=0.474907	lr=0.203614
Epoch[61] Batch [1099]	Speed: 4678.433915 samples/sec	accuracy=0.474858	lr=0.203396
Epoch[61] Batch [1149]	Speed: 5506.196930 samples/sec	accuracy=0.474701	lr=0.203177
Epoch[61] Batch [1199]	Speed: 596.910577 samples/sec	accuracy=0.474577	lr=0.202959
Epoch[61] Batch [1249]	Speed: 729.942690 samples/sec	accuracy=0.474623	lr=0.202740
[Epoch 61] training: accuracy=0.474606
[Epoch 61] speed: 1996 samples/sec	time cost: 676.155104
[Epoch 61] validation: err-top1=0.521763 err-top5=0.263712
Epoch[62] Batch [49]	Speed: 5157.266852 samples/sec	accuracy=0.481699	lr=0.202518
Epoch[62] Batch [99]	Speed: 375.800370 samples/sec	accuracy=0.481426	lr=0.202299
Epoch[62] Batch [149]	Speed: 873.698667 samples/sec	accuracy=0.480671	lr=0.202081
Epoch[62] Batch [199]	Speed: 507.248643 samples/sec	accuracy=0.479385	lr=0.201863
Epoch[62] Batch [249]	Speed: 2277.716653 samples/sec	accuracy=0.478648	lr=0.201644
Epoch[62] Batch [299]	Speed: 1957.048075 samples/sec	accuracy=0.478285	lr=0.201426
Epoch[62] Batch [349]	Speed: 1482.000437 samples/sec	accuracy=0.478686	lr=0.201208
Epoch[62] Batch [399]	Speed: 3931.218418 samples/sec	accuracy=0.478435	lr=0.200989
Epoch[62] Batch [449]	Speed: 1928.085437 samples/sec	accuracy=0.478210	lr=0.200771
Epoch[62] Batch [499]	Speed: 2222.219811 samples/sec	accuracy=0.478547	lr=0.200552
Epoch[62] Batch [549]	Speed: 2009.896617 samples/sec	accuracy=0.478217	lr=0.200334
Epoch[62] Batch [599]	Speed: 1561.548004 samples/sec	accuracy=0.478452	lr=0.200116
Epoch[62] Batch [649]	Speed: 5462.018975 samples/sec	accuracy=0.478359	lr=0.199897
Epoch[62] Batch [699]	Speed: 2387.126435 samples/sec	accuracy=0.478235	lr=0.199679
Epoch[62] Batch [749]	Speed: 797.604854 samples/sec	accuracy=0.478010	lr=0.199461
Epoch[62] Batch [799]	Speed: 934.334651 samples/sec	accuracy=0.477826	lr=0.199242
Epoch[62] Batch [849]	Speed: 5913.198834 samples/sec	accuracy=0.477543	lr=0.199024
Epoch[62] Batch [899]	Speed: 1231.759764 samples/sec	accuracy=0.477490	lr=0.198806
Epoch[62] Batch [949]	Speed: 1234.832241 samples/sec	accuracy=0.477488	lr=0.198587
Epoch[62] Batch [999]	Speed: 5003.303366 samples/sec	accuracy=0.477551	lr=0.198369
Epoch[62] Batch [1049]	Speed: 5819.105563 samples/sec	accuracy=0.477441	lr=0.198150
Epoch[62] Batch [1099]	Speed: 713.486465 samples/sec	accuracy=0.477329	lr=0.197932
Epoch[62] Batch [1149]	Speed: 1671.993585 samples/sec	accuracy=0.477376	lr=0.197714
Epoch[62] Batch [1199]	Speed: 5528.484250 samples/sec	accuracy=0.477366	lr=0.197495
Epoch[62] Batch [1249]	Speed: 5543.268588 samples/sec	accuracy=0.477259	lr=0.197277
[Epoch 62] training: accuracy=0.477230
[Epoch 62] speed: 1435 samples/sec	time cost: 940.496566
[Epoch 62] validation: err-top1=0.666095 err-top5=0.421357
Epoch[63] Batch [49]	Speed: 5538.451584 samples/sec	accuracy=0.484082	lr=0.197054
Epoch[63] Batch [99]	Speed: 1180.016979 samples/sec	accuracy=0.483477	lr=0.196836
Epoch[63] Batch [149]	Speed: 2579.001572 samples/sec	accuracy=0.482520	lr=0.196618
Epoch[63] Batch [199]	Speed: 5599.458610 samples/sec	accuracy=0.480518	lr=0.196399
Epoch[63] Batch [249]	Speed: 1866.749873 samples/sec	accuracy=0.480887	lr=0.196181
Epoch[63] Batch [299]	Speed: 3153.015932 samples/sec	accuracy=0.479857	lr=0.195963
Epoch[63] Batch [349]	Speed: 4319.895686 samples/sec	accuracy=0.479551	lr=0.195744
Epoch[63] Batch [399]	Speed: 3849.686873 samples/sec	accuracy=0.479773	lr=0.195526
Epoch[63] Batch [449]	Speed: 1307.402506 samples/sec	accuracy=0.479338	lr=0.195308
Epoch[63] Batch [499]	Speed: 2345.862290 samples/sec	accuracy=0.479531	lr=0.195089
Epoch[63] Batch [549]	Speed: 1792.678031 samples/sec	accuracy=0.479137	lr=0.194871
Epoch[63] Batch [599]	Speed: 3923.336468 samples/sec	accuracy=0.479142	lr=0.194653
Epoch[63] Batch [649]	Speed: 4113.081801 samples/sec	accuracy=0.479037	lr=0.194434
Epoch[63] Batch [699]	Speed: 2360.184748 samples/sec	accuracy=0.478901	lr=0.194216
Epoch[63] Batch [749]	Speed: 6016.111951 samples/sec	accuracy=0.478810	lr=0.193998
Epoch[63] Batch [799]	Speed: 5411.562471 samples/sec	accuracy=0.478719	lr=0.193780
Epoch[63] Batch [849]	Speed: 5808.760955 samples/sec	accuracy=0.478702	lr=0.193561
Epoch[63] Batch [899]	Speed: 3807.786725 samples/sec	accuracy=0.478757	lr=0.193343
Epoch[63] Batch [949]	Speed: 3448.606164 samples/sec	accuracy=0.478845	lr=0.193125
Epoch[63] Batch [999]	Speed: 3490.983016 samples/sec	accuracy=0.478687	lr=0.192907
Epoch[63] Batch [1049]	Speed: 5575.334791 samples/sec	accuracy=0.478493	lr=0.192688
Epoch[63] Batch [1099]	Speed: 5260.929252 samples/sec	accuracy=0.478351	lr=0.192470
Epoch[63] Batch [1149]	Speed: 3271.438836 samples/sec	accuracy=0.478376	lr=0.192252
Epoch[63] Batch [1199]	Speed: 4152.977494 samples/sec	accuracy=0.478394	lr=0.192034
Epoch[63] Batch [1249]	Speed: 1346.601886 samples/sec	accuracy=0.478285	lr=0.191816
[Epoch 63] training: accuracy=0.478274
[Epoch 63] speed: 2892 samples/sec	time cost: 454.156168
[Epoch 63] validation: err-top1=0.548910 err-top5=0.288818
Epoch[64] Batch [49]	Speed: 3522.115714 samples/sec	accuracy=0.485918	lr=0.191593
Epoch[64] Batch [99]	Speed: 3175.773046 samples/sec	accuracy=0.483018	lr=0.191375
Epoch[64] Batch [149]	Speed: 2917.511424 samples/sec	accuracy=0.483509	lr=0.191157
Epoch[64] Batch [199]	Speed: 4543.981672 samples/sec	accuracy=0.484512	lr=0.190939
Epoch[64] Batch [249]	Speed: 2641.471297 samples/sec	accuracy=0.483574	lr=0.190720
Epoch[64] Batch [299]	Speed: 2949.225497 samples/sec	accuracy=0.482686	lr=0.190502
Epoch[64] Batch [349]	Speed: 1912.027755 samples/sec	accuracy=0.482185	lr=0.190284
Epoch[64] Batch [399]	Speed: 4220.850678 samples/sec	accuracy=0.482065	lr=0.190066
Epoch[64] Batch [449]	Speed: 3396.544681 samples/sec	accuracy=0.481821	lr=0.189848
Epoch[64] Batch [499]	Speed: 2190.800305 samples/sec	accuracy=0.482182	lr=0.189630
Epoch[64] Batch [549]	Speed: 2372.158147 samples/sec	accuracy=0.482303	lr=0.189412
Epoch[64] Batch [599]	Speed: 3651.800008 samples/sec	accuracy=0.481940	lr=0.189194
Epoch[64] Batch [649]	Speed: 4868.223272 samples/sec	accuracy=0.481858	lr=0.188976
Epoch[64] Batch [699]	Speed: 3531.555412 samples/sec	accuracy=0.481762	lr=0.188758
Epoch[64] Batch [749]	Speed: 4452.319479 samples/sec	accuracy=0.481405	lr=0.188540
Epoch[64] Batch [799]	Speed: 1888.000287 samples/sec	accuracy=0.481008	lr=0.188322
Epoch[64] Batch [849]	Speed: 4241.502897 samples/sec	accuracy=0.481012	lr=0.188104
Epoch[64] Batch [899]	Speed: 2231.430649 samples/sec	accuracy=0.480928	lr=0.187886
Epoch[64] Batch [949]	Speed: 3797.615808 samples/sec	accuracy=0.480991	lr=0.187668
Epoch[64] Batch [999]	Speed: 2673.398638 samples/sec	accuracy=0.481005	lr=0.187450
Epoch[64] Batch [1049]	Speed: 2022.678327 samples/sec	accuracy=0.481031	lr=0.187232
Epoch[64] Batch [1099]	Speed: 3167.089368 samples/sec	accuracy=0.481065	lr=0.187014
Epoch[64] Batch [1149]	Speed: 2563.070420 samples/sec	accuracy=0.480883	lr=0.186796
Epoch[64] Batch [1199]	Speed: 3370.005362 samples/sec	accuracy=0.480707	lr=0.186578
Epoch[64] Batch [1249]	Speed: 1775.354605 samples/sec	accuracy=0.480715	lr=0.186360
[Epoch 64] training: accuracy=0.480723
[Epoch 64] speed: 2869 samples/sec	time cost: 457.635661
[Epoch 64] validation: err-top1=0.505879 err-top5=0.250837
Epoch[65] Batch [49]	Speed: 4428.317577 samples/sec	accuracy=0.487559	lr=0.186138
Epoch[65] Batch [99]	Speed: 4061.702042 samples/sec	accuracy=0.487510	lr=0.185920
Epoch[65] Batch [149]	Speed: 3457.133151 samples/sec	accuracy=0.486074	lr=0.185702
Epoch[65] Batch [199]	Speed: 1748.619600 samples/sec	accuracy=0.486016	lr=0.185485
Epoch[65] Batch [249]	Speed: 3877.746161 samples/sec	accuracy=0.485547	lr=0.185267
Epoch[65] Batch [299]	Speed: 2297.467274 samples/sec	accuracy=0.485794	lr=0.185049
Epoch[65] Batch [349]	Speed: 2872.206717 samples/sec	accuracy=0.485299	lr=0.184831
Epoch[65] Batch [399]	Speed: 3991.712506 samples/sec	accuracy=0.484983	lr=0.184614
Epoch[65] Batch [449]	Speed: 2642.625781 samples/sec	accuracy=0.484603	lr=0.184396
Epoch[65] Batch [499]	Speed: 3470.016531 samples/sec	accuracy=0.484672	lr=0.184178
Epoch[65] Batch [549]	Speed: 3303.103918 samples/sec	accuracy=0.484268	lr=0.183960
Epoch[65] Batch [599]	Speed: 4777.572332 samples/sec	accuracy=0.483926	lr=0.183743
Epoch[65] Batch [649]	Speed: 5623.419223 samples/sec	accuracy=0.483733	lr=0.183525
Epoch[65] Batch [699]	Speed: 5593.589747 samples/sec	accuracy=0.483726	lr=0.183308
Epoch[65] Batch [749]	Speed: 5349.629533 samples/sec	accuracy=0.483484	lr=0.183090
Epoch[65] Batch [799]	Speed: 920.334213 samples/sec	accuracy=0.483583	lr=0.182872
Epoch[65] Batch [849]	Speed: 3930.019262 samples/sec	accuracy=0.483514	lr=0.182655
Epoch[65] Batch [899]	Speed: 5523.097791 samples/sec	accuracy=0.483216	lr=0.182437
Epoch[65] Batch [949]	Speed: 4107.030877 samples/sec	accuracy=0.482871	lr=0.182220
Epoch[65] Batch [999]	Speed: 4004.564227 samples/sec	accuracy=0.482848	lr=0.182002
Epoch[65] Batch [1049]	Speed: 3980.229753 samples/sec	accuracy=0.482677	lr=0.181785
Epoch[65] Batch [1099]	Speed: 4314.076931 samples/sec	accuracy=0.482663	lr=0.181567
Epoch[65] Batch [1149]	Speed: 4595.346140 samples/sec	accuracy=0.482745	lr=0.181350
Epoch[65] Batch [1199]	Speed: 2631.286027 samples/sec	accuracy=0.482708	lr=0.181132
Epoch[65] Batch [1249]	Speed: 4730.833146 samples/sec	accuracy=0.482775	lr=0.180915
[Epoch 65] training: accuracy=0.482786
[Epoch 65] speed: 3263 samples/sec	time cost: 402.231229
[Epoch 65] validation: err-top1=0.542331 err-top5=0.287986
Epoch[66] Batch [49]	Speed: 3398.837755 samples/sec	accuracy=0.489277	lr=0.180693
Epoch[66] Batch [99]	Speed: 3866.395246 samples/sec	accuracy=0.489941	lr=0.180476
Epoch[66] Batch [149]	Speed: 3830.693250 samples/sec	accuracy=0.490391	lr=0.180259
Epoch[66] Batch [199]	Speed: 4081.130039 samples/sec	accuracy=0.489043	lr=0.180041
Epoch[66] Batch [249]	Speed: 3040.916718 samples/sec	accuracy=0.488641	lr=0.179824
Epoch[66] Batch [299]	Speed: 5459.179322 samples/sec	accuracy=0.488509	lr=0.179607
Epoch[66] Batch [349]	Speed: 1772.878601 samples/sec	accuracy=0.487447	lr=0.179390
Epoch[66] Batch [399]	Speed: 3846.533842 samples/sec	accuracy=0.487085	lr=0.179172
Epoch[66] Batch [449]	Speed: 5130.558595 samples/sec	accuracy=0.487422	lr=0.178955
Epoch[66] Batch [499]	Speed: 4334.857609 samples/sec	accuracy=0.487229	lr=0.178738
Epoch[66] Batch [549]	Speed: 2750.842598 samples/sec	accuracy=0.487157	lr=0.178521
Epoch[66] Batch [599]	Speed: 5845.404168 samples/sec	accuracy=0.486709	lr=0.178304
Epoch[66] Batch [649]	Speed: 5907.337657 samples/sec	accuracy=0.486068	lr=0.178087
Epoch[66] Batch [699]	Speed: 3639.597865 samples/sec	accuracy=0.485968	lr=0.177870
Epoch[66] Batch [749]	Speed: 4731.197314 samples/sec	accuracy=0.485833	lr=0.177653
Epoch[66] Batch [799]	Speed: 4267.814351 samples/sec	accuracy=0.485521	lr=0.177436
Epoch[66] Batch [849]	Speed: 3420.014245 samples/sec	accuracy=0.485250	lr=0.177219
Epoch[66] Batch [899]	Speed: 5065.200651 samples/sec	accuracy=0.485175	lr=0.177002
Epoch[66] Batch [949]	Speed: 3708.941165 samples/sec	accuracy=0.484969	lr=0.176785
Epoch[66] Batch [999]	Speed: 4193.008508 samples/sec	accuracy=0.484882	lr=0.176568
Epoch[66] Batch [1049]	Speed: 1575.435075 samples/sec	accuracy=0.484862	lr=0.176351
Epoch[66] Batch [1099]	Speed: 4407.165262 samples/sec	accuracy=0.484916	lr=0.176134
Epoch[66] Batch [1149]	Speed: 4934.058983 samples/sec	accuracy=0.484864	lr=0.175918
Epoch[66] Batch [1199]	Speed: 2871.263975 samples/sec	accuracy=0.484859	lr=0.175701
Epoch[66] Batch [1249]	Speed: 3676.514410 samples/sec	accuracy=0.484742	lr=0.175484
[Epoch 66] training: accuracy=0.484762
[Epoch 66] speed: 3602 samples/sec	time cost: 365.709117
[Epoch 66] validation: err-top1=0.596919 err-top5=0.341598
Epoch[67] Batch [49]	Speed: 3427.685529 samples/sec	accuracy=0.489258	lr=0.175263
Epoch[67] Batch [99]	Speed: 2130.064745 samples/sec	accuracy=0.487783	lr=0.175046
Epoch[67] Batch [149]	Speed: 5648.407380 samples/sec	accuracy=0.488034	lr=0.174830
Epoch[67] Batch [199]	Speed: 3487.669979 samples/sec	accuracy=0.488418	lr=0.174613
Epoch[67] Batch [249]	Speed: 5114.872328 samples/sec	accuracy=0.488320	lr=0.174397
Epoch[67] Batch [299]	Speed: 3978.826457 samples/sec	accuracy=0.487881	lr=0.174180
Epoch[67] Batch [349]	Speed: 4021.519795 samples/sec	accuracy=0.487974	lr=0.173964
Epoch[67] Batch [399]	Speed: 3115.626876 samples/sec	accuracy=0.487769	lr=0.173747
Epoch[67] Batch [449]	Speed: 3632.822110 samples/sec	accuracy=0.487231	lr=0.173531
Epoch[67] Batch [499]	Speed: 3967.392987 samples/sec	accuracy=0.486971	lr=0.173314
Epoch[67] Batch [549]	Speed: 2604.943856 samples/sec	accuracy=0.486967	lr=0.173098
Epoch[67] Batch [599]	Speed: 2780.387259 samples/sec	accuracy=0.486707	lr=0.172881
Epoch[67] Batch [649]	Speed: 4750.918766 samples/sec	accuracy=0.486499	lr=0.172665
Epoch[67] Batch [699]	Speed: 4792.638843 samples/sec	accuracy=0.486429	lr=0.172449
Epoch[67] Batch [749]	Speed: 5266.367149 samples/sec	accuracy=0.486255	lr=0.172232
Epoch[67] Batch [799]	Speed: 5075.133338 samples/sec	accuracy=0.486256	lr=0.172016
Epoch[67] Batch [849]	Speed: 2603.975437 samples/sec	accuracy=0.485857	lr=0.171800
Epoch[67] Batch [899]	Speed: 2508.693578 samples/sec	accuracy=0.485815	lr=0.171584
Epoch[67] Batch [949]	Speed: 3432.219922 samples/sec	accuracy=0.485940	lr=0.171368
Epoch[67] Batch [999]	Speed: 2475.623223 samples/sec	accuracy=0.485861	lr=0.171152
Epoch[67] Batch [1049]	Speed: 5251.090855 samples/sec	accuracy=0.486079	lr=0.170935
Epoch[67] Batch [1099]	Speed: 4394.018491 samples/sec	accuracy=0.485912	lr=0.170719
Epoch[67] Batch [1149]	Speed: 3991.917301 samples/sec	accuracy=0.485874	lr=0.170503
Epoch[67] Batch [1199]	Speed: 2315.227856 samples/sec	accuracy=0.485930	lr=0.170287
Epoch[67] Batch [1249]	Speed: 4505.961749 samples/sec	accuracy=0.485773	lr=0.170072
[Epoch 67] training: accuracy=0.485797
[Epoch 67] speed: 3508 samples/sec	time cost: 377.459511
[Epoch 67] validation: err-top1=0.601144 err-top5=0.345843
Epoch[68] Batch [49]	Speed: 4197.525027 samples/sec	accuracy=0.494004	lr=0.169851
Epoch[68] Batch [99]	Speed: 1624.751564 samples/sec	accuracy=0.493437	lr=0.169635
Epoch[68] Batch [149]	Speed: 5584.712392 samples/sec	accuracy=0.492493	lr=0.169420
Epoch[68] Batch [199]	Speed: 5463.380628 samples/sec	accuracy=0.491704	lr=0.169204
Epoch[68] Batch [249]	Speed: 5566.591425 samples/sec	accuracy=0.491102	lr=0.168988
Epoch[68] Batch [299]	Speed: 5332.769454 samples/sec	accuracy=0.491585	lr=0.168772
Epoch[68] Batch [349]	Speed: 5920.516667 samples/sec	accuracy=0.491353	lr=0.168557
Epoch[68] Batch [399]	Speed: 2481.512617 samples/sec	accuracy=0.490640	lr=0.168341
Epoch[68] Batch [449]	Speed: 5183.328291 samples/sec	accuracy=0.490386	lr=0.168126
Epoch[68] Batch [499]	Speed: 3276.344313 samples/sec	accuracy=0.490176	lr=0.167910
Epoch[68] Batch [549]	Speed: 2534.678320 samples/sec	accuracy=0.490099	lr=0.167694
Epoch[68] Batch [599]	Speed: 5100.712701 samples/sec	accuracy=0.489821	lr=0.167479
Epoch[68] Batch [649]	Speed: 5398.593982 samples/sec	accuracy=0.489695	lr=0.167263
Epoch[68] Batch [699]	Speed: 6064.200957 samples/sec	accuracy=0.489478	lr=0.167048
Epoch[68] Batch [749]	Speed: 5679.710470 samples/sec	accuracy=0.489393	lr=0.166833
Epoch[68] Batch [799]	Speed: 1813.710664 samples/sec	accuracy=0.489434	lr=0.166617
Epoch[68] Batch [849]	Speed: 3491.408011 samples/sec	accuracy=0.489318	lr=0.166402
Epoch[68] Batch [899]	Speed: 2661.744537 samples/sec	accuracy=0.489280	lr=0.166187
Epoch[68] Batch [949]	Speed: 1666.216156 samples/sec	accuracy=0.489195	lr=0.165972
Epoch[68] Batch [999]	Speed: 1720.259698 samples/sec	accuracy=0.488930	lr=0.165756
Epoch[68] Batch [1049]	Speed: 646.454412 samples/sec	accuracy=0.488841	lr=0.165541
Epoch[68] Batch [1099]	Speed: 2034.328576 samples/sec	accuracy=0.488841	lr=0.165326
Epoch[68] Batch [1149]	Speed: 5183.472545 samples/sec	accuracy=0.488671	lr=0.165111
Epoch[68] Batch [1199]	Speed: 2462.410231 samples/sec	accuracy=0.488525	lr=0.164896
Epoch[68] Batch [1249]	Speed: 731.230699 samples/sec	accuracy=0.488273	lr=0.164681
[Epoch 68] training: accuracy=0.488283
[Epoch 68] speed: 2456 samples/sec	time cost: 530.841383
[Epoch 68] validation: err-top1=0.670898 err-top5=0.422373
Epoch[69] Batch [49]	Speed: 2593.815955 samples/sec	accuracy=0.493887	lr=0.164462
Epoch[69] Batch [99]	Speed: 5731.643064 samples/sec	accuracy=0.492598	lr=0.164247
Epoch[69] Batch [149]	Speed: 5641.074586 samples/sec	accuracy=0.493086	lr=0.164032
Epoch[69] Batch [199]	Speed: 5728.859150 samples/sec	accuracy=0.492681	lr=0.163818
Epoch[69] Batch [249]	Speed: 3584.592534 samples/sec	accuracy=0.492840	lr=0.163603
Epoch[69] Batch [299]	Speed: 2644.726051 samples/sec	accuracy=0.492881	lr=0.163388
Epoch[69] Batch [349]	Speed: 2078.696021 samples/sec	accuracy=0.492441	lr=0.163173
Epoch[69] Batch [399]	Speed: 2163.705184 samples/sec	accuracy=0.491963	lr=0.162959
Epoch[69] Batch [449]	Speed: 5593.680809 samples/sec	accuracy=0.491638	lr=0.162744
Epoch[69] Batch [499]	Speed: 1503.447603 samples/sec	accuracy=0.491711	lr=0.162530
Epoch[69] Batch [549]	Speed: 5750.825295 samples/sec	accuracy=0.490977	lr=0.162315
Epoch[69] Batch [599]	Speed: 5902.350679 samples/sec	accuracy=0.491219	lr=0.162101
Epoch[69] Batch [649]	Speed: 2566.272994 samples/sec	accuracy=0.491017	lr=0.161886
Epoch[69] Batch [699]	Speed: 5332.084233 samples/sec	accuracy=0.490829	lr=0.161672
Epoch[69] Batch [749]	Speed: 2638.808008 samples/sec	accuracy=0.490569	lr=0.161458
Epoch[69] Batch [799]	Speed: 1272.679830 samples/sec	accuracy=0.490569	lr=0.161244
Epoch[69] Batch [849]	Speed: 2530.405034 samples/sec	accuracy=0.490395	lr=0.161029
Epoch[69] Batch [899]	Speed: 3358.043585 samples/sec	accuracy=0.490397	lr=0.160815
Epoch[69] Batch [949]	Speed: 1370.901829 samples/sec	accuracy=0.490298	lr=0.160601
Epoch[69] Batch [999]	Speed: 1245.023527 samples/sec	accuracy=0.490083	lr=0.160387
Epoch[69] Batch [1049]	Speed: 2599.298275 samples/sec	accuracy=0.490062	lr=0.160173
Epoch[69] Batch [1099]	Speed: 2480.018547 samples/sec	accuracy=0.489864	lr=0.159959
Epoch[69] Batch [1149]	Speed: 3307.098739 samples/sec	accuracy=0.489679	lr=0.159745
Epoch[69] Batch [1199]	Speed: 2518.071948 samples/sec	accuracy=0.489622	lr=0.159531
Epoch[69] Batch [1249]	Speed: 5511.854744 samples/sec	accuracy=0.489557	lr=0.159317
[Epoch 69] training: accuracy=0.489554
[Epoch 69] speed: 2676 samples/sec	time cost: 489.004678
[Epoch 69] validation: err-top1=0.498983 err-top5=0.244893
Epoch[70] Batch [49]	Speed: 2526.154572 samples/sec	accuracy=0.498398	lr=0.159099
Epoch[70] Batch [99]	Speed: 3951.775074 samples/sec	accuracy=0.497324	lr=0.158886
Epoch[70] Batch [149]	Speed: 5727.152708 samples/sec	accuracy=0.496725	lr=0.158672
Epoch[70] Batch [199]	Speed: 2538.878534 samples/sec	accuracy=0.496445	lr=0.158458
Epoch[70] Batch [249]	Speed: 1808.972764 samples/sec	accuracy=0.495680	lr=0.158245
Epoch[70] Batch [299]	Speed: 2767.835381 samples/sec	accuracy=0.495329	lr=0.158031
Epoch[70] Batch [349]	Speed: 5607.056982 samples/sec	accuracy=0.494944	lr=0.157818
Epoch[70] Batch [399]	Speed: 1118.894131 samples/sec	accuracy=0.494700	lr=0.157604
Epoch[70] Batch [449]	Speed: 2095.356153 samples/sec	accuracy=0.494036	lr=0.157391
Epoch[70] Batch [499]	Speed: 5631.961060 samples/sec	accuracy=0.493904	lr=0.157177
Epoch[70] Batch [549]	Speed: 1699.725220 samples/sec	accuracy=0.493601	lr=0.156964
Epoch[70] Batch [599]	Speed: 1801.155217 samples/sec	accuracy=0.493092	lr=0.156751
Epoch[70] Batch [649]	Speed: 2310.766076 samples/sec	accuracy=0.492679	lr=0.156538
Epoch[70] Batch [699]	Speed: 795.527916 samples/sec	accuracy=0.492607	lr=0.156325
Epoch[70] Batch [749]	Speed: 5822.536226 samples/sec	accuracy=0.492560	lr=0.156112
Epoch[70] Batch [799]	Speed: 5763.676431 samples/sec	accuracy=0.492378	lr=0.155899
Epoch[70] Batch [849]	Speed: 1672.482418 samples/sec	accuracy=0.492399	lr=0.155686
Epoch[70] Batch [899]	Speed: 2684.470865 samples/sec	accuracy=0.492372	lr=0.155473
Epoch[70] Batch [949]	Speed: 5471.824559 samples/sec	accuracy=0.492203	lr=0.155260
Epoch[70] Batch [999]	Speed: 2036.537890 samples/sec	accuracy=0.491999	lr=0.155047
Epoch[70] Batch [1049]	Speed: 1638.055922 samples/sec	accuracy=0.491917	lr=0.154834
Epoch[70] Batch [1099]	Speed: 1688.449255 samples/sec	accuracy=0.491890	lr=0.154622
Epoch[70] Batch [1149]	Speed: 2600.574264 samples/sec	accuracy=0.491703	lr=0.154409
Epoch[70] Batch [1199]	Speed: 4052.826706 samples/sec	accuracy=0.491760	lr=0.154196
Epoch[70] Batch [1249]	Speed: 4433.874739 samples/sec	accuracy=0.491788	lr=0.153984
[Epoch 70] training: accuracy=0.491780
[Epoch 70] speed: 2326 samples/sec	time cost: 560.310803
[Epoch 70] validation: err-top1=0.497828 err-top5=0.244041
Epoch[71] Batch [49]	Speed: 2257.851493 samples/sec	accuracy=0.496230	lr=0.153767
Epoch[71] Batch [99]	Speed: 1531.361148 samples/sec	accuracy=0.496709	lr=0.153555
Epoch[71] Batch [149]	Speed: 6051.038945 samples/sec	accuracy=0.496940	lr=0.153342
Epoch[71] Batch [199]	Speed: 1737.653247 samples/sec	accuracy=0.497095	lr=0.153130
Epoch[71] Batch [249]	Speed: 1681.283425 samples/sec	accuracy=0.497027	lr=0.152918
Epoch[71] Batch [299]	Speed: 4298.860172 samples/sec	accuracy=0.496735	lr=0.152705
Epoch[71] Batch [349]	Speed: 3208.713255 samples/sec	accuracy=0.496261	lr=0.152493
Epoch[71] Batch [399]	Speed: 5884.314209 samples/sec	accuracy=0.495986	lr=0.152281
Epoch[71] Batch [449]	Speed: 5841.360624 samples/sec	accuracy=0.496059	lr=0.152069
Epoch[71] Batch [499]	Speed: 5700.526810 samples/sec	accuracy=0.495887	lr=0.151857
Epoch[71] Batch [549]	Speed: 2546.311925 samples/sec	accuracy=0.495918	lr=0.151645
Epoch[71] Batch [599]	Speed: 5742.061090 samples/sec	accuracy=0.495451	lr=0.151433
Epoch[71] Batch [649]	Speed: 5741.212477 samples/sec	accuracy=0.495120	lr=0.151222
Epoch[71] Batch [699]	Speed: 5665.595682 samples/sec	accuracy=0.495117	lr=0.151010
Epoch[71] Batch [749]	Speed: 3695.883630 samples/sec	accuracy=0.495181	lr=0.150798
Epoch[71] Batch [799]	Speed: 2292.436836 samples/sec	accuracy=0.495142	lr=0.150587
Epoch[71] Batch [849]	Speed: 5597.638688 samples/sec	accuracy=0.494859	lr=0.150375
Epoch[71] Batch [899]	Speed: 2690.812813 samples/sec	accuracy=0.494425	lr=0.150163
Epoch[71] Batch [949]	Speed: 2011.178932 samples/sec	accuracy=0.494369	lr=0.149952
Epoch[71] Batch [999]	Speed: 5479.752328 samples/sec	accuracy=0.494252	lr=0.149741
Epoch[71] Batch [1049]	Speed: 2573.868920 samples/sec	accuracy=0.494395	lr=0.149529
Epoch[71] Batch [1099]	Speed: 5923.731188 samples/sec	accuracy=0.494196	lr=0.149318
Epoch[71] Batch [1149]	Speed: 2522.226384 samples/sec	accuracy=0.494063	lr=0.149107
Epoch[71] Batch [1199]	Speed: 2322.230946 samples/sec	accuracy=0.494060	lr=0.148896
Epoch[71] Batch [1249]	Speed: 3484.120706 samples/sec	accuracy=0.494065	lr=0.148685
[Epoch 71] training: accuracy=0.494089
[Epoch 71] speed: 3038 samples/sec	time cost: 455.301365
[Epoch 71] validation: err-top1=0.485352 err-top5=0.234435
Epoch[72] Batch [49]	Speed: 1671.285558 samples/sec	accuracy=0.502207	lr=0.148469
Epoch[72] Batch [99]	Speed: 5489.133855 samples/sec	accuracy=0.502441	lr=0.148258
Epoch[72] Batch [149]	Speed: 1658.164367 samples/sec	accuracy=0.501016	lr=0.148047
Epoch[72] Batch [199]	Speed: 1697.406194 samples/sec	accuracy=0.500244	lr=0.147837
Epoch[72] Batch [249]	Speed: 2616.358860 samples/sec	accuracy=0.499887	lr=0.147626
Epoch[72] Batch [299]	Speed: 5782.303872 samples/sec	accuracy=0.500098	lr=0.147415
Epoch[72] Batch [349]	Speed: 2559.930117 samples/sec	accuracy=0.499593	lr=0.147204
Epoch[72] Batch [399]	Speed: 2610.913607 samples/sec	accuracy=0.499766	lr=0.146994
Epoch[72] Batch [449]	Speed: 2512.366757 samples/sec	accuracy=0.499334	lr=0.146783
Epoch[72] Batch [499]	Speed: 2073.833244 samples/sec	accuracy=0.499322	lr=0.146573
Epoch[72] Batch [549]	Speed: 5264.210463 samples/sec	accuracy=0.498841	lr=0.146362
Epoch[72] Batch [599]	Speed: 3372.123091 samples/sec	accuracy=0.498359	lr=0.146152
Epoch[72] Batch [649]	Speed: 1432.636931 samples/sec	accuracy=0.498579	lr=0.145942
Epoch[72] Batch [699]	Speed: 5621.955149 samples/sec	accuracy=0.498334	lr=0.145732
Epoch[72] Batch [749]	Speed: 5327.187027 samples/sec	accuracy=0.497733	lr=0.145521
Epoch[72] Batch [799]	Speed: 5791.948830 samples/sec	accuracy=0.497770	lr=0.145311
Epoch[72] Batch [849]	Speed: 2587.920958 samples/sec	accuracy=0.497471	lr=0.145101
Epoch[72] Batch [899]	Speed: 3325.535825 samples/sec	accuracy=0.497314	lr=0.144891
Epoch[72] Batch [949]	Speed: 5891.602705 samples/sec	accuracy=0.497288	lr=0.144682
Epoch[72] Batch [999]	Speed: 1914.933851 samples/sec	accuracy=0.497069	lr=0.144472
Epoch[72] Batch [1049]	Speed: 5718.347926 samples/sec	accuracy=0.497081	lr=0.144262
Epoch[72] Batch [1099]	Speed: 5744.913948 samples/sec	accuracy=0.497067	lr=0.144052
Epoch[72] Batch [1149]	Speed: 1731.290752 samples/sec	accuracy=0.496916	lr=0.143843
Epoch[72] Batch [1199]	Speed: 1245.949262 samples/sec	accuracy=0.496775	lr=0.143633
Epoch[72] Batch [1249]	Speed: 5856.580425 samples/sec	accuracy=0.496734	lr=0.143424
[Epoch 72] training: accuracy=0.496733
[Epoch 72] speed: 2720 samples/sec	time cost: 510.126560
[Epoch 72] validation: err-top1=0.521106 err-top5=0.272481
Epoch[73] Batch [49]	Speed: 967.563034 samples/sec	accuracy=0.501016	lr=0.143210
Epoch[73] Batch [99]	Speed: 5761.445236 samples/sec	accuracy=0.501211	lr=0.143001
Epoch[73] Batch [149]	Speed: 3790.741845 samples/sec	accuracy=0.501374	lr=0.142791
Epoch[73] Batch [199]	Speed: 4887.790271 samples/sec	accuracy=0.500508	lr=0.142582
Epoch[73] Batch [249]	Speed: 2223.276638 samples/sec	accuracy=0.500371	lr=0.142373
Epoch[73] Batch [299]	Speed: 5450.767287 samples/sec	accuracy=0.499808	lr=0.142164
Epoch[73] Batch [349]	Speed: 5380.188559 samples/sec	accuracy=0.499849	lr=0.141955
Epoch[73] Batch [399]	Speed: 2665.724213 samples/sec	accuracy=0.499739	lr=0.141746
Epoch[73] Batch [449]	Speed: 5726.791353 samples/sec	accuracy=0.499212	lr=0.141537
Epoch[73] Batch [499]	Speed: 5656.101729 samples/sec	accuracy=0.499381	lr=0.141328
Epoch[73] Batch [549]	Speed: 2592.483679 samples/sec	accuracy=0.499134	lr=0.141120
Epoch[73] Batch [599]	Speed: 2420.094217 samples/sec	accuracy=0.499058	lr=0.140911
Epoch[73] Batch [649]	Speed: 5829.085641 samples/sec	accuracy=0.499258	lr=0.140702
Epoch[73] Batch [699]	Speed: 2727.478934 samples/sec	accuracy=0.499312	lr=0.140494
Epoch[73] Batch [749]	Speed: 2332.088180 samples/sec	accuracy=0.499284	lr=0.140285
Epoch[73] Batch [799]	Speed: 5592.794062 samples/sec	accuracy=0.499009	lr=0.140077
Epoch[73] Batch [849]	Speed: 1876.077189 samples/sec	accuracy=0.499023	lr=0.139869
Epoch[73] Batch [899]	Speed: 1868.514604 samples/sec	accuracy=0.499052	lr=0.139661
Epoch[73] Batch [949]	Speed: 3155.664491 samples/sec	accuracy=0.498992	lr=0.139452
Epoch[73] Batch [999]	Speed: 5073.108714 samples/sec	accuracy=0.498871	lr=0.139244
Epoch[73] Batch [1049]	Speed: 5362.480599 samples/sec	accuracy=0.498809	lr=0.139036
Epoch[73] Batch [1099]	Speed: 3647.363778 samples/sec	accuracy=0.498721	lr=0.138828
Epoch[73] Batch [1149]	Speed: 5743.052326 samples/sec	accuracy=0.498605	lr=0.138620
Epoch[73] Batch [1199]	Speed: 1724.840981 samples/sec	accuracy=0.498467	lr=0.138413
Epoch[73] Batch [1249]	Speed: 5444.033723 samples/sec	accuracy=0.498334	lr=0.138205
[Epoch 73] training: accuracy=0.498344
[Epoch 73] speed: 3066 samples/sec	time cost: 456.964034
[Epoch 73] validation: err-top1=0.493921 err-top5=0.244659
Epoch[74] Batch [49]	Speed: 5164.534292 samples/sec	accuracy=0.512578	lr=0.137993
Epoch[74] Batch [99]	Speed: 2700.439899 samples/sec	accuracy=0.507754	lr=0.137786
Epoch[74] Batch [149]	Speed: 5750.376410 samples/sec	accuracy=0.506133	lr=0.137578
Epoch[74] Batch [199]	Speed: 4631.944101 samples/sec	accuracy=0.505273	lr=0.137371
Epoch[74] Batch [249]	Speed: 5376.393196 samples/sec	accuracy=0.503633	lr=0.137163
Epoch[74] Batch [299]	Speed: 1300.631894 samples/sec	accuracy=0.502744	lr=0.136956
Epoch[74] Batch [349]	Speed: 2590.127596 samples/sec	accuracy=0.502302	lr=0.136749
Epoch[74] Batch [399]	Speed: 2549.942380 samples/sec	accuracy=0.501880	lr=0.136542
Epoch[74] Batch [449]	Speed: 2633.035674 samples/sec	accuracy=0.501363	lr=0.136335
Epoch[74] Batch [499]	Speed: 5547.362627 samples/sec	accuracy=0.501539	lr=0.136128
Epoch[74] Batch [549]	Speed: 2649.247200 samples/sec	accuracy=0.501396	lr=0.135921
Epoch[74] Batch [599]	Speed: 1768.933724 samples/sec	accuracy=0.501497	lr=0.135714
Epoch[74] Batch [649]	Speed: 5532.313317 samples/sec	accuracy=0.501465	lr=0.135507
Epoch[74] Batch [699]	Speed: 5489.996034 samples/sec	accuracy=0.501448	lr=0.135301
Epoch[74] Batch [749]	Speed: 2101.462995 samples/sec	accuracy=0.501052	lr=0.135094
Epoch[74] Batch [799]	Speed: 3997.009429 samples/sec	accuracy=0.501097	lr=0.134887
Epoch[74] Batch [849]	Speed: 3922.942068 samples/sec	accuracy=0.500819	lr=0.134681
Epoch[74] Batch [899]	Speed: 2521.331712 samples/sec	accuracy=0.500574	lr=0.134475
Epoch[74] Batch [949]	Speed: 2714.072466 samples/sec	accuracy=0.500453	lr=0.134268
Epoch[74] Batch [999]	Speed: 4003.516499 samples/sec	accuracy=0.500347	lr=0.134062
Epoch[74] Batch [1049]	Speed: 4772.778876 samples/sec	accuracy=0.500338	lr=0.133856
Epoch[74] Batch [1099]	Speed: 2117.256216 samples/sec	accuracy=0.500302	lr=0.133650
Epoch[74] Batch [1149]	Speed: 4216.086924 samples/sec	accuracy=0.500127	lr=0.133444
Epoch[74] Batch [1199]	Speed: 1977.291814 samples/sec	accuracy=0.500052	lr=0.133238
Epoch[74] Batch [1249]	Speed: 5531.788456 samples/sec	accuracy=0.500069	lr=0.133032
[Epoch 74] training: accuracy=0.500069
[Epoch 74] speed: 3064 samples/sec	time cost: 427.754345
[Epoch 74] validation: err-top1=0.494559 err-top5=0.244440
Epoch[75] Batch [49]	Speed: 2771.986871 samples/sec	accuracy=0.508379	lr=0.132822
Epoch[75] Batch [99]	Speed: 2866.082115 samples/sec	accuracy=0.506621	lr=0.132617
Epoch[75] Batch [149]	Speed: 5823.906370 samples/sec	accuracy=0.506855	lr=0.132411
Epoch[75] Batch [199]	Speed: 5776.985278 samples/sec	accuracy=0.506699	lr=0.132206
Epoch[75] Batch [249]	Speed: 1878.317254 samples/sec	accuracy=0.506168	lr=0.132000
Epoch[75] Batch [299]	Speed: 5999.326696 samples/sec	accuracy=0.506146	lr=0.131795
Epoch[75] Batch [349]	Speed: 5360.231646 samples/sec	accuracy=0.505636	lr=0.131590
Epoch[75] Batch [399]	Speed: 3430.871938 samples/sec	accuracy=0.505601	lr=0.131385
Epoch[75] Batch [449]	Speed: 2200.220401 samples/sec	accuracy=0.505436	lr=0.131180
Epoch[75] Batch [499]	Speed: 2299.730006 samples/sec	accuracy=0.504834	lr=0.130975
Epoch[75] Batch [549]	Speed: 3868.195478 samples/sec	accuracy=0.504721	lr=0.130770
Epoch[75] Batch [599]	Speed: 5895.410390 samples/sec	accuracy=0.504412	lr=0.130565
Epoch[75] Batch [649]	Speed: 2076.634125 samples/sec	accuracy=0.504223	lr=0.130360
Epoch[75] Batch [699]	Speed: 4885.373585 samples/sec	accuracy=0.504072	lr=0.130155
Epoch[75] Batch [749]	Speed: 1063.146416 samples/sec	accuracy=0.504047	lr=0.129951
Epoch[75] Batch [799]	Speed: 2394.163908 samples/sec	accuracy=0.503777	lr=0.129746
Epoch[75] Batch [849]	Speed: 6086.255690 samples/sec	accuracy=0.503557	lr=0.129542
Epoch[75] Batch [899]	Speed: 3702.648154 samples/sec	accuracy=0.503495	lr=0.129338
Epoch[75] Batch [949]	Speed: 3985.016979 samples/sec	accuracy=0.503447	lr=0.129133
Epoch[75] Batch [999]	Speed: 2058.277056 samples/sec	accuracy=0.503345	lr=0.128929
Epoch[75] Batch [1049]	Speed: 2945.237676 samples/sec	accuracy=0.503313	lr=0.128725
Epoch[75] Batch [1099]	Speed: 5451.957651 samples/sec	accuracy=0.503300	lr=0.128521
Epoch[75] Batch [1149]	Speed: 5885.722458 samples/sec	accuracy=0.503101	lr=0.128317
Epoch[75] Batch [1199]	Speed: 5392.713300 samples/sec	accuracy=0.502809	lr=0.128113
Epoch[75] Batch [1249]	Speed: 5327.389884 samples/sec	accuracy=0.502631	lr=0.127910
[Epoch 75] training: accuracy=0.502628
[Epoch 75] speed: 3190 samples/sec	time cost: 428.927694
[Epoch 75] validation: err-top1=0.482544 err-top5=0.233805
Epoch[76] Batch [49]	Speed: 2166.761761 samples/sec	accuracy=0.509062	lr=0.127702
Epoch[76] Batch [99]	Speed: 2882.143843 samples/sec	accuracy=0.509229	lr=0.127498
Epoch[76] Batch [149]	Speed: 1860.147694 samples/sec	accuracy=0.508848	lr=0.127295
Epoch[76] Batch [199]	Speed: 5564.715944 samples/sec	accuracy=0.507495	lr=0.127092
Epoch[76] Batch [249]	Speed: 2859.599846 samples/sec	accuracy=0.507086	lr=0.126888
Epoch[76] Batch [299]	Speed: 2833.536278 samples/sec	accuracy=0.506162	lr=0.126685
Epoch[76] Batch [349]	Speed: 1894.874014 samples/sec	accuracy=0.505625	lr=0.126482
Epoch[76] Batch [399]	Speed: 3565.401505 samples/sec	accuracy=0.505667	lr=0.126279
Epoch[76] Batch [449]	Speed: 4876.295442 samples/sec	accuracy=0.505543	lr=0.126076
Epoch[76] Batch [499]	Speed: 2882.229679 samples/sec	accuracy=0.505502	lr=0.125873
Epoch[76] Batch [549]	Speed: 5651.047013 samples/sec	accuracy=0.505664	lr=0.125670
Epoch[76] Batch [599]	Speed: 5609.022061 samples/sec	accuracy=0.505404	lr=0.125468
Epoch[76] Batch [649]	Speed: 3627.730505 samples/sec	accuracy=0.505446	lr=0.125265
Epoch[76] Batch [699]	Speed: 2830.665690 samples/sec	accuracy=0.505479	lr=0.125062
Epoch[76] Batch [749]	Speed: 2304.038280 samples/sec	accuracy=0.505345	lr=0.124860
Epoch[76] Batch [799]	Speed: 3489.688632 samples/sec	accuracy=0.505347	lr=0.124658
Epoch[76] Batch [849]	Speed: 5389.595554 samples/sec	accuracy=0.505402	lr=0.124455
Epoch[76] Batch [899]	Speed: 4529.376957 samples/sec	accuracy=0.505382	lr=0.124253
Epoch[76] Batch [949]	Speed: 5494.736675 samples/sec	accuracy=0.505159	lr=0.124051
Epoch[76] Batch [999]	Speed: 5643.714043 samples/sec	accuracy=0.505294	lr=0.123849
Epoch[76] Batch [1049]	Speed: 5651.450630 samples/sec	accuracy=0.505053	lr=0.123647
Epoch[76] Batch [1099]	Speed: 2845.208380 samples/sec	accuracy=0.505105	lr=0.123446
Epoch[76] Batch [1149]	Speed: 5343.250668 samples/sec	accuracy=0.504947	lr=0.123244
Epoch[76] Batch [1199]	Speed: 1411.654615 samples/sec	accuracy=0.504975	lr=0.123042
Epoch[76] Batch [1249]	Speed: 1730.100417 samples/sec	accuracy=0.504857	lr=0.122841
[Epoch 76] training: accuracy=0.504879
[Epoch 76] speed: 3099 samples/sec	time cost: 423.108741
[Epoch 76] validation: err-top1=0.522899 err-top5=0.273298
Epoch[77] Batch [49]	Speed: 5003.708360 samples/sec	accuracy=0.513262	lr=0.122635
Epoch[77] Batch [99]	Speed: 5627.166274 samples/sec	accuracy=0.512676	lr=0.122434
Epoch[77] Batch [149]	Speed: 2712.493390 samples/sec	accuracy=0.511608	lr=0.122233
Epoch[77] Batch [199]	Speed: 5866.280747 samples/sec	accuracy=0.510654	lr=0.122032
Epoch[77] Batch [249]	Speed: 5851.504784 samples/sec	accuracy=0.509527	lr=0.121831
Epoch[77] Batch [299]	Speed: 5360.727131 samples/sec	accuracy=0.509284	lr=0.121630
Epoch[77] Batch [349]	Speed: 3733.264630 samples/sec	accuracy=0.509001	lr=0.121429
Epoch[77] Batch [399]	Speed: 5585.532946 samples/sec	accuracy=0.508684	lr=0.121228
Epoch[77] Batch [449]	Speed: 3729.845499 samples/sec	accuracy=0.508550	lr=0.121027
Epoch[77] Batch [499]	Speed: 3983.150058 samples/sec	accuracy=0.508355	lr=0.120827
Epoch[77] Batch [549]	Speed: 5488.467761 samples/sec	accuracy=0.508173	lr=0.120626
Epoch[77] Batch [599]	Speed: 2684.674808 samples/sec	accuracy=0.508075	lr=0.120426
Epoch[77] Batch [649]	Speed: 2815.807396 samples/sec	accuracy=0.507939	lr=0.120226
Epoch[77] Batch [699]	Speed: 2912.358364 samples/sec	accuracy=0.507932	lr=0.120025
Epoch[77] Batch [749]	Speed: 4101.641196 samples/sec	accuracy=0.507745	lr=0.119825
Epoch[77] Batch [799]	Speed: 5871.558329 samples/sec	accuracy=0.507712	lr=0.119625
Epoch[77] Batch [849]	Speed: 5654.906631 samples/sec	accuracy=0.507734	lr=0.119425
Epoch[77] Batch [899]	Speed: 5681.479540 samples/sec	accuracy=0.507638	lr=0.119226
Epoch[77] Batch [949]	Speed: 2004.897447 samples/sec	accuracy=0.507412	lr=0.119026
Epoch[77] Batch [999]	Speed: 1339.274728 samples/sec	accuracy=0.507272	lr=0.118826
Epoch[77] Batch [1049]	Speed: 5765.730087 samples/sec	accuracy=0.507076	lr=0.118627
Epoch[77] Batch [1099]	Speed: 5095.281174 samples/sec	accuracy=0.507018	lr=0.118427
Epoch[77] Batch [1149]	Speed: 1195.397925 samples/sec	accuracy=0.507260	lr=0.118228
Epoch[77] Batch [1199]	Speed: 5796.113829 samples/sec	accuracy=0.507298	lr=0.118029
Epoch[77] Batch [1249]	Speed: 1078.877811 samples/sec	accuracy=0.507301	lr=0.117830
[Epoch 77] training: accuracy=0.507315
[Epoch 77] speed: 3197 samples/sec	time cost: 416.918147
[Epoch 77] validation: err-top1=0.470225 err-top5=0.224490
Epoch[78] Batch [49]	Speed: 3069.079785 samples/sec	accuracy=0.515898	lr=0.117627
Epoch[78] Batch [99]	Speed: 3282.935366 samples/sec	accuracy=0.515547	lr=0.117428
Epoch[78] Batch [149]	Speed: 5620.136462 samples/sec	accuracy=0.514753	lr=0.117229
Epoch[78] Batch [199]	Speed: 5688.210651 samples/sec	accuracy=0.513784	lr=0.117030
Epoch[78] Batch [249]	Speed: 5572.802999 samples/sec	accuracy=0.513836	lr=0.116831
Epoch[78] Batch [299]	Speed: 5670.159762 samples/sec	accuracy=0.512738	lr=0.116633
Epoch[78] Batch [349]	Speed: 3124.437612 samples/sec	accuracy=0.511833	lr=0.116434
Epoch[78] Batch [399]	Speed: 2060.575042 samples/sec	accuracy=0.511084	lr=0.116236
Epoch[78] Batch [449]	Speed: 2209.555543 samples/sec	accuracy=0.511428	lr=0.116038
Epoch[78] Batch [499]	Speed: 4221.238387 samples/sec	accuracy=0.510676	lr=0.115840
Epoch[78] Batch [549]	Speed: 2772.115331 samples/sec	accuracy=0.510334	lr=0.115642
Epoch[78] Batch [599]	Speed: 2504.548310 samples/sec	accuracy=0.510632	lr=0.115444
Epoch[78] Batch [649]	Speed: 2256.228705 samples/sec	accuracy=0.510627	lr=0.115246
Epoch[78] Batch [699]	Speed: 3180.835839 samples/sec	accuracy=0.510308	lr=0.115048
Epoch[78] Batch [749]	Speed: 3676.798490 samples/sec	accuracy=0.510152	lr=0.114850
Epoch[78] Batch [799]	Speed: 3385.862903 samples/sec	accuracy=0.510105	lr=0.114653
Epoch[78] Batch [849]	Speed: 5699.260834 samples/sec	accuracy=0.509817	lr=0.114455
Epoch[78] Batch [899]	Speed: 2772.806355 samples/sec	accuracy=0.509678	lr=0.114258
Epoch[78] Batch [949]	Speed: 3290.041846 samples/sec	accuracy=0.509423	lr=0.114061
Epoch[78] Batch [999]	Speed: 2401.785837 samples/sec	accuracy=0.509444	lr=0.113864
Epoch[78] Batch [1049]	Speed: 2643.719598 samples/sec	accuracy=0.509376	lr=0.113667
Epoch[78] Batch [1099]	Speed: 5763.224919 samples/sec	accuracy=0.509236	lr=0.113470
Epoch[78] Batch [1149]	Speed: 2823.903840 samples/sec	accuracy=0.509062	lr=0.113273
Epoch[78] Batch [1199]	Speed: 3328.058066 samples/sec	accuracy=0.509046	lr=0.113076
Epoch[78] Batch [1249]	Speed: 3364.133809 samples/sec	accuracy=0.509045	lr=0.112880
[Epoch 78] training: accuracy=0.509064
[Epoch 78] speed: 3253 samples/sec	time cost: 415.839741
[Epoch 78] validation: err-top1=0.495556 err-top5=0.248426
Epoch[79] Batch [49]	Speed: 3309.438591 samples/sec	accuracy=0.515859	lr=0.112679
Epoch[79] Batch [99]	Speed: 3384.669237 samples/sec	accuracy=0.516006	lr=0.112483
Epoch[79] Batch [149]	Speed: 2351.313643 samples/sec	accuracy=0.515326	lr=0.112287
Epoch[79] Batch [199]	Speed: 5231.151926 samples/sec	accuracy=0.513901	lr=0.112090
Epoch[79] Batch [249]	Speed: 2364.494598 samples/sec	accuracy=0.513414	lr=0.111894
Epoch[79] Batch [299]	Speed: 5625.501300 samples/sec	accuracy=0.513441	lr=0.111698
Epoch[79] Batch [349]	Speed: 5921.001977 samples/sec	accuracy=0.513516	lr=0.111502
Epoch[79] Batch [399]	Speed: 3379.254430 samples/sec	accuracy=0.513040	lr=0.111307
Epoch[79] Batch [449]	Speed: 2330.073471 samples/sec	accuracy=0.513422	lr=0.111111
Epoch[79] Batch [499]	Speed: 3711.021061 samples/sec	accuracy=0.512699	lr=0.110915
Epoch[79] Batch [549]	Speed: 3223.923154 samples/sec	accuracy=0.512550	lr=0.110720
Epoch[79] Batch [599]	Speed: 5847.091705 samples/sec	accuracy=0.512642	lr=0.110525
Epoch[79] Batch [649]	Speed: 3079.987546 samples/sec	accuracy=0.512453	lr=0.110329
Epoch[79] Batch [699]	Speed: 3197.803528 samples/sec	accuracy=0.512160	lr=0.110134
Epoch[79] Batch [749]	Speed: 2220.861466 samples/sec	accuracy=0.511995	lr=0.109939
Epoch[79] Batch [799]	Speed: 2955.606084 samples/sec	accuracy=0.511913	lr=0.109744
Epoch[79] Batch [849]	Speed: 4070.364887 samples/sec	accuracy=0.511657	lr=0.109549
Epoch[79] Batch [899]	Speed: 6062.759762 samples/sec	accuracy=0.511636	lr=0.109355
Epoch[79] Batch [949]	Speed: 5558.546547 samples/sec	accuracy=0.511608	lr=0.109160
Epoch[79] Batch [999]	Speed: 3387.450746 samples/sec	accuracy=0.511473	lr=0.108966
Epoch[79] Batch [1049]	Speed: 2506.430456 samples/sec	accuracy=0.511491	lr=0.108771
Epoch[79] Batch [1099]	Speed: 3404.687914 samples/sec	accuracy=0.511441	lr=0.108577
Epoch[79] Batch [1149]	Speed: 6005.357989 samples/sec	accuracy=0.511493	lr=0.108383
Epoch[79] Batch [1199]	Speed: 5698.913878 samples/sec	accuracy=0.511445	lr=0.108189
Epoch[79] Batch [1249]	Speed: 3367.359232 samples/sec	accuracy=0.511524	lr=0.107995
[Epoch 79] training: accuracy=0.511515
[Epoch 79] speed: 3515 samples/sec	time cost: 385.018352
[Epoch 79] validation: err-top1=0.566207 err-top5=0.316227
Epoch[80] Batch [49]	Speed: 3336.349947 samples/sec	accuracy=0.522090	lr=0.107797
Epoch[80] Batch [99]	Speed: 2429.807034 samples/sec	accuracy=0.520566	lr=0.107603
Epoch[80] Batch [149]	Speed: 5924.898608 samples/sec	accuracy=0.517318	lr=0.107410
Epoch[80] Batch [199]	Speed: 2446.248698 samples/sec	accuracy=0.517261	lr=0.107216
Epoch[80] Batch [249]	Speed: 2468.221760 samples/sec	accuracy=0.518172	lr=0.107023
Epoch[80] Batch [299]	Speed: 5715.267359 samples/sec	accuracy=0.517542	lr=0.106830
Epoch[80] Batch [349]	Speed: 6054.386903 samples/sec	accuracy=0.516627	lr=0.106636
Epoch[80] Batch [399]	Speed: 5106.143438 samples/sec	accuracy=0.516323	lr=0.106443
Epoch[80] Batch [449]	Speed: 2847.055187 samples/sec	accuracy=0.516560	lr=0.106250
Epoch[80] Batch [499]	Speed: 5251.985065 samples/sec	accuracy=0.516484	lr=0.106057
Epoch[80] Batch [549]	Speed: 2506.259655 samples/sec	accuracy=0.516241	lr=0.105865
Epoch[80] Batch [599]	Speed: 2457.927906 samples/sec	accuracy=0.516206	lr=0.105672
Epoch[80] Batch [649]	Speed: 3491.887561 samples/sec	accuracy=0.515834	lr=0.105480
Epoch[80] Batch [699]	Speed: 5468.484630 samples/sec	accuracy=0.515626	lr=0.105287
Epoch[80] Batch [749]	Speed: 5451.056320 samples/sec	accuracy=0.515309	lr=0.105095
Epoch[80] Batch [799]	Speed: 5800.036208 samples/sec	accuracy=0.515232	lr=0.104903
Epoch[80] Batch [849]	Speed: 5784.485190 samples/sec	accuracy=0.515123	lr=0.104711
Epoch[80] Batch [899]	Speed: 3232.582439 samples/sec	accuracy=0.515082	lr=0.104519
Epoch[80] Batch [949]	Speed: 3671.988143 samples/sec	accuracy=0.514946	lr=0.104327
Epoch[80] Batch [999]	Speed: 5362.176782 samples/sec	accuracy=0.514785	lr=0.104135
Epoch[80] Batch [1049]	Speed: 3492.617612 samples/sec	accuracy=0.514767	lr=0.103944
Epoch[80] Batch [1099]	Speed: 5723.825866 samples/sec	accuracy=0.514633	lr=0.103752
Epoch[80] Batch [1149]	Speed: 2846.348508 samples/sec	accuracy=0.514360	lr=0.103561
Epoch[80] Batch [1199]	Speed: 4094.465653 samples/sec	accuracy=0.514298	lr=0.103370
Epoch[80] Batch [1249]	Speed: 3378.411808 samples/sec	accuracy=0.514216	lr=0.103178
[Epoch 80] training: accuracy=0.514232
[Epoch 80] speed: 3729 samples/sec	time cost: 354.894901
[Epoch 80] validation: err-top1=0.493044 err-top5=0.242606
Epoch[81] Batch [49]	Speed: 3343.030736 samples/sec	accuracy=0.518516	lr=0.102984
Epoch[81] Batch [99]	Speed: 5576.894885 samples/sec	accuracy=0.520488	lr=0.102793
Epoch[81] Batch [149]	Speed: 5598.739199 samples/sec	accuracy=0.518633	lr=0.102602
Epoch[81] Batch [199]	Speed: 5630.486336 samples/sec	accuracy=0.518535	lr=0.102411
Epoch[81] Batch [249]	Speed: 6022.307267 samples/sec	accuracy=0.518520	lr=0.102221
Epoch[81] Batch [299]	Speed: 1920.489989 samples/sec	accuracy=0.517754	lr=0.102030
Epoch[81] Batch [349]	Speed: 1884.329414 samples/sec	accuracy=0.518401	lr=0.101840
Epoch[81] Batch [399]	Speed: 2419.388649 samples/sec	accuracy=0.518113	lr=0.101650
Epoch[81] Batch [449]	Speed: 5484.248386 samples/sec	accuracy=0.517776	lr=0.101460
Epoch[81] Batch [499]	Speed: 3611.218469 samples/sec	accuracy=0.517514	lr=0.101270
Epoch[81] Batch [549]	Speed: 4350.409877 samples/sec	accuracy=0.517647	lr=0.101080
Epoch[81] Batch [599]	Speed: 5518.803041 samples/sec	accuracy=0.517650	lr=0.100890
Epoch[81] Batch [649]	Speed: 3509.515825 samples/sec	accuracy=0.517631	lr=0.100701
Epoch[81] Batch [699]	Speed: 5163.980157 samples/sec	accuracy=0.517326	lr=0.100511
Epoch[81] Batch [749]	Speed: 3399.132462 samples/sec	accuracy=0.517078	lr=0.100322
Epoch[81] Batch [799]	Speed: 3476.763673 samples/sec	accuracy=0.516832	lr=0.100132
Epoch[81] Batch [849]	Speed: 1867.354709 samples/sec	accuracy=0.516438	lr=0.099943
Epoch[81] Batch [899]	Speed: 5781.002096 samples/sec	accuracy=0.516380	lr=0.099754
Epoch[81] Batch [949]	Speed: 3506.135078 samples/sec	accuracy=0.516196	lr=0.099565
Epoch[81] Batch [999]	Speed: 2503.944454 samples/sec	accuracy=0.516048	lr=0.099377
Epoch[81] Batch [1049]	Speed: 2509.652974 samples/sec	accuracy=0.516134	lr=0.099188
Epoch[81] Batch [1099]	Speed: 3095.372695 samples/sec	accuracy=0.515985	lr=0.098999
Epoch[81] Batch [1149]	Speed: 2892.473331 samples/sec	accuracy=0.516047	lr=0.098811
Epoch[81] Batch [1199]	Speed: 3332.547505 samples/sec	accuracy=0.516050	lr=0.098623
Epoch[81] Batch [1249]	Speed: 5513.970397 samples/sec	accuracy=0.515912	lr=0.098434
[Epoch 81] training: accuracy=0.515930
[Epoch 81] speed: 3404 samples/sec	time cost: 389.417149
[Epoch 81] validation: err-top1=0.461019 err-top5=0.218079
Epoch[82] Batch [49]	Speed: 5211.642084 samples/sec	accuracy=0.522969	lr=0.098243
Epoch[82] Batch [99]	Speed: 3617.909169 samples/sec	accuracy=0.523057	lr=0.098055
Epoch[82] Batch [149]	Speed: 4150.792106 samples/sec	accuracy=0.522799	lr=0.097867
Epoch[82] Batch [199]	Speed: 5510.997566 samples/sec	accuracy=0.523315	lr=0.097679
Epoch[82] Batch [249]	Speed: 5043.938021 samples/sec	accuracy=0.523656	lr=0.097492
Epoch[82] Batch [299]	Speed: 5530.140132 samples/sec	accuracy=0.523105	lr=0.097304
Epoch[82] Batch [349]	Speed: 3448.405200 samples/sec	accuracy=0.523022	lr=0.097117
Epoch[82] Batch [399]	Speed: 5671.239401 samples/sec	accuracy=0.522520	lr=0.096930
Epoch[82] Batch [449]	Speed: 5359.783339 samples/sec	accuracy=0.522313	lr=0.096743
Epoch[82] Batch [499]	Speed: 3400.278314 samples/sec	accuracy=0.522086	lr=0.096556
Epoch[82] Batch [549]	Speed: 2447.415103 samples/sec	accuracy=0.521603	lr=0.096369
Epoch[82] Batch [599]	Speed: 5233.014312 samples/sec	accuracy=0.521291	lr=0.096182
Epoch[82] Batch [649]	Speed: 2475.465327 samples/sec	accuracy=0.521005	lr=0.095996
Epoch[82] Batch [699]	Speed: 3536.901849 samples/sec	accuracy=0.520799	lr=0.095809
Epoch[82] Batch [749]	Speed: 2434.419275 samples/sec	accuracy=0.520613	lr=0.095623
Epoch[82] Batch [799]	Speed: 3259.414236 samples/sec	accuracy=0.520482	lr=0.095437
Epoch[82] Batch [849]	Speed: 4238.403178 samples/sec	accuracy=0.520162	lr=0.095250
Epoch[82] Batch [899]	Speed: 5984.438413 samples/sec	accuracy=0.519849	lr=0.095064
Epoch[82] Batch [949]	Speed: 3014.487348 samples/sec	accuracy=0.519715	lr=0.094879
Epoch[82] Batch [999]	Speed: 5185.135115 samples/sec	accuracy=0.519656	lr=0.094693
Epoch[82] Batch [1049]	Speed: 3671.137005 samples/sec	accuracy=0.519368	lr=0.094507
Epoch[82] Batch [1099]	Speed: 3432.355640 samples/sec	accuracy=0.519165	lr=0.094322
Epoch[82] Batch [1149]	Speed: 3447.625210 samples/sec	accuracy=0.519061	lr=0.094137
Epoch[82] Batch [1199]	Speed: 4984.661049 samples/sec	accuracy=0.519083	lr=0.093951
Epoch[82] Batch [1249]	Speed: 5111.645491 samples/sec	accuracy=0.519031	lr=0.093766
[Epoch 82] training: accuracy=0.519032
[Epoch 82] speed: 3903 samples/sec	time cost: 351.468665
[Epoch 82] validation: err-top1=0.449438 err-top5=0.208048
Epoch[83] Batch [49]	Speed: 3732.447648 samples/sec	accuracy=0.523379	lr=0.093578
Epoch[83] Batch [99]	Speed: 5374.307402 samples/sec	accuracy=0.523994	lr=0.093393
Epoch[83] Batch [149]	Speed: 3266.158074 samples/sec	accuracy=0.523275	lr=0.093208
Epoch[83] Batch [199]	Speed: 2551.445915 samples/sec	accuracy=0.522490	lr=0.093024
Epoch[83] Batch [249]	Speed: 3234.569657 samples/sec	accuracy=0.523449	lr=0.092839
Epoch[83] Batch [299]	Speed: 3221.066400 samples/sec	accuracy=0.523076	lr=0.092655
Epoch[83] Batch [349]	Speed: 3429.468985 samples/sec	accuracy=0.522762	lr=0.092471
Epoch[83] Batch [399]	Speed: 3303.208531 samples/sec	accuracy=0.522593	lr=0.092287
Epoch[83] Batch [449]	Speed: 5775.173948 samples/sec	accuracy=0.522452	lr=0.092103
Epoch[83] Batch [499]	Speed: 3094.246180 samples/sec	accuracy=0.522729	lr=0.091919
Epoch[83] Batch [549]	Speed: 4955.825885 samples/sec	accuracy=0.522532	lr=0.091735
Epoch[83] Batch [599]	Speed: 3396.341466 samples/sec	accuracy=0.522365	lr=0.091552
Epoch[83] Batch [649]	Speed: 2527.705539 samples/sec	accuracy=0.522266	lr=0.091368
Epoch[83] Batch [699]	Speed: 2326.480675 samples/sec	accuracy=0.522076	lr=0.091185
Epoch[83] Batch [749]	Speed: 3311.693708 samples/sec	accuracy=0.522070	lr=0.091002
Epoch[83] Batch [799]	Speed: 3464.095251 samples/sec	accuracy=0.522014	lr=0.090819
Epoch[83] Batch [849]	Speed: 3483.484892 samples/sec	accuracy=0.521892	lr=0.090636
Epoch[83] Batch [899]	Speed: 5488.579560 samples/sec	accuracy=0.521694	lr=0.090453
Epoch[83] Batch [949]	Speed: 2536.757480 samples/sec	accuracy=0.521571	lr=0.090270
Epoch[83] Batch [999]	Speed: 3520.379873 samples/sec	accuracy=0.521430	lr=0.090088
Epoch[83] Batch [1049]	Speed: 5101.234558 samples/sec	accuracy=0.521535	lr=0.089905
Epoch[83] Batch [1099]	Speed: 5742.741175 samples/sec	accuracy=0.521388	lr=0.089723
Epoch[83] Batch [1149]	Speed: 2738.429785 samples/sec	accuracy=0.521232	lr=0.089541
Epoch[83] Batch [1199]	Speed: 3365.299057 samples/sec	accuracy=0.521152	lr=0.089359
Epoch[83] Batch [1249]	Speed: 3462.430629 samples/sec	accuracy=0.520974	lr=0.089177
[Epoch 83] training: accuracy=0.520964
[Epoch 83] speed: 3448 samples/sec	time cost: 395.658321
[Epoch 83] validation: err-top1=0.527344 err-top5=0.279157
Epoch[84] Batch [49]	Speed: 2976.187159 samples/sec	accuracy=0.526035	lr=0.088992
Epoch[84] Batch [99]	Speed: 3490.717389 samples/sec	accuracy=0.526729	lr=0.088810
Epoch[84] Batch [149]	Speed: 3467.510596 samples/sec	accuracy=0.526810	lr=0.088629
Epoch[84] Batch [199]	Speed: 3477.618288 samples/sec	accuracy=0.526963	lr=0.088448
Epoch[84] Batch [249]	Speed: 1141.522933 samples/sec	accuracy=0.525945	lr=0.088266
Epoch[84] Batch [299]	Speed: 4653.930342 samples/sec	accuracy=0.525618	lr=0.088085
Epoch[84] Batch [349]	Speed: 2539.933136 samples/sec	accuracy=0.525232	lr=0.087904
Epoch[84] Batch [399]	Speed: 2945.482844 samples/sec	accuracy=0.525132	lr=0.087724
Epoch[84] Batch [449]	Speed: 2014.728134 samples/sec	accuracy=0.524946	lr=0.087543
Epoch[84] Batch [499]	Speed: 5575.209152 samples/sec	accuracy=0.524797	lr=0.087363
Epoch[84] Batch [549]	Speed: 2473.665853 samples/sec	accuracy=0.524659	lr=0.087182
Epoch[84] Batch [599]	Speed: 2325.884703 samples/sec	accuracy=0.524676	lr=0.087002
Epoch[84] Batch [649]	Speed: 3510.004206 samples/sec	accuracy=0.524507	lr=0.086822
Epoch[84] Batch [699]	Speed: 4320.669229 samples/sec	accuracy=0.524587	lr=0.086642
Epoch[84] Batch [749]	Speed: 5312.322535 samples/sec	accuracy=0.524699	lr=0.086462
Epoch[84] Batch [799]	Speed: 3458.945273 samples/sec	accuracy=0.524653	lr=0.086282
Epoch[84] Batch [849]	Speed: 4009.078392 samples/sec	accuracy=0.524685	lr=0.086103
Epoch[84] Batch [899]	Speed: 3267.304645 samples/sec	accuracy=0.524261	lr=0.085923
Epoch[84] Batch [949]	Speed: 3368.982509 samples/sec	accuracy=0.524127	lr=0.085744
Epoch[84] Batch [999]	Speed: 3839.138613 samples/sec	accuracy=0.524047	lr=0.085565
Epoch[84] Batch [1049]	Speed: 5564.937296 samples/sec	accuracy=0.524051	lr=0.085386
Epoch[84] Batch [1099]	Speed: 3134.794471 samples/sec	accuracy=0.523856	lr=0.085207
Epoch[84] Batch [1149]	Speed: 3421.789506 samples/sec	accuracy=0.523797	lr=0.085028
Epoch[84] Batch [1199]	Speed: 5839.944141 samples/sec	accuracy=0.523858	lr=0.084850
Epoch[84] Batch [1249]	Speed: 3527.209542 samples/sec	accuracy=0.523888	lr=0.084671
[Epoch 84] training: accuracy=0.523888
[Epoch 84] speed: 3168 samples/sec	time cost: 419.371057
[Epoch 84] validation: err-top1=0.446050 err-top5=0.201212
Epoch[85] Batch [49]	Speed: 3349.755211 samples/sec	accuracy=0.530937	lr=0.084489
Epoch[85] Batch [99]	Speed: 5611.633508 samples/sec	accuracy=0.530459	lr=0.084311
Epoch[85] Batch [149]	Speed: 5762.853893 samples/sec	accuracy=0.530430	lr=0.084133
Epoch[85] Batch [199]	Speed: 5529.933502 samples/sec	accuracy=0.529160	lr=0.083955
Epoch[85] Batch [249]	Speed: 4371.190187 samples/sec	accuracy=0.528457	lr=0.083777
Epoch[85] Batch [299]	Speed: 4299.653919 samples/sec	accuracy=0.529128	lr=0.083600
Epoch[85] Batch [349]	Speed: 2549.516254 samples/sec	accuracy=0.529241	lr=0.083422
Epoch[85] Batch [399]	Speed: 5368.099572 samples/sec	accuracy=0.529075	lr=0.083245
Epoch[85] Batch [449]	Speed: 3711.992362 samples/sec	accuracy=0.528438	lr=0.083067
Epoch[85] Batch [499]	Speed: 2741.623791 samples/sec	accuracy=0.527811	lr=0.082890
Epoch[85] Batch [549]	Speed: 3455.219798 samples/sec	accuracy=0.527388	lr=0.082713
Epoch[85] Batch [599]	Speed: 5611.447723 samples/sec	accuracy=0.527484	lr=0.082537
Epoch[85] Batch [649]	Speed: 2724.590460 samples/sec	accuracy=0.527488	lr=0.082360
Epoch[85] Batch [699]	Speed: 2665.262121 samples/sec	accuracy=0.527093	lr=0.082183
Epoch[85] Batch [749]	Speed: 3461.199613 samples/sec	accuracy=0.527116	lr=0.082007
Epoch[85] Batch [799]	Speed: 1937.584796 samples/sec	accuracy=0.527046	lr=0.081831
Epoch[85] Batch [849]	Speed: 5677.576522 samples/sec	accuracy=0.526711	lr=0.081655
Epoch[85] Batch [899]	Speed: 4515.707544 samples/sec	accuracy=0.526512	lr=0.081479
Epoch[85] Batch [949]	Speed: 3341.702849 samples/sec	accuracy=0.526620	lr=0.081303
Epoch[85] Batch [999]	Speed: 5430.442664 samples/sec	accuracy=0.526536	lr=0.081127
Epoch[85] Batch [1049]	Speed: 5296.445738 samples/sec	accuracy=0.526477	lr=0.080952
Epoch[85] Batch [1099]	Speed: 3406.337880 samples/sec	accuracy=0.526427	lr=0.080776
Epoch[85] Batch [1149]	Speed: 1555.495334 samples/sec	accuracy=0.526410	lr=0.080601
Epoch[85] Batch [1199]	Speed: 1950.883464 samples/sec	accuracy=0.526441	lr=0.080426
Epoch[85] Batch [1249]	Speed: 5251.323528 samples/sec	accuracy=0.526302	lr=0.080251
[Epoch 85] training: accuracy=0.526298
[Epoch 85] speed: 3457 samples/sec	time cost: 425.618314
[Epoch 85] validation: err-top1=0.544643 err-top5=0.295480
Epoch[86] Batch [49]	Speed: 1344.982704 samples/sec	accuracy=0.533672	lr=0.080073
Epoch[86] Batch [99]	Speed: 1985.835028 samples/sec	accuracy=0.530928	lr=0.079898
Epoch[86] Batch [149]	Speed: 4669.705056 samples/sec	accuracy=0.530527	lr=0.079723
Epoch[86] Batch [199]	Speed: 2797.630770 samples/sec	accuracy=0.529927	lr=0.079549
Epoch[86] Batch [249]	Speed: 3452.660740 samples/sec	accuracy=0.529168	lr=0.079375
Epoch[86] Batch [299]	Speed: 5847.084382 samples/sec	accuracy=0.529069	lr=0.079201
Epoch[86] Batch [349]	Speed: 3662.461574 samples/sec	accuracy=0.529102	lr=0.079027
Epoch[86] Batch [399]	Speed: 5593.455709 samples/sec	accuracy=0.529307	lr=0.078853
Epoch[86] Batch [449]	Speed: 5756.895736 samples/sec	accuracy=0.529564	lr=0.078679
Epoch[86] Batch [499]	Speed: 3406.585793 samples/sec	accuracy=0.528939	lr=0.078506
Epoch[86] Batch [549]	Speed: 5517.523771 samples/sec	accuracy=0.529212	lr=0.078332
Epoch[86] Batch [599]	Speed: 5837.731595 samples/sec	accuracy=0.529336	lr=0.078159
Epoch[86] Batch [649]	Speed: 3600.631463 samples/sec	accuracy=0.529163	lr=0.077986
Epoch[86] Batch [699]	Speed: 3366.713537 samples/sec	accuracy=0.529093	lr=0.077813
Epoch[86] Batch [749]	Speed: 5703.357015 samples/sec	accuracy=0.528974	lr=0.077640
Epoch[86] Batch [799]	Speed: 5386.358061 samples/sec	accuracy=0.529055	lr=0.077467
Epoch[86] Batch [849]	Speed: 3303.245826 samples/sec	accuracy=0.528998	lr=0.077295
Epoch[86] Batch [899]	Speed: 3437.107807 samples/sec	accuracy=0.528966	lr=0.077123
Epoch[86] Batch [949]	Speed: 5544.692962 samples/sec	accuracy=0.528765	lr=0.076950
Epoch[86] Batch [999]	Speed: 3431.335988 samples/sec	accuracy=0.528731	lr=0.076778
Epoch[86] Batch [1049]	Speed: 5180.292305 samples/sec	accuracy=0.528604	lr=0.076606
Epoch[86] Batch [1099]	Speed: 3721.787076 samples/sec	accuracy=0.528435	lr=0.076435
Epoch[86] Batch [1149]	Speed: 5625.665174 samples/sec	accuracy=0.528469	lr=0.076263
Epoch[86] Batch [1199]	Speed: 5764.222856 samples/sec	accuracy=0.528389	lr=0.076091
Epoch[86] Batch [1249]	Speed: 2906.229566 samples/sec	accuracy=0.528454	lr=0.075920
[Epoch 86] training: accuracy=0.528484
[Epoch 86] speed: 3723 samples/sec	time cost: 357.041701
[Epoch 86] validation: err-top1=0.535017 err-top5=0.283522
Epoch[87] Batch [49]	Speed: 3365.876259 samples/sec	accuracy=0.533301	lr=0.075745
Epoch[87] Batch [99]	Speed: 1912.586438 samples/sec	accuracy=0.534131	lr=0.075574
Epoch[87] Batch [149]	Speed: 5935.136433 samples/sec	accuracy=0.534375	lr=0.075404
Epoch[87] Batch [199]	Speed: 5500.794049 samples/sec	accuracy=0.534194	lr=0.075233
Epoch[87] Batch [249]	Speed: 3541.898125 samples/sec	accuracy=0.534039	lr=0.075062
Epoch[87] Batch [299]	Speed: 3392.629279 samples/sec	accuracy=0.533906	lr=0.074892
Epoch[87] Batch [349]	Speed: 5769.785829 samples/sec	accuracy=0.533426	lr=0.074721
Epoch[87] Batch [399]	Speed: 2459.428312 samples/sec	accuracy=0.533105	lr=0.074551
Epoch[87] Batch [449]	Speed: 2559.804794 samples/sec	accuracy=0.532799	lr=0.074381
Epoch[87] Batch [499]	Speed: 3283.917426 samples/sec	accuracy=0.532707	lr=0.074211
Epoch[87] Batch [549]	Speed: 6008.257151 samples/sec	accuracy=0.532697	lr=0.074042
Epoch[87] Batch [599]	Speed: 2451.950172 samples/sec	accuracy=0.532550	lr=0.073872
Epoch[87] Batch [649]	Speed: 5808.794736 samples/sec	accuracy=0.532435	lr=0.073703
Epoch[87] Batch [699]	Speed: 5766.058597 samples/sec	accuracy=0.532072	lr=0.073534
Epoch[87] Batch [749]	Speed: 4182.540116 samples/sec	accuracy=0.531721	lr=0.073364
Epoch[87] Batch [799]	Speed: 3625.909537 samples/sec	accuracy=0.531486	lr=0.073196
Epoch[87] Batch [849]	Speed: 3422.354454 samples/sec	accuracy=0.531342	lr=0.073027
Epoch[87] Batch [899]	Speed: 2808.172649 samples/sec	accuracy=0.531345	lr=0.072858
Epoch[87] Batch [949]	Speed: 2467.515299 samples/sec	accuracy=0.531208	lr=0.072690
Epoch[87] Batch [999]	Speed: 3367.630761 samples/sec	accuracy=0.531322	lr=0.072521
Epoch[87] Batch [1049]	Speed: 3126.701715 samples/sec	accuracy=0.531137	lr=0.072353
Epoch[87] Batch [1099]	Speed: 3413.193365 samples/sec	accuracy=0.531159	lr=0.072185
Epoch[87] Batch [1149]	Speed: 2996.399700 samples/sec	accuracy=0.531157	lr=0.072017
Epoch[87] Batch [1199]	Speed: 3769.676056 samples/sec	accuracy=0.531104	lr=0.071849
Epoch[87] Batch [1249]	Speed: 5404.902101 samples/sec	accuracy=0.531036	lr=0.071682
[Epoch 87] training: accuracy=0.531046
[Epoch 87] speed: 3466 samples/sec	time cost: 384.710828
[Epoch 87] validation: err-top1=0.442118 err-top5=0.204468
Epoch[88] Batch [49]	Speed: 5652.217868 samples/sec	accuracy=0.539219	lr=0.071511
Epoch[88] Batch [99]	Speed: 5498.565722 samples/sec	accuracy=0.538828	lr=0.071344
Epoch[88] Batch [149]	Speed: 3127.579715 samples/sec	accuracy=0.538053	lr=0.071177
Epoch[88] Batch [199]	Speed: 2269.864865 samples/sec	accuracy=0.537427	lr=0.071010
Epoch[88] Batch [249]	Speed: 2304.539910 samples/sec	accuracy=0.536656	lr=0.070843
Epoch[88] Batch [299]	Speed: 3374.349946 samples/sec	accuracy=0.535957	lr=0.070676
Epoch[88] Batch [349]	Speed: 5658.908114 samples/sec	accuracy=0.535318	lr=0.070510
Epoch[88] Batch [399]	Speed: 5706.833134 samples/sec	accuracy=0.535283	lr=0.070343
Epoch[88] Batch [449]	Speed: 3228.568577 samples/sec	accuracy=0.535352	lr=0.070177
Epoch[88] Batch [499]	Speed: 2542.360847 samples/sec	accuracy=0.535053	lr=0.070011
Epoch[88] Batch [549]	Speed: 2825.262812 samples/sec	accuracy=0.534966	lr=0.069845
Epoch[88] Batch [599]	Speed: 5593.775517 samples/sec	accuracy=0.535003	lr=0.069680
Epoch[88] Batch [649]	Speed: 5736.004564 samples/sec	accuracy=0.534701	lr=0.069514
Epoch[88] Batch [699]	Speed: 3802.050127 samples/sec	accuracy=0.534732	lr=0.069349
Epoch[88] Batch [749]	Speed: 5466.518251 samples/sec	accuracy=0.534564	lr=0.069183
Epoch[88] Batch [799]	Speed: 1946.053789 samples/sec	accuracy=0.534543	lr=0.069018
Epoch[88] Batch [849]	Speed: 5637.630286 samples/sec	accuracy=0.534594	lr=0.068853
Epoch[88] Batch [899]	Speed: 2779.457953 samples/sec	accuracy=0.534824	lr=0.068689
Epoch[88] Batch [949]	Speed: 5565.605205 samples/sec	accuracy=0.534543	lr=0.068524
Epoch[88] Batch [999]	Speed: 3515.371684 samples/sec	accuracy=0.534419	lr=0.068359
Epoch[88] Batch [1049]	Speed: 3800.329820 samples/sec	accuracy=0.534216	lr=0.068195
Epoch[88] Batch [1099]	Speed: 3339.005277 samples/sec	accuracy=0.534097	lr=0.068031
Epoch[88] Batch [1149]	Speed: 2155.477452 samples/sec	accuracy=0.534127	lr=0.067867
Epoch[88] Batch [1199]	Speed: 3155.477207 samples/sec	accuracy=0.534183	lr=0.067703
Epoch[88] Batch [1249]	Speed: 4975.346762 samples/sec	accuracy=0.533971	lr=0.067539
[Epoch 88] training: accuracy=0.533960
[Epoch 88] speed: 3510 samples/sec	time cost: 379.237479
[Epoch 88] validation: err-top1=0.435527 err-top5=0.199099
Epoch[89] Batch [49]	Speed: 4786.070168 samples/sec	accuracy=0.541035	lr=0.067373
Epoch[89] Batch [99]	Speed: 3449.311746 samples/sec	accuracy=0.540537	lr=0.067209
Epoch[89] Batch [149]	Speed: 3170.674239 samples/sec	accuracy=0.540286	lr=0.067046
Epoch[89] Batch [199]	Speed: 5531.789026 samples/sec	accuracy=0.539419	lr=0.066883
Epoch[89] Batch [249]	Speed: 5758.658247 samples/sec	accuracy=0.538555	lr=0.066720
Epoch[89] Batch [299]	Speed: 5555.580803 samples/sec	accuracy=0.538418	lr=0.066557
Epoch[89] Batch [349]	Speed: 3629.909260 samples/sec	accuracy=0.538156	lr=0.066395
Epoch[89] Batch [399]	Speed: 5531.235202 samples/sec	accuracy=0.537305	lr=0.066232
Epoch[89] Batch [449]	Speed: 3654.887505 samples/sec	accuracy=0.537190	lr=0.066070
Epoch[89] Batch [499]	Speed: 2462.832872 samples/sec	accuracy=0.537469	lr=0.065908
Epoch[89] Batch [549]	Speed: 5990.786555 samples/sec	accuracy=0.537069	lr=0.065746
Epoch[89] Batch [599]	Speed: 3441.950605 samples/sec	accuracy=0.536717	lr=0.065584
Epoch[89] Batch [649]	Speed: 5353.024090 samples/sec	accuracy=0.536653	lr=0.065423
Epoch[89] Batch [699]	Speed: 3447.639656 samples/sec	accuracy=0.536482	lr=0.065261
Epoch[89] Batch [749]	Speed: 3464.416139 samples/sec	accuracy=0.536628	lr=0.065100
Epoch[89] Batch [799]	Speed: 3496.107160 samples/sec	accuracy=0.536561	lr=0.064939
Epoch[89] Batch [849]	Speed: 3515.787098 samples/sec	accuracy=0.536680	lr=0.064778
Epoch[89] Batch [899]	Speed: 4139.571747 samples/sec	accuracy=0.536564	lr=0.064617
Epoch[89] Batch [949]	Speed: 3041.264945 samples/sec	accuracy=0.536437	lr=0.064456
Epoch[89] Batch [999]	Speed: 2599.244319 samples/sec	accuracy=0.536374	lr=0.064296
Epoch[89] Batch [1049]	Speed: 3421.938850 samples/sec	accuracy=0.536304	lr=0.064135
Epoch[89] Batch [1099]	Speed: 2482.848959 samples/sec	accuracy=0.536282	lr=0.063975
Epoch[89] Batch [1149]	Speed: 6004.888808 samples/sec	accuracy=0.536300	lr=0.063815
Epoch[89] Batch [1199]	Speed: 3298.722339 samples/sec	accuracy=0.536268	lr=0.063655
Epoch[89] Batch [1249]	Speed: 5726.319948 samples/sec	accuracy=0.536159	lr=0.063496
[Epoch 89] training: accuracy=0.536159
[Epoch 89] speed: 3799 samples/sec	time cost: 347.611667
[Epoch 89] validation: err-top1=0.431282 err-top5=0.195193
Epoch[90] Batch [49]	Speed: 5253.485345 samples/sec	accuracy=0.544199	lr=0.063333
Epoch[90] Batch [99]	Speed: 3443.697582 samples/sec	accuracy=0.542559	lr=0.063174
Epoch[90] Batch [149]	Speed: 5921.404586 samples/sec	accuracy=0.542845	lr=0.063015
Epoch[90] Batch [199]	Speed: 541.034218 samples/sec	accuracy=0.542314	lr=0.062855
Epoch[90] Batch [249]	Speed: 3483.608872 samples/sec	accuracy=0.541477	lr=0.062697
Epoch[90] Batch [299]	Speed: 3664.969393 samples/sec	accuracy=0.541634	lr=0.062538
Epoch[90] Batch [349]	Speed: 5653.664556 samples/sec	accuracy=0.541518	lr=0.062379
Epoch[90] Batch [399]	Speed: 3680.531555 samples/sec	accuracy=0.541318	lr=0.062221
Epoch[90] Batch [449]	Speed: 6074.035542 samples/sec	accuracy=0.541165	lr=0.062063
Epoch[90] Batch [499]	Speed: 3454.111181 samples/sec	accuracy=0.540559	lr=0.061905
Epoch[90] Batch [549]	Speed: 3794.704094 samples/sec	accuracy=0.540520	lr=0.061747
Epoch[90] Batch [599]	Speed: 3865.984162 samples/sec	accuracy=0.540314	lr=0.061589
Epoch[90] Batch [649]	Speed: 3598.482697 samples/sec	accuracy=0.540592	lr=0.061432
Epoch[90] Batch [699]	Speed: 2676.731196 samples/sec	accuracy=0.540248	lr=0.061274
Epoch[90] Batch [749]	Speed: 5359.523032 samples/sec	accuracy=0.540500	lr=0.061117
Epoch[90] Batch [799]	Speed: 2696.159120 samples/sec	accuracy=0.540287	lr=0.060960
Epoch[90] Batch [849]	Speed: 933.151096 samples/sec	accuracy=0.540191	lr=0.060803
Epoch[90] Batch [899]	Speed: 5345.810741 samples/sec	accuracy=0.540193	lr=0.060646
Epoch[90] Batch [949]	Speed: 5915.026270 samples/sec	accuracy=0.540191	lr=0.060490
Epoch[90] Batch [999]	Speed: 5680.745813 samples/sec	accuracy=0.539979	lr=0.060333
Epoch[90] Batch [1049]	Speed: 5894.438997 samples/sec	accuracy=0.539761	lr=0.060177
Epoch[90] Batch [1099]	Speed: 5472.282602 samples/sec	accuracy=0.539700	lr=0.060021
Epoch[90] Batch [1149]	Speed: 3766.610194 samples/sec	accuracy=0.539841	lr=0.059865
Epoch[90] Batch [1199]	Speed: 5566.318001 samples/sec	accuracy=0.539783	lr=0.059710
Epoch[90] Batch [1249]	Speed: 3548.402559 samples/sec	accuracy=0.539754	lr=0.059554
[Epoch 90] training: accuracy=0.539742
[Epoch 90] speed: 2983 samples/sec	time cost: 438.869548
[Epoch 90] validation: err-top1=0.446189 err-top5=0.205477
Epoch[91] Batch [49]	Speed: 5385.346745 samples/sec	accuracy=0.543848	lr=0.059396
Epoch[91] Batch [99]	Speed: 1070.815030 samples/sec	accuracy=0.544629	lr=0.059240
Epoch[91] Batch [149]	Speed: 1630.001659 samples/sec	accuracy=0.544115	lr=0.059085
Epoch[91] Batch [199]	Speed: 3398.288879 samples/sec	accuracy=0.545166	lr=0.058930
Epoch[91] Batch [249]	Speed: 5499.941999 samples/sec	accuracy=0.544836	lr=0.058776
Epoch[91] Batch [299]	Speed: 3810.444644 samples/sec	accuracy=0.544626	lr=0.058621
Epoch[91] Batch [349]	Speed: 5427.356593 samples/sec	accuracy=0.544263	lr=0.058467
Epoch[91] Batch [399]	Speed: 3617.520888 samples/sec	accuracy=0.544321	lr=0.058313
Epoch[91] Batch [449]	Speed: 5544.626822 samples/sec	accuracy=0.544013	lr=0.058158
Epoch[91] Batch [499]	Speed: 5812.131921 samples/sec	accuracy=0.543916	lr=0.058005
Epoch[91] Batch [549]	Speed: 3163.231606 samples/sec	accuracy=0.543770	lr=0.057851
Epoch[91] Batch [599]	Speed: 5822.101805 samples/sec	accuracy=0.543737	lr=0.057697
Epoch[91] Batch [649]	Speed: 5474.189917 samples/sec	accuracy=0.544046	lr=0.057544
Epoch[91] Batch [699]	Speed: 5052.317108 samples/sec	accuracy=0.543853	lr=0.057391
Epoch[91] Batch [749]	Speed: 5800.410156 samples/sec	accuracy=0.543673	lr=0.057238
Epoch[91] Batch [799]	Speed: 3858.660594 samples/sec	accuracy=0.543679	lr=0.057085
Epoch[91] Batch [849]	Speed: 2732.738446 samples/sec	accuracy=0.543688	lr=0.056932
Epoch[91] Batch [899]	Speed: 2149.813358 samples/sec	accuracy=0.543232	lr=0.056780
Epoch[91] Batch [949]	Speed: 5764.555063 samples/sec	accuracy=0.543057	lr=0.056627
Epoch[91] Batch [999]	Speed: 4042.326650 samples/sec	accuracy=0.542925	lr=0.056475
Epoch[91] Batch [1049]	Speed: 5531.149723 samples/sec	accuracy=0.542911	lr=0.056323
Epoch[91] Batch [1099]	Speed: 5444.712819 samples/sec	accuracy=0.542814	lr=0.056171
Epoch[91] Batch [1149]	Speed: 3684.177904 samples/sec	accuracy=0.542611	lr=0.056020
Epoch[91] Batch [1199]	Speed: 3769.846127 samples/sec	accuracy=0.542535	lr=0.055868
Epoch[91] Batch [1249]	Speed: 5629.589357 samples/sec	accuracy=0.542444	lr=0.055717
[Epoch 91] training: accuracy=0.542463
[Epoch 91] speed: 3612 samples/sec	time cost: 368.602293
[Epoch 91] validation: err-top1=0.425203 err-top5=0.190529
Epoch[92] Batch [49]	Speed: 4826.823054 samples/sec	accuracy=0.547188	lr=0.055563
Epoch[92] Batch [99]	Speed: 3716.647440 samples/sec	accuracy=0.548496	lr=0.055412
Epoch[92] Batch [149]	Speed: 4745.294092 samples/sec	accuracy=0.547422	lr=0.055261
Epoch[92] Batch [199]	Speed: 5397.181953 samples/sec	accuracy=0.547622	lr=0.055111
Epoch[92] Batch [249]	Speed: 5897.211138 samples/sec	accuracy=0.546891	lr=0.054960
Epoch[92] Batch [299]	Speed: 5799.358930 samples/sec	accuracy=0.546432	lr=0.054810
Epoch[92] Batch [349]	Speed: 3828.861595 samples/sec	accuracy=0.546336	lr=0.054660
Epoch[92] Batch [399]	Speed: 2746.186494 samples/sec	accuracy=0.546150	lr=0.054510
Epoch[92] Batch [449]	Speed: 5945.138942 samples/sec	accuracy=0.546105	lr=0.054360
Epoch[92] Batch [499]	Speed: 5589.076194 samples/sec	accuracy=0.546227	lr=0.054210
Epoch[92] Batch [549]	Speed: 3728.155141 samples/sec	accuracy=0.546122	lr=0.054061
Epoch[92] Batch [599]	Speed: 5669.065418 samples/sec	accuracy=0.546095	lr=0.053912
Epoch[92] Batch [649]	Speed: 5714.259537 samples/sec	accuracy=0.545944	lr=0.053763
Epoch[92] Batch [699]	Speed: 6117.174648 samples/sec	accuracy=0.545908	lr=0.053614
Epoch[92] Batch [749]	Speed: 3109.379490 samples/sec	accuracy=0.545823	lr=0.053465
Epoch[92] Batch [799]	Speed: 5460.757838 samples/sec	accuracy=0.545646	lr=0.053317
Epoch[92] Batch [849]	Speed: 2790.759378 samples/sec	accuracy=0.545568	lr=0.053168
Epoch[92] Batch [899]	Speed: 4371.553326 samples/sec	accuracy=0.545382	lr=0.053020
Epoch[92] Batch [949]	Speed: 5637.701031 samples/sec	accuracy=0.545213	lr=0.052872
Epoch[92] Batch [999]	Speed: 3608.897601 samples/sec	accuracy=0.545278	lr=0.052724
Epoch[92] Batch [1049]	Speed: 3047.315770 samples/sec	accuracy=0.545118	lr=0.052577
Epoch[92] Batch [1099]	Speed: 6063.192665 samples/sec	accuracy=0.545068	lr=0.052429
Epoch[92] Batch [1149]	Speed: 3695.660064 samples/sec	accuracy=0.544772	lr=0.052282
Epoch[92] Batch [1199]	Speed: 5335.537019 samples/sec	accuracy=0.544806	lr=0.052135
Epoch[92] Batch [1249]	Speed: 3439.585401 samples/sec	accuracy=0.544821	lr=0.051988
[Epoch 92] training: accuracy=0.544823
[Epoch 92] speed: 4338 samples/sec	time cost: 309.374053
[Epoch 92] validation: err-top1=0.418806 err-top5=0.185049
Epoch[93] Batch [49]	Speed: 5169.114365 samples/sec	accuracy=0.551387	lr=0.051838
Epoch[93] Batch [99]	Speed: 5575.564804 samples/sec	accuracy=0.553027	lr=0.051691
Epoch[93] Batch [149]	Speed: 4712.020089 samples/sec	accuracy=0.551973	lr=0.051545
Epoch[93] Batch [199]	Speed: 5607.211731 samples/sec	accuracy=0.551167	lr=0.051399
Epoch[93] Batch [249]	Speed: 2725.893532 samples/sec	accuracy=0.550898	lr=0.051253
Epoch[93] Batch [299]	Speed: 3852.356636 samples/sec	accuracy=0.550378	lr=0.051107
Epoch[93] Batch [349]	Speed: 3842.654987 samples/sec	accuracy=0.549780	lr=0.050961
Epoch[93] Batch [399]	Speed: 2339.504823 samples/sec	accuracy=0.549788	lr=0.050816
Epoch[93] Batch [449]	Speed: 3723.982793 samples/sec	accuracy=0.549944	lr=0.050670
Epoch[93] Batch [499]	Speed: 2755.744498 samples/sec	accuracy=0.549369	lr=0.050525
Epoch[93] Batch [549]	Speed: 3663.731805 samples/sec	accuracy=0.549260	lr=0.050380
Epoch[93] Batch [599]	Speed: 5536.419017 samples/sec	accuracy=0.549224	lr=0.050235
Epoch[93] Batch [649]	Speed: 5588.176946 samples/sec	accuracy=0.549081	lr=0.050091
Epoch[93] Batch [699]	Speed: 5362.007281 samples/sec	accuracy=0.549290	lr=0.049946
Epoch[93] Batch [749]	Speed: 3765.699378 samples/sec	accuracy=0.549027	lr=0.049802
Epoch[93] Batch [799]	Speed: 1935.573465 samples/sec	accuracy=0.549012	lr=0.049658
Epoch[93] Batch [849]	Speed: 3686.725372 samples/sec	accuracy=0.548788	lr=0.049514
Epoch[93] Batch [899]	Speed: 2685.426750 samples/sec	accuracy=0.548528	lr=0.049370
Epoch[93] Batch [949]	Speed: 5676.201440 samples/sec	accuracy=0.548514	lr=0.049227
Epoch[93] Batch [999]	Speed: 1329.613562 samples/sec	accuracy=0.548448	lr=0.049083
Epoch[93] Batch [1049]	Speed: 2638.273875 samples/sec	accuracy=0.548378	lr=0.048940
Epoch[93] Batch [1099]	Speed: 5862.576844 samples/sec	accuracy=0.548436	lr=0.048797
Epoch[93] Batch [1149]	Speed: 5700.331914 samples/sec	accuracy=0.548124	lr=0.048654
Epoch[93] Batch [1199]	Speed: 6018.342894 samples/sec	accuracy=0.547889	lr=0.048511
Epoch[93] Batch [1249]	Speed: 3524.688093 samples/sec	accuracy=0.547738	lr=0.048369
[Epoch 93] training: accuracy=0.547738
[Epoch 93] speed: 3539 samples/sec	time cost: 371.768600
[Epoch 93] validation: err-top1=0.463888 err-top5=0.223002
Epoch[94] Batch [49]	Speed: 4199.342473 samples/sec	accuracy=0.554004	lr=0.048224
Epoch[94] Batch [99]	Speed: 5810.322697 samples/sec	accuracy=0.554482	lr=0.048082
Epoch[94] Batch [149]	Speed: 3513.278844 samples/sec	accuracy=0.555931	lr=0.047940
Epoch[94] Batch [199]	Speed: 5420.633093 samples/sec	accuracy=0.555161	lr=0.047798
Epoch[94] Batch [249]	Speed: 3313.834589 samples/sec	accuracy=0.554813	lr=0.047656
Epoch[94] Batch [299]	Speed: 5931.713560 samples/sec	accuracy=0.554502	lr=0.047515
Epoch[94] Batch [349]	Speed: 5596.451684 samples/sec	accuracy=0.553948	lr=0.047374
Epoch[94] Batch [399]	Speed: 3722.295486 samples/sec	accuracy=0.553774	lr=0.047233
Epoch[94] Batch [449]	Speed: 5828.441268 samples/sec	accuracy=0.553637	lr=0.047092
Epoch[94] Batch [499]	Speed: 4149.342952 samples/sec	accuracy=0.553350	lr=0.046951
Epoch[94] Batch [549]	Speed: 5925.943842 samples/sec	accuracy=0.553015	lr=0.046811
Epoch[94] Batch [599]	Speed: 5529.684313 samples/sec	accuracy=0.553083	lr=0.046670
Epoch[94] Batch [649]	Speed: 5528.994106 samples/sec	accuracy=0.552743	lr=0.046530
Epoch[94] Batch [699]	Speed: 3547.198194 samples/sec	accuracy=0.552210	lr=0.046390
Epoch[94] Batch [749]	Speed: 5383.593005 samples/sec	accuracy=0.552052	lr=0.046251
Epoch[94] Batch [799]	Speed: 2754.343319 samples/sec	accuracy=0.552133	lr=0.046111
Epoch[94] Batch [849]	Speed: 2201.225467 samples/sec	accuracy=0.551886	lr=0.045972
Epoch[94] Batch [899]	Speed: 2382.596085 samples/sec	accuracy=0.551606	lr=0.045833
Epoch[94] Batch [949]	Speed: 3859.905047 samples/sec	accuracy=0.551600	lr=0.045693
Epoch[94] Batch [999]	Speed: 2708.355738 samples/sec	accuracy=0.551481	lr=0.045555
Epoch[94] Batch [1049]	Speed: 4307.370974 samples/sec	accuracy=0.551389	lr=0.045416
Epoch[94] Batch [1099]	Speed: 3886.347122 samples/sec	accuracy=0.551318	lr=0.045278
Epoch[94] Batch [1149]	Speed: 3673.363325 samples/sec	accuracy=0.551260	lr=0.045139
Epoch[94] Batch [1199]	Speed: 5585.466555 samples/sec	accuracy=0.551379	lr=0.045001
Epoch[94] Batch [1249]	Speed: 3624.041652 samples/sec	accuracy=0.551272	lr=0.044863
[Epoch 94] training: accuracy=0.551268
[Epoch 94] speed: 3971 samples/sec	time cost: 337.309268
[Epoch 94] validation: err-top1=0.431501 err-top5=0.194475
Epoch[95] Batch [49]	Speed: 5440.065645 samples/sec	accuracy=0.559980	lr=0.044723
Epoch[95] Batch [99]	Speed: 5593.849974 samples/sec	accuracy=0.556729	lr=0.044585
Epoch[95] Batch [149]	Speed: 3601.301583 samples/sec	accuracy=0.555117	lr=0.044448
Epoch[95] Batch [199]	Speed: 5126.218603 samples/sec	accuracy=0.555518	lr=0.044311
Epoch[95] Batch [249]	Speed: 5653.732727 samples/sec	accuracy=0.555586	lr=0.044174
Epoch[95] Batch [299]	Speed: 2673.244789 samples/sec	accuracy=0.555371	lr=0.044037
Epoch[95] Batch [349]	Speed: 1337.404093 samples/sec	accuracy=0.555374	lr=0.043900
Epoch[95] Batch [399]	Speed: 2096.317424 samples/sec	accuracy=0.555046	lr=0.043764
Epoch[95] Batch [449]	Speed: 3750.027805 samples/sec	accuracy=0.555091	lr=0.043628
Epoch[95] Batch [499]	Speed: 5609.120952 samples/sec	accuracy=0.555166	lr=0.043492
Epoch[95] Batch [549]	Speed: 3743.875880 samples/sec	accuracy=0.554728	lr=0.043356
Epoch[95] Batch [599]	Speed: 2677.428254 samples/sec	accuracy=0.554422	lr=0.043220
Epoch[95] Batch [649]	Speed: 4280.552023 samples/sec	accuracy=0.553924	lr=0.043085
Epoch[95] Batch [699]	Speed: 4429.931448 samples/sec	accuracy=0.553929	lr=0.042949
Epoch[95] Batch [749]	Speed: 2738.948514 samples/sec	accuracy=0.554012	lr=0.042814
Epoch[95] Batch [799]	Speed: 3570.915951 samples/sec	accuracy=0.553859	lr=0.042679
Epoch[95] Batch [849]	Speed: 6082.674437 samples/sec	accuracy=0.553921	lr=0.042545
Epoch[95] Batch [899]	Speed: 5410.316483 samples/sec	accuracy=0.554025	lr=0.042410
Epoch[95] Batch [949]	Speed: 3703.014187 samples/sec	accuracy=0.553747	lr=0.042276
Epoch[95] Batch [999]	Speed: 3668.377025 samples/sec	accuracy=0.553496	lr=0.042141
Epoch[95] Batch [1049]	Speed: 5572.188013 samples/sec	accuracy=0.553252	lr=0.042007
Epoch[95] Batch [1099]	Speed: 5802.753642 samples/sec	accuracy=0.553359	lr=0.041874
Epoch[95] Batch [1149]	Speed: 3684.714024 samples/sec	accuracy=0.553278	lr=0.041740
Epoch[95] Batch [1199]	Speed: 5478.196911 samples/sec	accuracy=0.553162	lr=0.041607
Epoch[95] Batch [1249]	Speed: 2703.867799 samples/sec	accuracy=0.553059	lr=0.041473
[Epoch 95] training: accuracy=0.553043
[Epoch 95] speed: 3636 samples/sec	time cost: 362.464822
[Epoch 95] validation: err-top1=0.432956 err-top5=0.194117
Epoch[96] Batch [49]	Speed: 4027.428371 samples/sec	accuracy=0.558340	lr=0.041338
Epoch[96] Batch [99]	Speed: 5288.650194 samples/sec	accuracy=0.560645	lr=0.041205
Epoch[96] Batch [149]	Speed: 5842.371979 samples/sec	accuracy=0.558906	lr=0.041072
Epoch[96] Batch [199]	Speed: 5895.159865 samples/sec	accuracy=0.559097	lr=0.040940
Epoch[96] Batch [249]	Speed: 2639.925180 samples/sec	accuracy=0.558957	lr=0.040807
Epoch[96] Batch [299]	Speed: 4434.505302 samples/sec	accuracy=0.558792	lr=0.040675
Epoch[96] Batch [349]	Speed: 2800.273090 samples/sec	accuracy=0.558496	lr=0.040543
Epoch[96] Batch [399]	Speed: 4485.346388 samples/sec	accuracy=0.558333	lr=0.040412
Epoch[96] Batch [449]	Speed: 5662.239771 samples/sec	accuracy=0.557791	lr=0.040280
Epoch[96] Batch [499]	Speed: 6024.174215 samples/sec	accuracy=0.557740	lr=0.040149
Epoch[96] Batch [549]	Speed: 3489.448434 samples/sec	accuracy=0.557624	lr=0.040018
Epoch[96] Batch [599]	Speed: 4342.281753 samples/sec	accuracy=0.557422	lr=0.039887
Epoch[96] Batch [649]	Speed: 3133.425648 samples/sec	accuracy=0.557300	lr=0.039756
Epoch[96] Batch [699]	Speed: 5518.872679 samples/sec	accuracy=0.557259	lr=0.039625
Epoch[96] Batch [749]	Speed: 1543.231872 samples/sec	accuracy=0.557048	lr=0.039495
Epoch[96] Batch [799]	Speed: 3205.081269 samples/sec	accuracy=0.556947	lr=0.039365
Epoch[96] Batch [849]	Speed: 5687.661669 samples/sec	accuracy=0.557024	lr=0.039235
Epoch[96] Batch [899]	Speed: 4547.606771 samples/sec	accuracy=0.556682	lr=0.039105
Epoch[96] Batch [949]	Speed: 3462.639932 samples/sec	accuracy=0.556587	lr=0.038975
Epoch[96] Batch [999]	Speed: 2749.638294 samples/sec	accuracy=0.556459	lr=0.038846
Epoch[96] Batch [1049]	Speed: 5572.601122 samples/sec	accuracy=0.556567	lr=0.038717
Epoch[96] Batch [1099]	Speed: 3766.089344 samples/sec	accuracy=0.556554	lr=0.038588
Epoch[96] Batch [1149]	Speed: 3949.026712 samples/sec	accuracy=0.556508	lr=0.038459
Epoch[96] Batch [1199]	Speed: 5423.527697 samples/sec	accuracy=0.556420	lr=0.038330
Epoch[96] Batch [1249]	Speed: 3568.288901 samples/sec	accuracy=0.556466	lr=0.038202
[Epoch 96] training: accuracy=0.556470
[Epoch 96] speed: 3847 samples/sec	time cost: 356.729524
[Epoch 96] validation: err-top1=0.424805 err-top5=0.190091
Epoch[97] Batch [49]	Speed: 4368.088575 samples/sec	accuracy=0.564355	lr=0.038071
Epoch[97] Batch [99]	Speed: 3309.707082 samples/sec	accuracy=0.562344	lr=0.037943
Epoch[97] Batch [149]	Speed: 4386.625485 samples/sec	accuracy=0.562598	lr=0.037815
Epoch[97] Batch [199]	Speed: 5860.545592 samples/sec	accuracy=0.561792	lr=0.037687
Epoch[97] Batch [249]	Speed: 5682.443202 samples/sec	accuracy=0.561789	lr=0.037560
Epoch[97] Batch [299]	Speed: 1056.780617 samples/sec	accuracy=0.560885	lr=0.037432
Epoch[97] Batch [349]	Speed: 5920.007282 samples/sec	accuracy=0.560265	lr=0.037305
Epoch[97] Batch [399]	Speed: 3774.379112 samples/sec	accuracy=0.560198	lr=0.037178
Epoch[97] Batch [449]	Speed: 5409.447670 samples/sec	accuracy=0.560681	lr=0.037052
Epoch[97] Batch [499]	Speed: 2572.767985 samples/sec	accuracy=0.560398	lr=0.036925
Epoch[97] Batch [549]	Speed: 5598.501139 samples/sec	accuracy=0.560309	lr=0.036799
Epoch[97] Batch [599]	Speed: 5032.226879 samples/sec	accuracy=0.560114	lr=0.036673
Epoch[97] Batch [649]	Speed: 2181.506622 samples/sec	accuracy=0.560152	lr=0.036547
Epoch[97] Batch [699]	Speed: 3678.165939 samples/sec	accuracy=0.560113	lr=0.036421
Epoch[97] Batch [749]	Speed: 2823.217921 samples/sec	accuracy=0.560016	lr=0.036296
Epoch[97] Batch [799]	Speed: 3688.424353 samples/sec	accuracy=0.559829	lr=0.036170
Epoch[97] Batch [849]	Speed: 1385.783200 samples/sec	accuracy=0.559643	lr=0.036045
Epoch[97] Batch [899]	Speed: 5234.592449 samples/sec	accuracy=0.559480	lr=0.035920
Epoch[97] Batch [949]	Speed: 3752.757224 samples/sec	accuracy=0.559476	lr=0.035795
Epoch[97] Batch [999]	Speed: 3605.709466 samples/sec	accuracy=0.559536	lr=0.035671
Epoch[97] Batch [1049]	Speed: 3671.578187 samples/sec	accuracy=0.559434	lr=0.035546
Epoch[97] Batch [1099]	Speed: 3824.949677 samples/sec	accuracy=0.559469	lr=0.035422
Epoch[97] Batch [1149]	Speed: 2888.013985 samples/sec	accuracy=0.559452	lr=0.035298
Epoch[97] Batch [1199]	Speed: 4943.422412 samples/sec	accuracy=0.559464	lr=0.035174
Epoch[97] Batch [1249]	Speed: 743.609457 samples/sec	accuracy=0.559462	lr=0.035051
[Epoch 97] training: accuracy=0.559469
[Epoch 97] speed: 2880 samples/sec	time cost: 458.880728
[Epoch 97] validation: err-top1=0.427097 err-top5=0.191346
Epoch[98] Batch [49]	Speed: 2784.029931 samples/sec	accuracy=0.566445	lr=0.034925
Epoch[98] Batch [99]	Speed: 3626.237224 samples/sec	accuracy=0.566475	lr=0.034802
Epoch[98] Batch [149]	Speed: 2807.488111 samples/sec	accuracy=0.565729	lr=0.034679
Epoch[98] Batch [199]	Speed: 3761.698464 samples/sec	accuracy=0.564141	lr=0.034556
Epoch[98] Batch [249]	Speed: 5840.579146 samples/sec	accuracy=0.563676	lr=0.034433
Epoch[98] Batch [299]	Speed: 2690.181493 samples/sec	accuracy=0.563236	lr=0.034311
Epoch[98] Batch [349]	Speed: 5433.468182 samples/sec	accuracy=0.563019	lr=0.034189
Epoch[98] Batch [399]	Speed: 3771.284273 samples/sec	accuracy=0.563193	lr=0.034067
Epoch[98] Batch [449]	Speed: 2712.159586 samples/sec	accuracy=0.563166	lr=0.033945
Epoch[98] Batch [499]	Speed: 5892.134858 samples/sec	accuracy=0.562920	lr=0.033823
Epoch[98] Batch [549]	Speed: 5901.583937 samples/sec	accuracy=0.562923	lr=0.033702
Epoch[98] Batch [599]	Speed: 2497.847325 samples/sec	accuracy=0.562912	lr=0.033581
Epoch[98] Batch [649]	Speed: 3360.559126 samples/sec	accuracy=0.563071	lr=0.033460
Epoch[98] Batch [699]	Speed: 5518.616969 samples/sec	accuracy=0.562778	lr=0.033339
Epoch[98] Batch [749]	Speed: 5625.294261 samples/sec	accuracy=0.562664	lr=0.033218
Epoch[98] Batch [799]	Speed: 3636.890569 samples/sec	accuracy=0.562616	lr=0.033098
Epoch[98] Batch [849]	Speed: 4261.442528 samples/sec	accuracy=0.562638	lr=0.032978
Epoch[98] Batch [899]	Speed: 5699.104290 samples/sec	accuracy=0.562478	lr=0.032858
Epoch[98] Batch [949]	Speed: 5953.469623 samples/sec	accuracy=0.562436	lr=0.032738
Epoch[98] Batch [999]	Speed: 2685.502948 samples/sec	accuracy=0.562424	lr=0.032618
Epoch[98] Batch [1049]	Speed: 3557.625621 samples/sec	accuracy=0.562471	lr=0.032499
Epoch[98] Batch [1099]	Speed: 5767.066654 samples/sec	accuracy=0.562310	lr=0.032380
Epoch[98] Batch [1149]	Speed: 3588.818877 samples/sec	accuracy=0.562189	lr=0.032261
Epoch[98] Batch [1199]	Speed: 5386.918521 samples/sec	accuracy=0.562222	lr=0.032142
Epoch[98] Batch [1249]	Speed: 3669.636377 samples/sec	accuracy=0.562060	lr=0.032023
[Epoch 98] training: accuracy=0.562062
[Epoch 98] speed: 3884 samples/sec	time cost: 348.569193
[Epoch 98] validation: err-top1=0.408264 err-top5=0.177592
Epoch[99] Batch [49]	Speed: 3923.416175 samples/sec	accuracy=0.567754	lr=0.031902
Epoch[99] Batch [99]	Speed: 3877.556203 samples/sec	accuracy=0.566777	lr=0.031784
Epoch[99] Batch [149]	Speed: 5761.948569 samples/sec	accuracy=0.567949	lr=0.031666
Epoch[99] Batch [199]	Speed: 3843.335139 samples/sec	accuracy=0.567852	lr=0.031548
Epoch[99] Batch [249]	Speed: 3531.052714 samples/sec	accuracy=0.567754	lr=0.031431
Epoch[99] Batch [299]	Speed: 3658.658589 samples/sec	accuracy=0.567868	lr=0.031313
Epoch[99] Batch [349]	Speed: 3717.104260 samples/sec	accuracy=0.567843	lr=0.031196
Epoch[99] Batch [399]	Speed: 3708.551545 samples/sec	accuracy=0.567290	lr=0.031079
Epoch[99] Batch [449]	Speed: 5476.714308 samples/sec	accuracy=0.566842	lr=0.030962
Epoch[99] Batch [499]	Speed: 2380.003333 samples/sec	accuracy=0.566840	lr=0.030846
Epoch[99] Batch [549]	Speed: 3691.100964 samples/sec	accuracy=0.566676	lr=0.030729
Epoch[99] Batch [599]	Speed: 4976.319140 samples/sec	accuracy=0.567017	lr=0.030613
Epoch[99] Batch [649]	Speed: 3617.278125 samples/sec	accuracy=0.566920	lr=0.030497
Epoch[99] Batch [699]	Speed: 3776.236224 samples/sec	accuracy=0.566641	lr=0.030381
Epoch[99] Batch [749]	Speed: 5767.857242 samples/sec	accuracy=0.566520	lr=0.030266
Epoch[99] Batch [799]	Speed: 5598.222528 samples/sec	accuracy=0.566290	lr=0.030150
Epoch[99] Batch [849]	Speed: 3668.948234 samples/sec	accuracy=0.566406	lr=0.030035
Epoch[99] Batch [899]	Speed: 3592.252392 samples/sec	accuracy=0.566135	lr=0.029920
Epoch[99] Batch [949]	Speed: 4889.420832 samples/sec	accuracy=0.566061	lr=0.029805
Epoch[99] Batch [999]	Speed: 5867.169944 samples/sec	accuracy=0.566141	lr=0.029691
Epoch[99] Batch [1049]	Speed: 5439.300498 samples/sec	accuracy=0.566167	lr=0.029576
Epoch[99] Batch [1099]	Speed: 3783.323757 samples/sec	accuracy=0.566091	lr=0.029462
Epoch[99] Batch [1149]	Speed: 5864.443585 samples/sec	accuracy=0.566050	lr=0.029348
Epoch[99] Batch [1199]	Speed: 3321.826523 samples/sec	accuracy=0.565921	lr=0.029234
Epoch[99] Batch [1249]	Speed: 3646.022784 samples/sec	accuracy=0.565603	lr=0.029121
[Epoch 99] training: accuracy=0.565582
[Epoch 99] speed: 4074 samples/sec	time cost: 333.648462
[Epoch 99] validation: err-top1=0.403280 err-top5=0.174207
Epoch[100] Batch [49]	Speed: 2692.623978 samples/sec	accuracy=0.568867	lr=0.029005
Epoch[100] Batch [99]	Speed: 3638.969346 samples/sec	accuracy=0.571357	lr=0.028892
Epoch[100] Batch [149]	Speed: 3752.381750 samples/sec	accuracy=0.572474	lr=0.028779
Epoch[100] Batch [199]	Speed: 3347.040298 samples/sec	accuracy=0.570962	lr=0.028666
Epoch[100] Batch [249]	Speed: 5841.977344 samples/sec	accuracy=0.570125	lr=0.028554
Epoch[100] Batch [299]	Speed: 3651.476564 samples/sec	accuracy=0.569561	lr=0.028441
Epoch[100] Batch [349]	Speed: 3608.377434 samples/sec	accuracy=0.569328	lr=0.028329
Epoch[100] Batch [399]	Speed: 3554.909506 samples/sec	accuracy=0.569270	lr=0.028217
Epoch[100] Batch [449]	Speed: 5589.014374 samples/sec	accuracy=0.569115	lr=0.028105
Epoch[100] Batch [499]	Speed: 5797.484869 samples/sec	accuracy=0.569014	lr=0.027994
Epoch[100] Batch [549]	Speed: 5881.880553 samples/sec	accuracy=0.569180	lr=0.027883
Epoch[100] Batch [599]	Speed: 5780.502742 samples/sec	accuracy=0.569115	lr=0.027772
Epoch[100] Batch [649]	Speed: 5727.847140 samples/sec	accuracy=0.569313	lr=0.027661
Epoch[100] Batch [699]	Speed: 2762.607844 samples/sec	accuracy=0.569275	lr=0.027550
Epoch[100] Batch [749]	Speed: 3672.368486 samples/sec	accuracy=0.569241	lr=0.027439
Epoch[100] Batch [799]	Speed: 5774.913970 samples/sec	accuracy=0.569011	lr=0.027329
Epoch[100] Batch [849]	Speed: 2143.145472 samples/sec	accuracy=0.569027	lr=0.027219
Epoch[100] Batch [899]	Speed: 5800.439767 samples/sec	accuracy=0.568889	lr=0.027109
Epoch[100] Batch [949]	Speed: 3654.937082 samples/sec	accuracy=0.568795	lr=0.026999
Epoch[100] Batch [999]	Speed: 2720.521789 samples/sec	accuracy=0.568869	lr=0.026890
Epoch[100] Batch [1049]	Speed: 2665.091611 samples/sec	accuracy=0.568890	lr=0.026781
Epoch[100] Batch [1099]	Speed: 4724.829444 samples/sec	accuracy=0.568825	lr=0.026672
Epoch[100] Batch [1149]	Speed: 1723.526083 samples/sec	accuracy=0.568801	lr=0.026563
Epoch[100] Batch [1199]	Speed: 3524.666746 samples/sec	accuracy=0.568641	lr=0.026454
Epoch[100] Batch [1249]	Speed: 4336.715818 samples/sec	accuracy=0.568626	lr=0.026346
[Epoch 100] training: accuracy=0.568625
[Epoch 100] speed: 3646 samples/sec	time cost: 370.868830
[Epoch 100] validation: err-top1=0.417431 err-top5=0.185527
Epoch[101] Batch [49]	Speed: 5624.893193 samples/sec	accuracy=0.578340	lr=0.026235
Epoch[101] Batch [99]	Speed: 2774.615974 samples/sec	accuracy=0.576982	lr=0.026127
Epoch[101] Batch [149]	Speed: 5708.220671 samples/sec	accuracy=0.576803	lr=0.026020
Epoch[101] Batch [199]	Speed: 3626.935655 samples/sec	accuracy=0.574644	lr=0.025912
Epoch[101] Batch [249]	Speed: 2715.316362 samples/sec	accuracy=0.573820	lr=0.025805
Epoch[101] Batch [299]	Speed: 4939.242026 samples/sec	accuracy=0.573031	lr=0.025697
Epoch[101] Batch [349]	Speed: 3634.386141 samples/sec	accuracy=0.573111	lr=0.025590
Epoch[101] Batch [399]	Speed: 5794.862121 samples/sec	accuracy=0.573074	lr=0.025484
Epoch[101] Batch [449]	Speed: 3615.396903 samples/sec	accuracy=0.572934	lr=0.025377
Epoch[101] Batch [499]	Speed: 5859.245921 samples/sec	accuracy=0.572592	lr=0.025271
Epoch[101] Batch [549]	Speed: 3617.511747 samples/sec	accuracy=0.572358	lr=0.025165
Epoch[101] Batch [599]	Speed: 6023.385634 samples/sec	accuracy=0.572380	lr=0.025059
Epoch[101] Batch [649]	Speed: 5732.512109 samples/sec	accuracy=0.572028	lr=0.024953
Epoch[101] Batch [699]	Speed: 2952.070530 samples/sec	accuracy=0.572091	lr=0.024847
Epoch[101] Batch [749]	Speed: 3637.376540 samples/sec	accuracy=0.572020	lr=0.024742
Epoch[101] Batch [799]	Speed: 2846.683974 samples/sec	accuracy=0.571869	lr=0.024637
Epoch[101] Batch [849]	Speed: 5745.686329 samples/sec	accuracy=0.571837	lr=0.024532
Epoch[101] Batch [899]	Speed: 2692.004193 samples/sec	accuracy=0.571627	lr=0.024427
Epoch[101] Batch [949]	Speed: 5283.483103 samples/sec	accuracy=0.571522	lr=0.024323
Epoch[101] Batch [999]	Speed: 2695.307988 samples/sec	accuracy=0.571539	lr=0.024219
Epoch[101] Batch [1049]	Speed: 5388.259341 samples/sec	accuracy=0.571677	lr=0.024115
Epoch[101] Batch [1099]	Speed: 3671.517925 samples/sec	accuracy=0.571612	lr=0.024011
Epoch[101] Batch [1149]	Speed: 2126.523947 samples/sec	accuracy=0.571620	lr=0.023907
Epoch[101] Batch [1199]	Speed: 5522.362080 samples/sec	accuracy=0.571671	lr=0.023804
Epoch[101] Batch [1249]	Speed: 3725.176579 samples/sec	accuracy=0.571633	lr=0.023700
[Epoch 101] training: accuracy=0.571640
[Epoch 101] speed: 3830 samples/sec	time cost: 353.287133
[Epoch 101] validation: err-top1=0.401168 err-top5=0.173569
Epoch[102] Batch [49]	Speed: 3600.344844 samples/sec	accuracy=0.581270	lr=0.023595
Epoch[102] Batch [99]	Speed: 2728.452457 samples/sec	accuracy=0.577783	lr=0.023493
Epoch[102] Batch [149]	Speed: 2916.851745 samples/sec	accuracy=0.578340	lr=0.023390
Epoch[102] Batch [199]	Speed: 3666.508210 samples/sec	accuracy=0.578662	lr=0.023288
Epoch[102] Batch [249]	Speed: 3480.900624 samples/sec	accuracy=0.577391	lr=0.023185
Epoch[102] Batch [299]	Speed: 5850.801407 samples/sec	accuracy=0.577018	lr=0.023083
Epoch[102] Batch [349]	Speed: 3543.574330 samples/sec	accuracy=0.576521	lr=0.022982
Epoch[102] Batch [399]	Speed: 3520.268785 samples/sec	accuracy=0.576714	lr=0.022880
Epoch[102] Batch [449]	Speed: 4617.969908 samples/sec	accuracy=0.576608	lr=0.022779
Epoch[102] Batch [499]	Speed: 5558.430009 samples/sec	accuracy=0.576436	lr=0.022678
Epoch[102] Batch [549]	Speed: 2046.820719 samples/sec	accuracy=0.576399	lr=0.022577
Epoch[102] Batch [599]	Speed: 4381.957746 samples/sec	accuracy=0.576323	lr=0.022476
Epoch[102] Batch [649]	Speed: 5541.404498 samples/sec	accuracy=0.576276	lr=0.022376
Epoch[102] Batch [699]	Speed: 2543.665472 samples/sec	accuracy=0.576091	lr=0.022276
Epoch[102] Batch [749]	Speed: 2672.846253 samples/sec	accuracy=0.576073	lr=0.022175
Epoch[102] Batch [799]	Speed: 5774.988824 samples/sec	accuracy=0.575989	lr=0.022076
Epoch[102] Batch [849]	Speed: 2688.871023 samples/sec	accuracy=0.575956	lr=0.021976
Epoch[102] Batch [899]	Speed: 3601.726198 samples/sec	accuracy=0.576005	lr=0.021877
Epoch[102] Batch [949]	Speed: 6044.376073 samples/sec	accuracy=0.575866	lr=0.021777
Epoch[102] Batch [999]	Speed: 5610.613381 samples/sec	accuracy=0.575608	lr=0.021678
Epoch[102] Batch [1049]	Speed: 3666.515284 samples/sec	accuracy=0.575472	lr=0.021580
Epoch[102] Batch [1099]	Speed: 5611.519719 samples/sec	accuracy=0.575401	lr=0.021481
Epoch[102] Batch [1149]	Speed: 5481.097937 samples/sec	accuracy=0.575145	lr=0.021383
Epoch[102] Batch [1199]	Speed: 5694.596046 samples/sec	accuracy=0.574963	lr=0.021285
Epoch[102] Batch [1249]	Speed: 3931.819206 samples/sec	accuracy=0.574963	lr=0.021187
[Epoch 102] training: accuracy=0.574953
[Epoch 102] speed: 3798 samples/sec	time cost: 351.616007
[Epoch 102] validation: err-top1=0.563277 err-top5=0.318280
Epoch[103] Batch [49]	Speed: 5564.455249 samples/sec	accuracy=0.584688	lr=0.021087
Epoch[103] Batch [99]	Speed: 2192.967325 samples/sec	accuracy=0.584365	lr=0.020990
Epoch[103] Batch [149]	Speed: 5651.519789 samples/sec	accuracy=0.582799	lr=0.020892
Epoch[103] Batch [199]	Speed: 6005.750149 samples/sec	accuracy=0.581885	lr=0.020795
Epoch[103] Batch [249]	Speed: 5742.843608 samples/sec	accuracy=0.581332	lr=0.020698
Epoch[103] Batch [299]	Speed: 6011.507773 samples/sec	accuracy=0.580651	lr=0.020602
Epoch[103] Batch [349]	Speed: 2743.841489 samples/sec	accuracy=0.580092	lr=0.020505
Epoch[103] Batch [399]	Speed: 5727.755477 samples/sec	accuracy=0.579746	lr=0.020409
Epoch[103] Batch [449]	Speed: 5924.623015 samples/sec	accuracy=0.579390	lr=0.020313
Epoch[103] Batch [499]	Speed: 5632.628313 samples/sec	accuracy=0.579178	lr=0.020217
Epoch[103] Batch [549]	Speed: 4087.519236 samples/sec	accuracy=0.579309	lr=0.020122
Epoch[103] Batch [599]	Speed: 2706.448517 samples/sec	accuracy=0.579339	lr=0.020026
Epoch[103] Batch [649]	Speed: 3684.634174 samples/sec	accuracy=0.579133	lr=0.019931
Epoch[103] Batch [699]	Speed: 3665.137028 samples/sec	accuracy=0.578887	lr=0.019836
Epoch[103] Batch [749]	Speed: 3248.818598 samples/sec	accuracy=0.578952	lr=0.019742
Epoch[103] Batch [799]	Speed: 3685.989681 samples/sec	accuracy=0.578849	lr=0.019647
Epoch[103] Batch [849]	Speed: 3767.260651 samples/sec	accuracy=0.578724	lr=0.019553
Epoch[103] Batch [899]	Speed: 3674.879707 samples/sec	accuracy=0.578652	lr=0.019459
Epoch[103] Batch [949]	Speed: 3126.696161 samples/sec	accuracy=0.578568	lr=0.019365
Epoch[103] Batch [999]	Speed: 6027.344983 samples/sec	accuracy=0.578509	lr=0.019271
Epoch[103] Batch [1049]	Speed: 5493.027033 samples/sec	accuracy=0.578498	lr=0.019178
Epoch[103] Batch [1099]	Speed: 3620.940944 samples/sec	accuracy=0.578597	lr=0.019085
Epoch[103] Batch [1149]	Speed: 5719.398622 samples/sec	accuracy=0.578578	lr=0.018992
Epoch[103] Batch [1199]	Speed: 6044.735062 samples/sec	accuracy=0.578470	lr=0.018899
Epoch[103] Batch [1249]	Speed: 5575.700882 samples/sec	accuracy=0.578445	lr=0.018806
[Epoch 103] training: accuracy=0.578436
[Epoch 103] speed: 4202 samples/sec	time cost: 319.129276
[Epoch 103] validation: err-top1=0.387277 err-top5=0.161133
Epoch[104] Batch [49]	Speed: 5080.868639 samples/sec	accuracy=0.582559	lr=0.018712
Epoch[104] Batch [99]	Speed: 3824.124628 samples/sec	accuracy=0.583652	lr=0.018620
Epoch[104] Batch [149]	Speed: 3640.671124 samples/sec	accuracy=0.583516	lr=0.018528
Epoch[104] Batch [199]	Speed: 3618.817272 samples/sec	accuracy=0.583779	lr=0.018436
Epoch[104] Batch [249]	Speed: 3113.422878 samples/sec	accuracy=0.583359	lr=0.018345
Epoch[104] Batch [299]	Speed: 2935.731646 samples/sec	accuracy=0.582607	lr=0.018254
Epoch[104] Batch [349]	Speed: 3853.953812 samples/sec	accuracy=0.583083	lr=0.018163
Epoch[104] Batch [399]	Speed: 2672.314015 samples/sec	accuracy=0.583005	lr=0.018072
Epoch[104] Batch [449]	Speed: 5743.385937 samples/sec	accuracy=0.583053	lr=0.017981
Epoch[104] Batch [499]	Speed: 3798.960770 samples/sec	accuracy=0.582617	lr=0.017891
Epoch[104] Batch [549]	Speed: 3788.502679 samples/sec	accuracy=0.582564	lr=0.017801
Epoch[104] Batch [599]	Speed: 5660.943288 samples/sec	accuracy=0.582619	lr=0.017711
Epoch[104] Batch [649]	Speed: 5302.175827 samples/sec	accuracy=0.582435	lr=0.017621
Epoch[104] Batch [699]	Speed: 3931.462036 samples/sec	accuracy=0.582422	lr=0.017532
Epoch[104] Batch [749]	Speed: 5966.158933 samples/sec	accuracy=0.582409	lr=0.017442
Epoch[104] Batch [799]	Speed: 3740.437554 samples/sec	accuracy=0.582067	lr=0.017353
Epoch[104] Batch [849]	Speed: 5836.715342 samples/sec	accuracy=0.582005	lr=0.017264
Epoch[104] Batch [899]	Speed: 5817.885044 samples/sec	accuracy=0.581849	lr=0.017176
Epoch[104] Batch [949]	Speed: 5200.627945 samples/sec	accuracy=0.581762	lr=0.017087
Epoch[104] Batch [999]	Speed: 5402.270319 samples/sec	accuracy=0.581648	lr=0.016999
Epoch[104] Batch [1049]	Speed: 4493.278032 samples/sec	accuracy=0.581573	lr=0.016911
Epoch[104] Batch [1099]	Speed: 5766.810038 samples/sec	accuracy=0.581626	lr=0.016823
Epoch[104] Batch [1149]	Speed: 2672.957437 samples/sec	accuracy=0.581405	lr=0.016736
Epoch[104] Batch [1199]	Speed: 2713.695305 samples/sec	accuracy=0.581361	lr=0.016648
Epoch[104] Batch [1249]	Speed: 5945.691181 samples/sec	accuracy=0.581297	lr=0.016561
[Epoch 104] training: accuracy=0.581287
[Epoch 104] speed: 4101 samples/sec	time cost: 331.219645
[Epoch 104] validation: err-top1=0.373271 err-top5=0.153727
Epoch[105] Batch [49]	Speed: 3693.079153 samples/sec	accuracy=0.588516	lr=0.016473
Epoch[105] Batch [99]	Speed: 5972.246095 samples/sec	accuracy=0.587773	lr=0.016386
Epoch[105] Batch [149]	Speed: 2234.046377 samples/sec	accuracy=0.587520	lr=0.016299
Epoch[105] Batch [199]	Speed: 5593.457020 samples/sec	accuracy=0.587144	lr=0.016213
Epoch[105] Batch [249]	Speed: 2720.063899 samples/sec	accuracy=0.586156	lr=0.016127
Epoch[105] Batch [299]	Speed: 5024.711939 samples/sec	accuracy=0.585771	lr=0.016041
Epoch[105] Batch [349]	Speed: 4838.547532 samples/sec	accuracy=0.585262	lr=0.015956
Epoch[105] Batch [399]	Speed: 2800.056244 samples/sec	accuracy=0.585266	lr=0.015870
Epoch[105] Batch [449]	Speed: 5899.773711 samples/sec	accuracy=0.585414	lr=0.015785
Epoch[105] Batch [499]	Speed: 3499.901070 samples/sec	accuracy=0.585264	lr=0.015700
Epoch[105] Batch [549]	Speed: 5451.429520 samples/sec	accuracy=0.585265	lr=0.015616
Epoch[105] Batch [599]	Speed: 2678.192541 samples/sec	accuracy=0.585216	lr=0.015531
Epoch[105] Batch [649]	Speed: 2716.298811 samples/sec	accuracy=0.584713	lr=0.015447
Epoch[105] Batch [699]	Speed: 2709.870190 samples/sec	accuracy=0.584978	lr=0.015363
Epoch[105] Batch [749]	Speed: 3484.182604 samples/sec	accuracy=0.584780	lr=0.015279
Epoch[105] Batch [799]	Speed: 3752.465678 samples/sec	accuracy=0.584752	lr=0.015195
Epoch[105] Batch [849]	Speed: 3592.296258 samples/sec	accuracy=0.584709	lr=0.015112
Epoch[105] Batch [899]	Speed: 5831.844148 samples/sec	accuracy=0.584682	lr=0.015029
Epoch[105] Batch [949]	Speed: 1558.604440 samples/sec	accuracy=0.584353	lr=0.014946
Epoch[105] Batch [999]	Speed: 4445.527257 samples/sec	accuracy=0.584290	lr=0.014863
Epoch[105] Batch [1049]	Speed: 5656.748490 samples/sec	accuracy=0.584326	lr=0.014781
Epoch[105] Batch [1099]	Speed: 5655.331351 samples/sec	accuracy=0.584128	lr=0.014699
Epoch[105] Batch [1149]	Speed: 5437.448529 samples/sec	accuracy=0.584049	lr=0.014616
Epoch[105] Batch [1199]	Speed: 3722.468729 samples/sec	accuracy=0.584041	lr=0.014535
Epoch[105] Batch [1249]	Speed: 2339.928111 samples/sec	accuracy=0.584098	lr=0.014453
[Epoch 105] training: accuracy=0.584108
[Epoch 105] speed: 3544 samples/sec	time cost: 385.613245
[Epoch 105] validation: err-top1=0.369440 err-top5=0.150470
Epoch[106] Batch [49]	Speed: 3341.732697 samples/sec	accuracy=0.587148	lr=0.014370
Epoch[106] Batch [99]	Speed: 3502.307085 samples/sec	accuracy=0.586934	lr=0.014289
Epoch[106] Batch [149]	Speed: 2696.183864 samples/sec	accuracy=0.588620	lr=0.014208
Epoch[106] Batch [199]	Speed: 1793.284840 samples/sec	accuracy=0.588179	lr=0.014127
Epoch[106] Batch [249]	Speed: 1967.343560 samples/sec	accuracy=0.588785	lr=0.014047
Epoch[106] Batch [299]	Speed: 5380.088006 samples/sec	accuracy=0.587852	lr=0.013966
Epoch[106] Batch [349]	Speed: 4474.753867 samples/sec	accuracy=0.587762	lr=0.013886
Epoch[106] Batch [399]	Speed: 5558.638774 samples/sec	accuracy=0.588081	lr=0.013806
Epoch[106] Batch [449]	Speed: 2694.506788 samples/sec	accuracy=0.587884	lr=0.013727
Epoch[106] Batch [499]	Speed: 5899.619411 samples/sec	accuracy=0.587922	lr=0.013647
Epoch[106] Batch [549]	Speed: 4115.385455 samples/sec	accuracy=0.587988	lr=0.013568
Epoch[106] Batch [599]	Speed: 3767.749962 samples/sec	accuracy=0.587733	lr=0.013489
Epoch[106] Batch [649]	Speed: 3735.244429 samples/sec	accuracy=0.587733	lr=0.013411
Epoch[106] Batch [699]	Speed: 2810.466075 samples/sec	accuracy=0.587609	lr=0.013332
Epoch[106] Batch [749]	Speed: 6353.106258 samples/sec	accuracy=0.587622	lr=0.013254
Epoch[106] Batch [799]	Speed: 5782.458636 samples/sec	accuracy=0.587434	lr=0.013176
Epoch[106] Batch [849]	Speed: 5825.018971 samples/sec	accuracy=0.587378	lr=0.013098
Epoch[106] Batch [899]	Speed: 2768.335798 samples/sec	accuracy=0.587283	lr=0.013020
Epoch[106] Batch [949]	Speed: 3833.158755 samples/sec	accuracy=0.587279	lr=0.012943
Epoch[106] Batch [999]	Speed: 3614.779208 samples/sec	accuracy=0.587258	lr=0.012866
Epoch[106] Batch [1049]	Speed: 2693.050589 samples/sec	accuracy=0.587294	lr=0.012789
Epoch[106] Batch [1099]	Speed: 3586.591514 samples/sec	accuracy=0.587275	lr=0.012712
Epoch[106] Batch [1149]	Speed: 5082.188544 samples/sec	accuracy=0.587252	lr=0.012636
Epoch[106] Batch [1199]	Speed: 4548.930058 samples/sec	accuracy=0.587279	lr=0.012559
Epoch[106] Batch [1249]	Speed: 6029.814516 samples/sec	accuracy=0.587295	lr=0.012483
[Epoch 106] training: accuracy=0.587291
[Epoch 106] speed: 3611 samples/sec	time cost: 373.826566
[Epoch 106] validation: err-top1=0.372250 err-top5=0.153619
Epoch[107] Batch [49]	Speed: 3629.545636 samples/sec	accuracy=0.591875	lr=0.012406
Epoch[107] Batch [99]	Speed: 3759.926585 samples/sec	accuracy=0.592451	lr=0.012330
Epoch[107] Batch [149]	Speed: 2726.756856 samples/sec	accuracy=0.591986	lr=0.012255
Epoch[107] Batch [199]	Speed: 5631.497163 samples/sec	accuracy=0.593169	lr=0.012180
Epoch[107] Batch [249]	Speed: 2251.338548 samples/sec	accuracy=0.592629	lr=0.012105
Epoch[107] Batch [299]	Speed: 3566.605589 samples/sec	accuracy=0.592139	lr=0.012030
Epoch[107] Batch [349]	Speed: 5952.177412 samples/sec	accuracy=0.592229	lr=0.011956
Epoch[107] Batch [399]	Speed: 2706.870171 samples/sec	accuracy=0.591868	lr=0.011881
Epoch[107] Batch [449]	Speed: 5726.252611 samples/sec	accuracy=0.591439	lr=0.011807
Epoch[107] Batch [499]	Speed: 5304.836894 samples/sec	accuracy=0.591236	lr=0.011734
Epoch[107] Batch [549]	Speed: 2381.714372 samples/sec	accuracy=0.590835	lr=0.011660
Epoch[107] Batch [599]	Speed: 3762.708345 samples/sec	accuracy=0.591007	lr=0.011587
Epoch[107] Batch [649]	Speed: 5239.388311 samples/sec	accuracy=0.590978	lr=0.011513
Epoch[107] Batch [699]	Speed: 5900.340574 samples/sec	accuracy=0.590795	lr=0.011441
Epoch[107] Batch [749]	Speed: 3589.059154 samples/sec	accuracy=0.590693	lr=0.011368
Epoch[107] Batch [799]	Speed: 3712.227021 samples/sec	accuracy=0.590754	lr=0.011295
Epoch[107] Batch [849]	Speed: 3708.008275 samples/sec	accuracy=0.590703	lr=0.011223
Epoch[107] Batch [899]	Speed: 3865.674759 samples/sec	accuracy=0.590710	lr=0.011151
Epoch[107] Batch [949]	Speed: 2569.092390 samples/sec	accuracy=0.590679	lr=0.011079
Epoch[107] Batch [999]	Speed: 3627.206489 samples/sec	accuracy=0.590596	lr=0.011008
Epoch[107] Batch [1049]	Speed: 2504.695448 samples/sec	accuracy=0.590614	lr=0.010936
Epoch[107] Batch [1099]	Speed: 3585.271004 samples/sec	accuracy=0.590650	lr=0.010865
Epoch[107] Batch [1149]	Speed: 5207.861834 samples/sec	accuracy=0.590549	lr=0.010794
Epoch[107] Batch [1199]	Speed: 6032.003958 samples/sec	accuracy=0.590562	lr=0.010724
Epoch[107] Batch [1249]	Speed: 3766.407452 samples/sec	accuracy=0.590502	lr=0.010653
[Epoch 107] training: accuracy=0.590501
[Epoch 107] speed: 3673 samples/sec	time cost: 373.098471
[Epoch 107] validation: err-top1=0.365214 err-top5=0.147979
Epoch[108] Batch [49]	Speed: 2789.867343 samples/sec	accuracy=0.597207	lr=0.010582
Epoch[108] Batch [99]	Speed: 2628.221112 samples/sec	accuracy=0.595518	lr=0.010512
Epoch[108] Batch [149]	Speed: 5607.060788 samples/sec	accuracy=0.594941	lr=0.010442
Epoch[108] Batch [199]	Speed: 5653.333844 samples/sec	accuracy=0.595601	lr=0.010373
Epoch[108] Batch [249]	Speed: 3551.099928 samples/sec	accuracy=0.595977	lr=0.010303
Epoch[108] Batch [299]	Speed: 5691.929009 samples/sec	accuracy=0.595439	lr=0.010234
Epoch[108] Batch [349]	Speed: 6341.711587 samples/sec	accuracy=0.595167	lr=0.010165
Epoch[108] Batch [399]	Speed: 5907.154200 samples/sec	accuracy=0.594685	lr=0.010097
Epoch[108] Batch [449]	Speed: 3629.363207 samples/sec	accuracy=0.594651	lr=0.010028
Epoch[108] Batch [499]	Speed: 3831.905093 samples/sec	accuracy=0.594473	lr=0.009960
Epoch[108] Batch [549]	Speed: 5663.186762 samples/sec	accuracy=0.594496	lr=0.009892
Epoch[108] Batch [599]	Speed: 6026.846989 samples/sec	accuracy=0.594338	lr=0.009824
Epoch[108] Batch [649]	Speed: 3613.387583 samples/sec	accuracy=0.594255	lr=0.009757
Epoch[108] Batch [699]	Speed: 5818.572646 samples/sec	accuracy=0.593908	lr=0.009690
Epoch[108] Batch [749]	Speed: 1639.105579 samples/sec	accuracy=0.593820	lr=0.009623
Epoch[108] Batch [799]	Speed: 3526.944456 samples/sec	accuracy=0.593777	lr=0.009556
Epoch[108] Batch [849]	Speed: 3804.690401 samples/sec	accuracy=0.593912	lr=0.009489
Epoch[108] Batch [899]	Speed: 2825.977616 samples/sec	accuracy=0.594001	lr=0.009423
Epoch[108] Batch [949]	Speed: 5553.050690 samples/sec	accuracy=0.594039	lr=0.009357
Epoch[108] Batch [999]	Speed: 5140.729956 samples/sec	accuracy=0.593886	lr=0.009291
Epoch[108] Batch [1049]	Speed: 3791.918697 samples/sec	accuracy=0.593960	lr=0.009225
Epoch[108] Batch [1099]	Speed: 5573.446037 samples/sec	accuracy=0.593948	lr=0.009160
Epoch[108] Batch [1149]	Speed: 3598.769260 samples/sec	accuracy=0.593768	lr=0.009095
Epoch[108] Batch [1199]	Speed: 5606.977049 samples/sec	accuracy=0.593738	lr=0.009030
Epoch[108] Batch [1249]	Speed: 2887.775183 samples/sec	accuracy=0.593759	lr=0.008965
[Epoch 108] training: accuracy=0.593741
[Epoch 108] speed: 3941 samples/sec	time cost: 334.675156
[Epoch 108] validation: err-top1=0.362265 err-top5=0.146265
Epoch[109] Batch [49]	Speed: 3042.924964 samples/sec	accuracy=0.596992	lr=0.008899
Epoch[109] Batch [99]	Speed: 6052.817124 samples/sec	accuracy=0.595371	lr=0.008835
Epoch[109] Batch [149]	Speed: 2830.531896 samples/sec	accuracy=0.595846	lr=0.008771
Epoch[109] Batch [199]	Speed: 2051.475105 samples/sec	accuracy=0.595708	lr=0.008707
Epoch[109] Batch [249]	Speed: 3725.397721 samples/sec	accuracy=0.596531	lr=0.008643
Epoch[109] Batch [299]	Speed: 3701.317817 samples/sec	accuracy=0.596546	lr=0.008580
Epoch[109] Batch [349]	Speed: 5171.398047 samples/sec	accuracy=0.596164	lr=0.008517
Epoch[109] Batch [399]	Speed: 3747.510631 samples/sec	accuracy=0.596125	lr=0.008454
Epoch[109] Batch [449]	Speed: 5416.096990 samples/sec	accuracy=0.596306	lr=0.008391
Epoch[109] Batch [499]	Speed: 3357.719209 samples/sec	accuracy=0.596232	lr=0.008329
Epoch[109] Batch [549]	Speed: 5702.586431 samples/sec	accuracy=0.596046	lr=0.008266
Epoch[109] Batch [599]	Speed: 5944.352652 samples/sec	accuracy=0.595905	lr=0.008204
Epoch[109] Batch [649]	Speed: 2165.163311 samples/sec	accuracy=0.596140	lr=0.008143
Epoch[109] Batch [699]	Speed: 3666.480353 samples/sec	accuracy=0.595997	lr=0.008081
Epoch[109] Batch [749]	Speed: 4994.399365 samples/sec	accuracy=0.596115	lr=0.008020
Epoch[109] Batch [799]	Speed: 5354.729394 samples/sec	accuracy=0.596393	lr=0.007959
Epoch[109] Batch [849]	Speed: 2796.771161 samples/sec	accuracy=0.596142	lr=0.007898
Epoch[109] Batch [899]	Speed: 5005.186550 samples/sec	accuracy=0.596023	lr=0.007837
Epoch[109] Batch [949]	Speed: 5597.057304 samples/sec	accuracy=0.595999	lr=0.007777
Epoch[109] Batch [999]	Speed: 3699.690881 samples/sec	accuracy=0.595854	lr=0.007716
Epoch[109] Batch [1049]	Speed: 3760.207176 samples/sec	accuracy=0.595967	lr=0.007656
Epoch[109] Batch [1099]	Speed: 3550.852611 samples/sec	accuracy=0.595850	lr=0.007597
Epoch[109] Batch [1149]	Speed: 2835.241730 samples/sec	accuracy=0.595726	lr=0.007537
Epoch[109] Batch [1199]	Speed: 5197.817989 samples/sec	accuracy=0.595677	lr=0.007478
Epoch[109] Batch [1249]	Speed: 3796.148320 samples/sec	accuracy=0.595645	lr=0.007419
[Epoch 109] training: accuracy=0.595649
[Epoch 109] speed: 3760 samples/sec	time cost: 359.865319
[Epoch 109] validation: err-top1=0.360192 err-top5=0.144651
Epoch[110] Batch [49]	Speed: 5702.537065 samples/sec	accuracy=0.602891	lr=0.007359
Epoch[110] Batch [99]	Speed: 3593.252211 samples/sec	accuracy=0.602822	lr=0.007300
Epoch[110] Batch [149]	Speed: 4260.152900 samples/sec	accuracy=0.601491	lr=0.007242
Epoch[110] Batch [199]	Speed: 4553.173581 samples/sec	accuracy=0.600317	lr=0.007184
Epoch[110] Batch [249]	Speed: 2684.988519 samples/sec	accuracy=0.600711	lr=0.007126
Epoch[110] Batch [299]	Speed: 3511.400690 samples/sec	accuracy=0.600889	lr=0.007068
Epoch[110] Batch [349]	Speed: 3464.118497 samples/sec	accuracy=0.600862	lr=0.007011
Epoch[110] Batch [399]	Speed: 3779.839371 samples/sec	accuracy=0.600847	lr=0.006954
Epoch[110] Batch [449]	Speed: 3599.472896 samples/sec	accuracy=0.600653	lr=0.006897
Epoch[110] Batch [499]	Speed: 3847.270300 samples/sec	accuracy=0.600066	lr=0.006840
Epoch[110] Batch [549]	Speed: 5680.873848 samples/sec	accuracy=0.600055	lr=0.006784
Epoch[110] Batch [599]	Speed: 5557.059972 samples/sec	accuracy=0.599528	lr=0.006727
Epoch[110] Batch [649]	Speed: 5746.856133 samples/sec	accuracy=0.599154	lr=0.006671
Epoch[110] Batch [699]	Speed: 4106.799571 samples/sec	accuracy=0.599195	lr=0.006615
Epoch[110] Batch [749]	Speed: 3706.831863 samples/sec	accuracy=0.599154	lr=0.006560
Epoch[110] Batch [799]	Speed: 5790.209287 samples/sec	accuracy=0.598999	lr=0.006504
Epoch[110] Batch [849]	Speed: 3894.905717 samples/sec	accuracy=0.598824	lr=0.006449
Epoch[110] Batch [899]	Speed: 3793.080937 samples/sec	accuracy=0.598841	lr=0.006394
Epoch[110] Batch [949]	Speed: 5545.512107 samples/sec	accuracy=0.598945	lr=0.006340
Epoch[110] Batch [999]	Speed: 4748.354316 samples/sec	accuracy=0.599034	lr=0.006285
Epoch[110] Batch [1049]	Speed: 5724.206683 samples/sec	accuracy=0.598974	lr=0.006231
Epoch[110] Batch [1099]	Speed: 5931.700289 samples/sec	accuracy=0.598995	lr=0.006177
Epoch[110] Batch [1149]	Speed: 5758.813755 samples/sec	accuracy=0.598916	lr=0.006123
Epoch[110] Batch [1199]	Speed: 4499.266385 samples/sec	accuracy=0.598893	lr=0.006070
Epoch[110] Batch [1249]	Speed: 5600.081528 samples/sec	accuracy=0.598779	lr=0.006017
[Epoch 110] training: accuracy=0.598777
[Epoch 110] speed: 4381 samples/sec	time cost: 319.787739
[Epoch 110] validation: err-top1=0.360189 err-top5=0.145589
Epoch[111] Batch [49]	Speed: 5565.246928 samples/sec	accuracy=0.603457	lr=0.005963
Epoch[111] Batch [99]	Speed: 2647.267011 samples/sec	accuracy=0.601396	lr=0.005910
Epoch[111] Batch [149]	Speed: 5518.965438 samples/sec	accuracy=0.602637	lr=0.005857
Epoch[111] Batch [199]	Speed: 4014.631744 samples/sec	accuracy=0.601050	lr=0.005805
Epoch[111] Batch [249]	Speed: 5883.877292 samples/sec	accuracy=0.600941	lr=0.005753
Epoch[111] Batch [299]	Speed: 2822.539348 samples/sec	accuracy=0.600560	lr=0.005701
Epoch[111] Batch [349]	Speed: 3969.612020 samples/sec	accuracy=0.599992	lr=0.005649
Epoch[111] Batch [399]	Speed: 5447.689449 samples/sec	accuracy=0.599939	lr=0.005598
Epoch[111] Batch [449]	Speed: 1275.537910 samples/sec	accuracy=0.600200	lr=0.005547
Epoch[111] Batch [499]	Speed: 2836.192201 samples/sec	accuracy=0.600322	lr=0.005496
Epoch[111] Batch [549]	Speed: 3994.160443 samples/sec	accuracy=0.600142	lr=0.005445
Epoch[111] Batch [599]	Speed: 1163.181747 samples/sec	accuracy=0.599834	lr=0.005394
Epoch[111] Batch [649]	Speed: 2952.282784 samples/sec	accuracy=0.599827	lr=0.005344
Epoch[111] Batch [699]	Speed: 2946.790895 samples/sec	accuracy=0.599527	lr=0.005294
Epoch[111] Batch [749]	Speed: 6122.828694 samples/sec	accuracy=0.599792	lr=0.005244
Epoch[111] Batch [799]	Speed: 5657.741790 samples/sec	accuracy=0.599978	lr=0.005195
Epoch[111] Batch [849]	Speed: 3005.530383 samples/sec	accuracy=0.600056	lr=0.005145
Epoch[111] Batch [899]	Speed: 2700.289339 samples/sec	accuracy=0.600038	lr=0.005096
Epoch[111] Batch [949]	Speed: 1322.937657 samples/sec	accuracy=0.599860	lr=0.005047
Epoch[111] Batch [999]	Speed: 5326.918380 samples/sec	accuracy=0.599805	lr=0.004999
Epoch[111] Batch [1049]	Speed: 5055.344135 samples/sec	accuracy=0.599893	lr=0.004950
Epoch[111] Batch [1099]	Speed: 909.371098 samples/sec	accuracy=0.600040	lr=0.004902
Epoch[111] Batch [1149]	Speed: 3042.756083 samples/sec	accuracy=0.600204	lr=0.004854
Epoch[111] Batch [1199]	Speed: 5539.946798 samples/sec	accuracy=0.600243	lr=0.004807
Epoch[111] Batch [1249]	Speed: 3995.784751 samples/sec	accuracy=0.600322	lr=0.004759
[Epoch 111] training: accuracy=0.600324
[Epoch 111] speed: 2802 samples/sec	time cost: 467.070983
[Epoch 111] validation: err-top1=0.355230 err-top5=0.140366
Epoch[112] Batch [49]	Speed: 3704.518662 samples/sec	accuracy=0.604492	lr=0.004711
Epoch[112] Batch [99]	Speed: 5341.350854 samples/sec	accuracy=0.604873	lr=0.004664
Epoch[112] Batch [149]	Speed: 5786.915935 samples/sec	accuracy=0.605098	lr=0.004617
Epoch[112] Batch [199]	Speed: 5706.183057 samples/sec	accuracy=0.604829	lr=0.004571
Epoch[112] Batch [249]	Speed: 5515.886769 samples/sec	accuracy=0.604680	lr=0.004524
Epoch[112] Batch [299]	Speed: 3950.321642 samples/sec	accuracy=0.604303	lr=0.004478
Epoch[112] Batch [349]	Speed: 5986.470190 samples/sec	accuracy=0.604325	lr=0.004432
Epoch[112] Batch [399]	Speed: 5461.407918 samples/sec	accuracy=0.603772	lr=0.004387
Epoch[112] Batch [449]	Speed: 5310.086823 samples/sec	accuracy=0.603770	lr=0.004341
Epoch[112] Batch [499]	Speed: 5700.517276 samples/sec	accuracy=0.604006	lr=0.004296
Epoch[112] Batch [549]	Speed: 2580.074000 samples/sec	accuracy=0.603853	lr=0.004251
Epoch[112] Batch [599]	Speed: 5592.399362 samples/sec	accuracy=0.603996	lr=0.004207
Epoch[112] Batch [649]	Speed: 5926.590492 samples/sec	accuracy=0.603818	lr=0.004162
Epoch[112] Batch [699]	Speed: 3586.749539 samples/sec	accuracy=0.603624	lr=0.004118
Epoch[112] Batch [749]	Speed: 3425.542157 samples/sec	accuracy=0.603536	lr=0.004074
Epoch[112] Batch [799]	Speed: 4088.663633 samples/sec	accuracy=0.603553	lr=0.004030
Epoch[112] Batch [849]	Speed: 5695.040795 samples/sec	accuracy=0.603523	lr=0.003987
Epoch[112] Batch [899]	Speed: 2987.255344 samples/sec	accuracy=0.603416	lr=0.003944
Epoch[112] Batch [949]	Speed: 4115.513775 samples/sec	accuracy=0.603571	lr=0.003901
Epoch[112] Batch [999]	Speed: 3934.548202 samples/sec	accuracy=0.603459	lr=0.003858
Epoch[112] Batch [1049]	Speed: 3007.972003 samples/sec	accuracy=0.603485	lr=0.003815
Epoch[112] Batch [1099]	Speed: 3601.470572 samples/sec	accuracy=0.603387	lr=0.003773
Epoch[112] Batch [1149]	Speed: 3982.552685 samples/sec	accuracy=0.603347	lr=0.003731
Epoch[112] Batch [1199]	Speed: 3687.167965 samples/sec	accuracy=0.603348	lr=0.003689
Epoch[112] Batch [1249]	Speed: 3051.957717 samples/sec	accuracy=0.603366	lr=0.003647
[Epoch 112] training: accuracy=0.603356
[Epoch 112] speed: 4182 samples/sec	time cost: 315.621342
[Epoch 112] validation: err-top1=0.354612 err-top5=0.140067
Epoch[113] Batch [49]	Speed: 3471.750868 samples/sec	accuracy=0.608672	lr=0.003605
Epoch[113] Batch [99]	Speed: 2929.983860 samples/sec	accuracy=0.608086	lr=0.003564
Epoch[113] Batch [149]	Speed: 2970.462325 samples/sec	accuracy=0.607148	lr=0.003523
Epoch[113] Batch [199]	Speed: 3050.958756 samples/sec	accuracy=0.606836	lr=0.003482
Epoch[113] Batch [249]	Speed: 3976.895732 samples/sec	accuracy=0.606578	lr=0.003442
Epoch[113] Batch [299]	Speed: 3973.689199 samples/sec	accuracy=0.606292	lr=0.003402
Epoch[113] Batch [349]	Speed: 5745.192289 samples/sec	accuracy=0.606443	lr=0.003362
Epoch[113] Batch [399]	Speed: 3959.767512 samples/sec	accuracy=0.606333	lr=0.003322
Epoch[113] Batch [449]	Speed: 4055.570153 samples/sec	accuracy=0.606330	lr=0.003282
Epoch[113] Batch [499]	Speed: 5631.618557 samples/sec	accuracy=0.605691	lr=0.003243
Epoch[113] Batch [549]	Speed: 3513.578785 samples/sec	accuracy=0.605376	lr=0.003204
Epoch[113] Batch [599]	Speed: 5615.735791 samples/sec	accuracy=0.605524	lr=0.003165
Epoch[113] Batch [649]	Speed: 5290.800887 samples/sec	accuracy=0.605529	lr=0.003127
Epoch[113] Batch [699]	Speed: 3701.828690 samples/sec	accuracy=0.605190	lr=0.003088
Epoch[113] Batch [749]	Speed: 2912.576717 samples/sec	accuracy=0.604785	lr=0.003050
Epoch[113] Batch [799]	Speed: 4047.458033 samples/sec	accuracy=0.604659	lr=0.003012
Epoch[113] Batch [849]	Speed: 1249.957800 samples/sec	accuracy=0.604601	lr=0.002975
Epoch[113] Batch [899]	Speed: 3943.218919 samples/sec	accuracy=0.604489	lr=0.002937
Epoch[113] Batch [949]	Speed: 2495.695283 samples/sec	accuracy=0.604690	lr=0.002900
Epoch[113] Batch [999]	Speed: 5833.292040 samples/sec	accuracy=0.604615	lr=0.002863
Epoch[113] Batch [1049]	Speed: 5746.202749 samples/sec	accuracy=0.604634	lr=0.002826
Epoch[113] Batch [1099]	Speed: 5856.146818 samples/sec	accuracy=0.604666	lr=0.002790
Epoch[113] Batch [1149]	Speed: 3138.214612 samples/sec	accuracy=0.604684	lr=0.002754
Epoch[113] Batch [1199]	Speed: 4028.338726 samples/sec	accuracy=0.604701	lr=0.002718
Epoch[113] Batch [1249]	Speed: 5816.255120 samples/sec	accuracy=0.604675	lr=0.002682
[Epoch 113] training: accuracy=0.604690
[Epoch 113] speed: 3637 samples/sec	time cost: 368.642300
[Epoch 113] validation: err-top1=0.351722 err-top5=0.138293
Epoch[114] Batch [49]	Speed: 3966.727571 samples/sec	accuracy=0.607559	lr=0.002646
Epoch[114] Batch [99]	Speed: 4073.726161 samples/sec	accuracy=0.606650	lr=0.002611
Epoch[114] Batch [149]	Speed: 4713.278906 samples/sec	accuracy=0.607253	lr=0.002575
Epoch[114] Batch [199]	Speed: 5532.227948 samples/sec	accuracy=0.606772	lr=0.002541
Epoch[114] Batch [249]	Speed: 3714.600300 samples/sec	accuracy=0.607000	lr=0.002506
Epoch[114] Batch [299]	Speed: 5898.904905 samples/sec	accuracy=0.606556	lr=0.002472
Epoch[114] Batch [349]	Speed: 2513.616882 samples/sec	accuracy=0.606616	lr=0.002438
Epoch[114] Batch [399]	Speed: 5697.637281 samples/sec	accuracy=0.606704	lr=0.002404
Epoch[114] Batch [449]	Speed: 2514.522721 samples/sec	accuracy=0.607025	lr=0.002370
Epoch[114] Batch [499]	Speed: 5585.804048 samples/sec	accuracy=0.607135	lr=0.002337
Epoch[114] Batch [549]	Speed: 6014.192561 samples/sec	accuracy=0.607109	lr=0.002304
Epoch[114] Batch [599]	Speed: 4096.443095 samples/sec	accuracy=0.606803	lr=0.002271
Epoch[114] Batch [649]	Speed: 3831.549233 samples/sec	accuracy=0.607124	lr=0.002238
Epoch[114] Batch [699]	Speed: 5000.029216 samples/sec	accuracy=0.607069	lr=0.002205
Epoch[114] Batch [749]	Speed: 6066.572768 samples/sec	accuracy=0.607027	lr=0.002173
Epoch[114] Batch [799]	Speed: 5379.917910 samples/sec	accuracy=0.607153	lr=0.002141
Epoch[114] Batch [849]	Speed: 5204.195522 samples/sec	accuracy=0.607263	lr=0.002110
Epoch[114] Batch [899]	Speed: 3935.043505 samples/sec	accuracy=0.607358	lr=0.002078
Epoch[114] Batch [949]	Speed: 3735.455916 samples/sec	accuracy=0.607403	lr=0.002047
Epoch[114] Batch [999]	Speed: 3125.573666 samples/sec	accuracy=0.607336	lr=0.002016
Epoch[114] Batch [1049]	Speed: 5625.395642 samples/sec	accuracy=0.607249	lr=0.001985
Epoch[114] Batch [1099]	Speed: 5866.297573 samples/sec	accuracy=0.607294	lr=0.001954
Epoch[114] Batch [1149]	Speed: 3497.828541 samples/sec	accuracy=0.607210	lr=0.001924
Epoch[114] Batch [1199]	Speed: 5691.800474 samples/sec	accuracy=0.607058	lr=0.001894
Epoch[114] Batch [1249]	Speed: 5135.501503 samples/sec	accuracy=0.607099	lr=0.001864
[Epoch 114] training: accuracy=0.607069
[Epoch 114] speed: 4350 samples/sec	time cost: 304.166447
[Epoch 114] validation: err-top1=0.351224 err-top5=0.138732
Epoch[115] Batch [49]	Speed: 3804.980748 samples/sec	accuracy=0.610000	lr=0.001834
Epoch[115] Batch [99]	Speed: 5807.270720 samples/sec	accuracy=0.608027	lr=0.001804
Epoch[115] Batch [149]	Speed: 5699.813570 samples/sec	accuracy=0.608092	lr=0.001775
Epoch[115] Batch [199]	Speed: 5627.821330 samples/sec	accuracy=0.608740	lr=0.001746
Epoch[115] Batch [249]	Speed: 5242.845184 samples/sec	accuracy=0.609035	lr=0.001718
Epoch[115] Batch [299]	Speed: 5256.973077 samples/sec	accuracy=0.609206	lr=0.001689
Epoch[115] Batch [349]	Speed: 2424.013810 samples/sec	accuracy=0.609540	lr=0.001661
Epoch[115] Batch [399]	Speed: 3094.066160 samples/sec	accuracy=0.609370	lr=0.001633
Epoch[115] Batch [449]	Speed: 6021.657123 samples/sec	accuracy=0.609544	lr=0.001605
Epoch[115] Batch [499]	Speed: 5865.451895 samples/sec	accuracy=0.609334	lr=0.001578
Epoch[115] Batch [549]	Speed: 4073.479583 samples/sec	accuracy=0.609315	lr=0.001551
Epoch[115] Batch [599]	Speed: 5758.814373 samples/sec	accuracy=0.609163	lr=0.001524
Epoch[115] Batch [649]	Speed: 4103.965617 samples/sec	accuracy=0.609135	lr=0.001497
Epoch[115] Batch [699]	Speed: 4002.817647 samples/sec	accuracy=0.609039	lr=0.001470
Epoch[115] Batch [749]	Speed: 6168.088448 samples/sec	accuracy=0.608742	lr=0.001444
Epoch[115] Batch [799]	Speed: 2295.780263 samples/sec	accuracy=0.608623	lr=0.001418
Epoch[115] Batch [849]	Speed: 1028.120562 samples/sec	accuracy=0.608433	lr=0.001392
Epoch[115] Batch [899]	Speed: 3949.778316 samples/sec	accuracy=0.608349	lr=0.001366
Epoch[115] Batch [949]	Speed: 4114.871785 samples/sec	accuracy=0.608346	lr=0.001341
Epoch[115] Batch [999]	Speed: 5506.599747 samples/sec	accuracy=0.608529	lr=0.001316
Epoch[115] Batch [1049]	Speed: 3067.750603 samples/sec	accuracy=0.608593	lr=0.001291
Epoch[115] Batch [1099]	Speed: 3636.933438 samples/sec	accuracy=0.608533	lr=0.001266
Epoch[115] Batch [1149]	Speed: 1817.694084 samples/sec	accuracy=0.608284	lr=0.001242
Epoch[115] Batch [1199]	Speed: 5414.147352 samples/sec	accuracy=0.608237	lr=0.001218
Epoch[115] Batch [1249]	Speed: 4189.677516 samples/sec	accuracy=0.608282	lr=0.001194
[Epoch 115] training: accuracy=0.608297
[Epoch 115] speed: 3580 samples/sec	time cost: 371.082360
[Epoch 115] validation: err-top1=0.349290 err-top5=0.137655
Epoch[116] Batch [49]	Speed: 2955.555441 samples/sec	accuracy=0.608262	lr=0.001170
Epoch[116] Batch [99]	Speed: 4104.493043 samples/sec	accuracy=0.607656	lr=0.001146
Epoch[116] Batch [149]	Speed: 2977.848270 samples/sec	accuracy=0.607663	lr=0.001123
Epoch[116] Batch [199]	Speed: 3058.385861 samples/sec	accuracy=0.607705	lr=0.001100
Epoch[116] Batch [249]	Speed: 4031.939225 samples/sec	accuracy=0.607437	lr=0.001077
Epoch[116] Batch [299]	Speed: 5840.453818 samples/sec	accuracy=0.607930	lr=0.001055
Epoch[116] Batch [349]	Speed: 4146.065580 samples/sec	accuracy=0.607893	lr=0.001032
Epoch[116] Batch [399]	Speed: 5786.127909 samples/sec	accuracy=0.607847	lr=0.001010
Epoch[116] Batch [449]	Speed: 5905.858951 samples/sec	accuracy=0.607899	lr=0.000989
Epoch[116] Batch [499]	Speed: 3778.376147 samples/sec	accuracy=0.607838	lr=0.000967
Epoch[116] Batch [549]	Speed: 4066.854268 samples/sec	accuracy=0.607844	lr=0.000946
Epoch[116] Batch [599]	Speed: 5633.213417 samples/sec	accuracy=0.607554	lr=0.000925
Epoch[116] Batch [649]	Speed: 5530.488917 samples/sec	accuracy=0.607704	lr=0.000904
Epoch[116] Batch [699]	Speed: 5654.102934 samples/sec	accuracy=0.608118	lr=0.000883
Epoch[116] Batch [749]	Speed: 5798.101281 samples/sec	accuracy=0.608609	lr=0.000863
Epoch[116] Batch [799]	Speed: 5743.682257 samples/sec	accuracy=0.608662	lr=0.000843
Epoch[116] Batch [849]	Speed: 2113.296803 samples/sec	accuracy=0.608644	lr=0.000823
Epoch[116] Batch [899]	Speed: 4392.656911 samples/sec	accuracy=0.608663	lr=0.000803
Epoch[116] Batch [949]	Speed: 4121.291142 samples/sec	accuracy=0.608473	lr=0.000784
Epoch[116] Batch [999]	Speed: 5385.495170 samples/sec	accuracy=0.608704	lr=0.000764
Epoch[116] Batch [1049]	Speed: 3798.136213 samples/sec	accuracy=0.608761	lr=0.000745
Epoch[116] Batch [1099]	Speed: 5839.257512 samples/sec	accuracy=0.609034	lr=0.000727
Epoch[116] Batch [1149]	Speed: 3874.322494 samples/sec	accuracy=0.609126	lr=0.000708
Epoch[116] Batch [1199]	Speed: 3980.125222 samples/sec	accuracy=0.609202	lr=0.000690
Epoch[116] Batch [1249]	Speed: 3022.118499 samples/sec	accuracy=0.609359	lr=0.000672
[Epoch 116] training: accuracy=0.609373
[Epoch 116] speed: 4132 samples/sec	time cost: 319.729633
[Epoch 116] validation: err-top1=0.348735 err-top5=0.137126
Epoch[117] Batch [49]	Speed: 4502.100192 samples/sec	accuracy=0.610410	lr=0.000654
Epoch[117] Batch [99]	Speed: 6005.238084 samples/sec	accuracy=0.609307	lr=0.000636
Epoch[117] Batch [149]	Speed: 3901.304028 samples/sec	accuracy=0.609811	lr=0.000619
Epoch[117] Batch [199]	Speed: 4705.184888 samples/sec	accuracy=0.610493	lr=0.000602
Epoch[117] Batch [249]	Speed: 5333.471143 samples/sec	accuracy=0.610129	lr=0.000585
Epoch[117] Batch [299]	Speed: 5207.893029 samples/sec	accuracy=0.610046	lr=0.000569
Epoch[117] Batch [349]	Speed: 3944.467152 samples/sec	accuracy=0.610784	lr=0.000552
Epoch[117] Batch [399]	Speed: 5573.828952 samples/sec	accuracy=0.611006	lr=0.000536
Epoch[117] Batch [449]	Speed: 3157.587114 samples/sec	accuracy=0.610710	lr=0.000520
Epoch[117] Batch [499]	Speed: 3982.452906 samples/sec	accuracy=0.610535	lr=0.000505
Epoch[117] Batch [549]	Speed: 5460.251742 samples/sec	accuracy=0.610616	lr=0.000489
Epoch[117] Batch [599]	Speed: 1800.582972 samples/sec	accuracy=0.610597	lr=0.000474
Epoch[117] Batch [649]	Speed: 3091.545139 samples/sec	accuracy=0.610669	lr=0.000459
Epoch[117] Batch [699]	Speed: 2896.313712 samples/sec	accuracy=0.610449	lr=0.000445
Epoch[117] Batch [749]	Speed: 4046.757788 samples/sec	accuracy=0.610215	lr=0.000430
Epoch[117] Batch [799]	Speed: 4039.698581 samples/sec	accuracy=0.609949	lr=0.000416
Epoch[117] Batch [849]	Speed: 5529.362537 samples/sec	accuracy=0.610124	lr=0.000402
Epoch[117] Batch [899]	Speed: 4110.025047 samples/sec	accuracy=0.610112	lr=0.000388
Epoch[117] Batch [949]	Speed: 3116.235012 samples/sec	accuracy=0.610034	lr=0.000375
Epoch[117] Batch [999]	Speed: 6020.277762 samples/sec	accuracy=0.610228	lr=0.000362
Epoch[117] Batch [1049]	Speed: 5523.042393 samples/sec	accuracy=0.610239	lr=0.000349
Epoch[117] Batch [1099]	Speed: 2968.624473 samples/sec	accuracy=0.610308	lr=0.000336
Epoch[117] Batch [1149]	Speed: 2278.744480 samples/sec	accuracy=0.610388	lr=0.000323
Epoch[117] Batch [1199]	Speed: 5472.476719 samples/sec	accuracy=0.610396	lr=0.000311
Epoch[117] Batch [1249]	Speed: 5556.851900 samples/sec	accuracy=0.610485	lr=0.000299
[Epoch 117] training: accuracy=0.610491
[Epoch 117] speed: 3920 samples/sec	time cost: 341.050297
[Epoch 117] validation: err-top1=0.348334 err-top5=0.136918
Epoch[118] Batch [49]	Speed: 4762.153297 samples/sec	accuracy=0.613848	lr=0.000287
Epoch[118] Batch [99]	Speed: 4311.544062 samples/sec	accuracy=0.610752	lr=0.000275
Epoch[118] Batch [149]	Speed: 3123.268906 samples/sec	accuracy=0.611100	lr=0.000264
Epoch[118] Batch [199]	Speed: 6063.002481 samples/sec	accuracy=0.610171	lr=0.000253
Epoch[118] Batch [249]	Speed: 5898.798287 samples/sec	accuracy=0.609664	lr=0.000242
Epoch[118] Batch [299]	Speed: 3410.214549 samples/sec	accuracy=0.610140	lr=0.000231
Epoch[118] Batch [349]	Speed: 5238.574547 samples/sec	accuracy=0.610711	lr=0.000221
Epoch[118] Batch [399]	Speed: 5820.091243 samples/sec	accuracy=0.610659	lr=0.000211
Epoch[118] Batch [449]	Speed: 3961.571501 samples/sec	accuracy=0.610742	lr=0.000201
Epoch[118] Batch [499]	Speed: 5174.704905 samples/sec	accuracy=0.610488	lr=0.000191
Epoch[118] Batch [549]	Speed: 3482.696580 samples/sec	accuracy=0.610634	lr=0.000182
Epoch[118] Batch [599]	Speed: 5660.688569 samples/sec	accuracy=0.610708	lr=0.000173
Epoch[118] Batch [649]	Speed: 5664.813451 samples/sec	accuracy=0.611028	lr=0.000164
Epoch[118] Batch [699]	Speed: 5534.670219 samples/sec	accuracy=0.610991	lr=0.000155
Epoch[118] Batch [749]	Speed: 4040.764727 samples/sec	accuracy=0.610958	lr=0.000147
Epoch[118] Batch [799]	Speed: 3340.048863 samples/sec	accuracy=0.610985	lr=0.000138
Epoch[118] Batch [849]	Speed: 3322.137371 samples/sec	accuracy=0.611029	lr=0.000130
Epoch[118] Batch [899]	Speed: 2878.695510 samples/sec	accuracy=0.611075	lr=0.000123
Epoch[118] Batch [949]	Speed: 4176.471933 samples/sec	accuracy=0.611068	lr=0.000115
Epoch[118] Batch [999]	Speed: 2369.132716 samples/sec	accuracy=0.610768	lr=0.000108
Epoch[118] Batch [1049]	Speed: 4111.978102 samples/sec	accuracy=0.610933	lr=0.000101
Epoch[118] Batch [1099]	Speed: 5139.779487 samples/sec	accuracy=0.610888	lr=0.000094
Epoch[118] Batch [1149]	Speed: 4215.601019 samples/sec	accuracy=0.610963	lr=0.000087
Epoch[118] Batch [1199]	Speed: 6046.359042 samples/sec	accuracy=0.610837	lr=0.000081
Epoch[118] Batch [1249]	Speed: 4264.125643 samples/sec	accuracy=0.610787	lr=0.000075
[Epoch 118] training: accuracy=0.610774
[Epoch 118] speed: 4196 samples/sec	time cost: 317.776568
[Epoch 118] validation: err-top1=0.347816 err-top5=0.136719
Epoch[119] Batch [49]	Speed: 5172.143488 samples/sec	accuracy=0.610742	lr=0.000069
Epoch[119] Batch [99]	Speed: 5524.269369 samples/sec	accuracy=0.609717	lr=0.000063
Epoch[119] Batch [149]	Speed: 5968.124741 samples/sec	accuracy=0.609284	lr=0.000058
Epoch[119] Batch [199]	Speed: 4122.555511 samples/sec	accuracy=0.609004	lr=0.000053
Epoch[119] Batch [249]	Speed: 5339.963695 samples/sec	accuracy=0.609582	lr=0.000048
Epoch[119] Batch [299]	Speed: 4280.434707 samples/sec	accuracy=0.609290	lr=0.000043
Epoch[119] Batch [349]	Speed: 4204.274536 samples/sec	accuracy=0.609718	lr=0.000039
Epoch[119] Batch [399]	Speed: 4043.948266 samples/sec	accuracy=0.610110	lr=0.000035
Epoch[119] Batch [449]	Speed: 4179.549130 samples/sec	accuracy=0.610117	lr=0.000031
Epoch[119] Batch [499]	Speed: 3429.870807 samples/sec	accuracy=0.610041	lr=0.000027
Epoch[119] Batch [549]	Speed: 2913.927856 samples/sec	accuracy=0.610087	lr=0.000023
Epoch[119] Batch [599]	Speed: 6158.496055 samples/sec	accuracy=0.610192	lr=0.000020
Epoch[119] Batch [649]	Speed: 1587.118099 samples/sec	accuracy=0.610385	lr=0.000017
Epoch[119] Batch [699]	Speed: 4024.699394 samples/sec	accuracy=0.610589	lr=0.000015
Epoch[119] Batch [749]	Speed: 4065.031320 samples/sec	accuracy=0.610702	lr=0.000012
Epoch[119] Batch [799]	Speed: 3955.024624 samples/sec	accuracy=0.610481	lr=0.000010
Epoch[119] Batch [849]	Speed: 5111.319430 samples/sec	accuracy=0.610484	lr=0.000008
Epoch[119] Batch [899]	Speed: 4291.367440 samples/sec	accuracy=0.610525	lr=0.000006
Epoch[119] Batch [949]	Speed: 4270.904847 samples/sec	accuracy=0.610536	lr=0.000004
Epoch[119] Batch [999]	Speed: 4684.654980 samples/sec	accuracy=0.610610	lr=0.000003
Epoch[119] Batch [1049]	Speed: 5433.769132 samples/sec	accuracy=0.610746	lr=0.000002
Epoch[119] Batch [1099]	Speed: 4066.580028 samples/sec	accuracy=0.610771	lr=0.000001
Epoch[119] Batch [1149]	Speed: 4259.793500 samples/sec	accuracy=0.610771	lr=0.000000
Epoch[119] Batch [1199]	Speed: 3320.190250 samples/sec	accuracy=0.610830	lr=0.000000
Epoch[119] Batch [1249]	Speed: 5445.493436 samples/sec	accuracy=0.610905	lr=0.000000
[Epoch 119] training: accuracy=0.610926
[Epoch 119] speed: 4074 samples/sec	time cost: 326.844471
[Epoch 119] validation: err-top1=0.347517 err-top5=0.137237